article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
the first pulsar catalog released by _ fermi
_ large area telescope ( lat ) included 46 pulsars @xcite , 8 of which were millisecond pulsars ( msps ) @xcite . currently , there are @xmath2 msps @xcite and @xmath3 pulsars in total @xcite
. the discovery of @xcite revealed it to be the first msp to have ( nearly ) phase - aligned radio and light curves ( lcs ) .
this rare phenomenon has only been observed for the crab pulsar @xcite .
however , this behavior has now also been observed for ( b1937 + 21 ) , ( b1957 + 20 ) @xcite , and @xcite , and more msps will be added to this subclass .
two classes of pulsar models have been used to describe high - energy ( he ) pulsar emission . in polar cap ( pc )
models @xcite , primary electrons are ejected from the neutron star ( ns ) surface and accelerated along curved magnetic field lines , producing curvature radiation gamma rays .
thermal x - rays may also be upscattered to energies .
subsequently , these gamma rays are converted into electron - positron pairs via magnetic pair production in the intense magnetic fields close to the stellar surface ( at radius @xmath4 ) .
in addition , a slot gap ( sg ) @xcite may form along the last open magnetic field lines of the pulsar magnetosphere in the absence of pair creation along those lines .
this corresponds to a two - pole caustic ( tpc ) geometry @xcite which may extend from the stellar surface up to near the light cylinder ( at radius @xmath0 ) .
outer gap ( og ) models @xcite represent the second model class . in these models , he radiation is produced along the last open field lines above the null charge surface ( ncs ) where the goldreich - julian charge density changes sign .
the narrow gaps in both the og and tpc models require screening of the electric field parallel to the local magnetic field , and therefore presupposes copious pair production .
lastly , he lcs were also modeled in the context of og and tpc models in a force - free magnetic field geometry , proposing a separatrix layer model close to @xmath0 @xcite .
he photons escaping from the magnetosphere are subject to two relativistic effects : their traveling direction is aberrated due to the large corotation velocity , while their arrival time at the observer is determined by their emission height , due to the finite speed of light .
lastly , it is assumed that these photons are emitted tangent to the local magnetic field lines in the co - rotating frame .
the combination of these three effects result in the formation of caustics , i.e. , the accumulation of photons in narrow phase bands @xcite .
these caustics manifest themselves as bright peaks in the observed pulse profiles . due to their much lower surface dipole magnetic field strengths , msps have been thought to have pair - starved magnetospheres , where the magnetic pair multiplicity is not high enough to screen the accelerating electric field in the open volume above the pc @xcite . in this case ,
primary electrons are accelerated up to very high altitudes above the full pc , while pair formation is suppressed .
this model is called the pair - starved polar cap ( pspc ) model @xcite , an extension of the traditional pc model .
msp lcs and spectra have been modeled using this framework @xcite .
alternatively , msp spectra and energetics have also been modeled in the context of an og model @xcite .
an annular gap model @xcite can furthermore reproduce the main characteristics of the lcs of three msps , although this model does not attempt to model the nonzero phase offsets between the and radio profiles .
the first 8 _ fermi_-detected msps have been modeled @xcite .
two distinct msp subclasses were found : those whose lcs are well fit by a standard og or tpc model , and those whose lcs are well fit by a pspc model ( with these fits being mutually exclusive ) . these models yielded
the correct radio - to - gamma phase lags when the radio emission was modeled as a cone beam at lower altitude .
such fits implied that msps have screened magnetospheres with large amounts of pairs available , as these conditions are needed to set up the gap structure presupposed by the og / tpc models .
small distortions of the dipole magnetic field causing offsets of the pc may provide a mechanism for enhancing pair creation , even in low - spin - down pulsars @xcite . this paper discusses a third sublcass of msps : those having phase - aligned radio and lcs .
in contrast to the first two msp sublcasses , the near phase - alignment of the and radio lcs of msps in the third subclass argues for overlapping emission regions .
these co - located emission regions may occur at high altitudes , so that the radio emission will be subject to the same relativistic effects as the emission , as described in section [ sec : caustics ] .
discoveries of new msps exhibiting phase - aligned lcs therefore motivate the investigation of high - altitude @xcite caustic radio emission . a second argument
motivating caustic radio emission comes from investigating the beaming properties of normal pulsars and msps detectable using blind searches on data as well as radio data @xcite .
the relative number of to radio pulsars for each of these ` gamma - ray - selected ' and ` radio - selected ' samples implies that radio and beams must have comparable sky coverage of @xmath5 sr for pulsars with high spin - down luminosities ( @xmath6 ) , but radio beams should shrink for pulsars having lower values of @xmath6 . the radio emission for high-@xmath6 pulsars should therefore originate in wide beams at a significant fraction of @xmath0 .
one should however bear in mind that lcs resulting from radio and caustics would generally be nearly phase - aligned ( although small phase differences could result if the radio and emission regions are at different altitudes ) .
caustic radio emission is therefore plausible for young pulsars with nearly aligned lcs .
radio caustics may however be more common in the case of the msps , as there are many more examples of msps with phase - aligned lcs .
we investigate the possibility of reproducing phase - aligned radio and lcs using `` altitude - limited '' og / tpc models ( alog / altpc ) in which we limit the extent of the emission regions ( section [ sec : alt - lim ] ) vs. a low - altitude sg ( lasg ) model ( section [ sec : lasg ] ) . by modeling the lcs of and we can infer values for the magnetic inclination and observer angles @xmath7 and @xmath8 ( section [ sec : results ] ) , and also constrain the emission altitudes .
[ [ section ] ] was discovered using the parkes radio telescope @xcite .
it follows a circular orbit around a low - mass companion ( _ hubble space telescope _
observations revealed an optical white dwarf companion with a mass of about @xmath9 ; @xcite ) .
s period of @xmath10 ms implies a rotational age of @xmath11 gyr , dipolar surface field of @xmath12 g , and spin - down power of @xmath13 erg s@xmath14 @xcite , typical among the radio msp population .
it is also relatively close , lying at 0.5 kpc @xcite .
no rotating vector model ( rvm ) fits exist for @xcite because there is no detected linear polarization .
a @xmath15 detection of an x - ray source 0.2@xmath16 from the pulsar position by _ xmm - newton _ was reported @xcite , while _ egret _ obtained a @xmath17 flux upper limit above 100 mev @xcite exceeding the recent _ fermi _ flux measurements @xcite by an order of magnitude .
is the first msp ever discovered @xcite .
it has a period @xmath18 ms , and remains one of the fastest - spinning msps discovered .
this msp has a very high spin - down luminosity of @xmath19 erg s@xmath14 , surface magnetic field of @xmath20 g , and characteristic age of @xmath21 gyr @xcite , and lies at a distance of @xmath22 kpc @xcite .
_ rxte _ observations @xcite revealed a double - peaked x - ray profile with phase separation of about half a rotation , closely aligned with the phases of the giant radio pulses @xcite , but slightly lagging the radio peaks .
the _ egret _ @xmath17 upper limit to the unpulsed flux above 100 mev was @xmath23 cm s@xmath14 @xcite .
pulsations with a significance well above @xmath24 have now been detected from by _ fermi _ lat @xcite .
was the first `` black widow '' pulsar discovered .
it is in a nearly circular eclipsing binary orbit , ablating its low - mass tidally - locked companion star @xcite .
lying at a distance of @xmath25 kpc , its period @xmath26 ms and intrinsic @xmath27 is @xmath28 @xcite lead to a spin - down luminosity of @xmath29 erg s@xmath14 , surface magnetic field of @xmath12 g , and characteristic age of @xmath30 gyr .
_ xmm - newton _ observations revealed a phase dependence of the x - ray emission on the binary orbital period @xcite , although no pulsations were detected at the pulsar spin period @xmath31 .
a @xmath32 pulsed x - ray signal have now been observed from with the x - ray peaks seemingly in close alignment with the radio peaks @xcite .
significant pulsed emission has also been detected by _
lat @xcite .
as in in our previous work @xcite , we assumed a retarded vacuum dipole magnetic field as the basic structure of the pulsar magnetosphere @xcite . in the case of young pulsars
, this field may actually be closer to the force - free solution @xcite , but it is not clear whether msps produce enough pairs to facilitate a force - free magnetosphere .
we furthermore use a photon emission rate that is constant along magnetic field lines in the corotating frame , and treat relativistic effects ( i.e. , aberration of photon directions and time - of - flight delays ; section [ sec : caustics ] ) consistently to first order in @xmath33 ( with @xmath34 the radial distance from the ns center ) .
we lastly include the lorentz transformation ( a second - order effect in @xmath33 ) of the local magnetic field between lab and corotating frames @xcite .
we use the same framework as previously @xcite , but the radio emission region is extended in altitude .
we free the minimum and maximum radii of the radio ( @xmath35 and @xmath36 ) and the maximum radius of the ( @xmath37 ) emission regions , and restrict the emission gaps extent to a cylindrical radius @xmath38 .
importantly , we do not use an axis - centered radio conal model , but investigate radio photons coming from an og / tpc - like structure . in the alog radio models , the minimum radius is actually @xmath39 , so it is a function of magnetic azimuth @xmath40 and co - latitude @xmath41 when @xmath42 . here , @xmath43 is the radius of the ncs .
we always set @xmath44 for altpc ( and tpc ) and @xmath45 for alog ( and alog ) fits , while @xmath35 may vary and may even be quite close to @xmath0 .
our alog and altpc models have 9 and 8 free parameters respectively , describing the pulsar geometry ( @xmath7 and @xmath8 ) and and radio gap locations , apart from @xmath31 which determines the size of the pc ( see figure [ fig : emission_layers ] ) .
more details are provided in @xcite .
note that the radio and emission layers are fit independently .
, and the spin axis by @xmath46 .
the two concentric circles indicate the limiting minimum and maximum emission radii imposed for the altitude - limited models.,width=302 ] this model provides a non - caustic explanation for the emission , and may be viewed as a geometric low - altitude sg model @xcite resembling a hollow cone beam close to the stellar surface .
we modulate the emissivity according to @xmath47 with @xmath48 the distance above the ns surface along a magnetic field line , @xmath49 , and @xmath50 and @xmath51 setting the rate at which the intensity rises and falls along the magnetic field lines .
the peak intensity occurs at a distance @xmath52 ( i.e. , @xmath53 ) along the field lines .
the lasg models have 5 free parameters ( more details in @xcite ) .
in order to statistically pick the best - fit parameters , for the alog and altpc models and for the three msps considered here , we have developed an markov chain monte carlo ( mcmc ) maximum likelihood procedure @xcite .
the gamma - ray lcs are fit using poisson likelihood while the radio lcs are fit using a @xmath54 statistic , and the two values are then combined . for a given parameter state
the likelihood value is calculated by independently optimizing the radio and model normalizations using the _ scipy _ python module and the _ scipy.optimize.fmin_l_fbgs_b _ multivariate , bound optimizer @xcite .
an mcmc involves taking random steps in parameter space and accepting a step based on the likelihood ratio with respect to the previous step @xcite .
the likelihood surfaces can be very multimodal which can lead to poor mixing of the chain and slow convergence .
therefore , we have implemented small - world chain steps @xcite and simulated annealing @xcite to speed up the convergence and ensure that the mcmc fully explores the parameter space and does not get stuck in a local maximum .
we verify that our chains have converged using the criteria proposed by @xcite . in order to balance the gamma - ray and radio contributions to the likelihood
, we have chosen to use a radio uncertainty which is equal to the average relative uncertainty in the on - peak region times the maximum radio value .
the choice of radio uncertainty can strongly affect the best - fit results ; in particular , a smaller uncertainty will decrease the overall likelihood and can , in some cases , lead to a different best - fit geometry which favors the radio lc more strongly . when varying the radio uncertainty by a factor of 2 , the best - fit @xmath7 and @xmath8 values of were found to change by @xmath55 .
for the best - fit @xmath7 and @xmath8 were found to vary by @xmath56 when varying the radio uncertainty .
the best - fit geometry of was found to be the most sensitive to changes in the radio uncertainty , with either the best - fit @xmath7 or @xmath8 value changing by @xmath57 , while the other parameter changed by @xmath58 .
starting from the best - fit parameters found by the mcmc , we produced confidence contours in @xmath7 and @xmath8 by performing likelihood profile scans over the other parameters , allowing for the possibility of finding a better fit .
the uncertainties on @xmath7 and @xmath8 quoted in table [ tab2 ] are approximate 95% confidence level uncertainties .
we can also estimate uncertainties on the emission altitude parameters using the information from the likelihood profile scans which generated the confidence contours .
note that we used manually - selected lc fits for the lasg models .
the effects of letting @xmath59 and @xmath60 be free parameters in the alog / altpc model context , as well as using different fading parameters in our lasg models is discussed elsewhere @xcite .
our best - fit lc parameters are summarized in table [ tab2 ] . as an example , the alog / altpc lc fits for are shown in figure [ fig : j0034_lim ] , while figure [ fig : pc_j0034 ] shows fits for in the case of lasg models .
for both and radio lcs , as the data and model zero phases do not coincide .
this implies that the definition of ` leading peak ' and ` first peak ' do not coincide anymore .
see table [ tab2 ] for more details.,width=302 ] we studied a third subclass of msp lcs for which the and radio profiles are phase - aligned , and the and radio emission should therefore be co - located .
we introduced free parameters @xmath59 and @xmath60 in the alog / altpc models , ( both are free for the radio emission region , but only @xmath37 is free for the the emission region ) and found fits from these models which could reproduce the salient features of the profiles , although not perfectly . as a second option
, we implemented the the lasg models and demonstrated that a modulated emissivity at low altitudes can reproduce main features of the profiles quite well . at the moment
, it is difficult to quantitatively favor one class of models above the other , since searching for optimal lasg lc fits has been done manually .
however , we did calculate the likelihood of the best - fit lasg lcs .
the altpc models provide slightly better lc fits than the alog models , and both of these give better fits than the lasg models for the parameters listed ( see table [ tab2 ] ) . favoring the alog / altpc models over the lasg model therefore implies that the phase - aligned and radio lcs are most probably of caustic origin , produced in the outer magnetosphere , and the radio emission is most likely originating near the light cylinder .
thus , we can now divide the msp population into three subclasses on the basis of their lcs : those with lcs fit by standard og / tpc models , those with phase - aligned lcs fit by alog / altpc or lasg models , and those with lcs fit by pspc models .
radio polarization measurements can be used to give independent constraints on the pulsar viewing geometry @xcite , complementing the model fits , although the traditional rvm @xcite is not valid for the altpc or alog models where the radio peaks are caustics .
furthermore , the rvm is not expected to yield good results in the case of radio cone beam emission in msps , as these beams should suffer significant distortions due to retardation and aberration @xcite .
this may account for the generally poor or non - existent rvm fits of msp polarization data .
caustic models predict rapid pa swings coupled with depolarization @xcite , since although the emission originates from a large range of altitudes and magnetic field orientations , it is restricted to a narrow phase interval to form the peaks .
these features seem to be present in radio polarization measurements of the modeled msps @xcite .
polarization signatures are important to help discriminate between models with caustic emission ( such as occurs in alog / altpc models ) and non - caustic emission ( e.g. , in the lasg model ) .
future studies include development of full radiation models which will be able to reproduce the multiwavelength lc shapes , polarization properties , as well as the energy - dependent behavior of the spectra of the msps .
ways to increase pair production also need to be found , which may include investigation of offset - pc dipole magnetic fields @xcite and higher - multipole magnetic fields near the ns surface @xcite .
@xmath37 & @xmath35 & @xmath36 & @xmath61 + + alog & 12@xmath62 & 69@xmath63 & 0.9@xmath64 & 0.2@xmath65 & 1.1@xmath66 & 96.1 + altpc & 30@xmath67 & 70@xmath68 & 0.9@xmath69 & 0.7@xmath70 & 0.8@xmath71 & 87.0 + lasg1 & 10 & 34 & & & & 97.3 + lasg2 & 10 & 37 & & & & 98.7 + + alog & 84@xmath72 & 84@xmath73 & 1.0@xmath74 & 0.6@xmath69 & 0.9@xmath69 & 130.9 + altpc & 75@xmath75 & 80@xmath73 & 1.0@xmath76 & 0.7@xmath77 & 0.9@xmath74 & 126.3 + lasg1 & 30 & 32 & & & & 146.9 + lasg2 & 35 & 25 & & & & 154.6 + + alog & 31@xmath78 & 89@xmath79 & 1.1@xmath80 & 0.7@xmath69 & 0.9@xmath74 & 128.3 + altpc & 47@xmath81 & 85@xmath82 & 1.2@xmath66 & 0.8@xmath69 & 1.0@xmath74 & 123.7 + lasg1 & 20 & 43 & & & & 129.0 + lasg2 & 25 & 45 & & & & 141.7 + cv is supported by the south african national research foundation .
akh acknowledges support from the nasa astrophysics theory program .
cv , tjj , and akh acknowledge support from the _ fermi _ guest investigator program as well as fruitful discussions with dick manchester and matthew kerr .
part of this work was performed at the naval research laboratory and is sponsored by nasa dpr s-15633-y .
the @xmath83 lat collaboration acknowledges support from a number of agencies and institutes for both development and the operation of the lat as well as scientific data analysis .
these include nasa and doe in the united states , cea / irfu and in2p3/cnrs in france , asi and infn in italy , mext , kek , and jaxa in japan , and the k. a. wallenberg foundation , the swedish research council and the national space board in sweden . additional support from inaf in italy and cnes in france for science analysis during the operations phase is also gratefully acknowledged .
99 abdo , a. a. et al .
2009a , science 325 , 848 abdo , a. a. et al .
2010a , apjs , 187 , 460 abdo , a. a. et al .
2010b , apj , 712 , 957 arons , j. 1983 , apj , 266 , 215 backer , d. c. , kulkarni , s. r. , heiles , c. , davis , m. m. , & goss , w. m. 1982 , nature , 300 , 615 bailes , m. , et al .
1994 , apj , 425 , l41 bai , x .-
n . , & spitkovsky , a. 2010 , 715 , 1270 bell , j. f. , kulkarni , s. r. , bailes , m. , leitch , e. m. , & lyne , a. g. 1995 , apj , 452 , l121 blaskiewicz , m. , cordes , j. m. , wasserman , i. 1991 , apj , 370 , 643 cheng , k. s. , ho , c. , & ruderman , m. 1986a , apj , 300 , 500 cusumano , g. et al .
2003 , a&a , 410 , l9 daugherty , j. k. , & harding , a. k. 1982 , apj , 252 , 337 daugherty , j. k. , & harding , a. k. 1996 , apj , 458 , 278 deutsch , a. j. 1955 , ann .
dastrophys . , 18 , 1 du , y. j. , qiao , g. j. , han , j. l. , lee , k. j. , & xu , r. x. , 2010 , mnras , 406 , 2671 dyks , j. , & rudak , b. 2003 , apj , 598 , 1201 dyks , j. , harding , a. k. , & rudak , b. 2004 , apj , 606 , 1125 fierro , j. m. et al . 1995 , apj , 447 , 807 frackowiak , m. , & rudak , b. 2005 , adv .
space res .
, 35 , 1152 fruchter , a. s. , stinebring , d. r. , & taylor , j. h. 1988 , nature , 333 , 237 gelman , a. , & rubin , d. 1992 , stat .
, 7 , 457 guan , y. et al .
2006 , stat . comp .
, 16 , 193 guillemot , l. et al .
2011a , in press ( arxiv:1101.0669 ) guillemot , l. et al .
2011b , submitted .
harding , a. k. , usov , v. v. , & muslimov , a. g. 2005 , apj , 622 , 531 harding , a. k. , & muslimov , a. g. 2011 , apj , 726 , l10 hastings , w. k. 1970 , biometrika , 57 , 97 hobbs , g. , lorimer , d. r. , lyne , a. g. , & kramer , m. 2005 , mnras , 360 , 974 huang , h. h. , & becker , w. 2007 , a&a , 463 , l5 johnson , t. j. , harding , a. k. , & venter , c. 2011 , in preparation .
kinkhabwala , a. & thorsett , s. e. 2000 , apj , 535 , 365 kniffen , d. a. , hartman , r. c. , thompson , d. j. , bignami , g. f. , & fichtel , c. e. 1974 , nature , 251 , 397 manchester , r. n. 2005 , ap&ss , 297 , 101 manchester , r. n. , hobbs , g. b. , teoh , a. , & hobbs , m. 2005 , aj , 129 , 1993 marinari , e. , & parisi , g. 1992 , europhys .
, 19 , 451 morini , m. 1983 , mnras , 202 , 495 muslimov , a. g. , & harding , a. k. 2003 , apj , 588 , 430 muslimov , a. g. , & harding , a. k. 2004b , apj , 617 , 471 muslimov , a. g. , & harding , a. k. 2009 , apj , 692 , 140 radhakrishnan , v. , & cooke , d. j. 1969 , apj , 3 , 225 ransom , s. m. et al .
2011 , apj , 727 , l16 ravi , v. , manchester , r. n. , & hobbs , g. 2010 , apj , 716 , l85 romani , r. w. 1996 , apj , 470 , 469 romani , r. w. et al .
2011 , in press spitkovsky , a. 2006 , apj , 648 , l51 stairs , i. h. , thorsett , s. e. , & camilo , f. 1999 , apjs , 123 , 627 thorsett s. e. & stinebring , d. r. 1990 , apj , 361 , 644 van paradijs , et al .
1988 , nature , 334 , 684 venter , c. , & de jager , o.c .
2005 , apj , 619 , l167 venter , c. , harding , a. k. , & guillemot , l. 2009 , apj , 707 , 800 venter , c. , johnson , t. j. , & harding , a. k. 2011 , apj , accepted verbiest , j. p. w. et al .
2009 , mnras , 400 , 951 weltevrede , p. et al .
2010 , apj , 708 , 1426 yan , w. et al .
2011 , accepted for publication in mnras ( arxiv:1102.2274 ) zavlin , v. e. 2006 , apj , 638 , 951 zhang , l. , & cheng , k. s. 2003 , a&a , 398 , 639 zhang , l. et al . 2007 , apj , 666 , 1165 zhu , c. , et al .
1997 , acm trans . on math . | the population of millisecond pulsars ( msps ) detected by the _ fermi _ large area telescope ( lat ) has been steadily increasing . a number of the more recent detections , including ( b1937 + 21 ; the first msp ever discovered ) , ( b1957 + 20 ; the first black widow system ) , and exhibit an unusual phenomenon : nearly phase - aligned radio and gamma - ray light curves ( lcs ) . to account for the phase alignment , we explore geometric models where both the radio and gamma - ray emission originate either in the outer magnetosphere near the light cylinder ( @xmath0 ) or near the polar caps ( pcs ) .
we obtain reasonable fits for the first three of these msps in the context of `` altitude - limited '' outer gap ( alog ) and two - pole caustic ( altpc ) geometries .
the outer magnetosphere phase - aligned models differ from the standard outer gap ( og ) / two - pole caustic ( tpc ) models in two respects : first , the radio emission originates in caustics at relatively high altitudes compared to the usual low - altitude conal radio beams ; second , we allow the maximum altitude of the gamma - ray emission region as well as both the minimum and maximum altitudes of the radio emission region to vary within a limited range .
alternatively , there also exist phase - aligned lc solutions for emission originating near the stellar surface in a slot gap ( sg ) scenario ( `` low - altitude slot gap '' ( lasg ) models ) .
we find best - fit lcs using a markov chain monte carlo ( mcmc ) maximum likelihood approach @xcite .
our fits imply that the phase - aligned lcs are likely of caustic origin , produced in the outer magnetosphere , and that the radio emission may come from close to @xmath0 .
we lastly constrain the emission altitudes with typical uncertainties of @xmath1 .
our results describe a third gamma - ray msp subclass , in addition to the two ( with non - aligned lcs ) previously found @xcite : those with lcs fit by standard og / tpc models , and those with lcs fit by pair - starved polar cap ( pspc ) models . |
the initial mass function of primordial black holes ( or , it is better to say , the initial mass spectrum ) is determined , first of all , by the peculiarities of a gravitational collapse near the threshold of a black hole formation . according to analytic calculations of seventies @xcite
a critical size of the density contrast needed for the pbh formation , @xmath2 , is about @xmath3 . besides
, it was argued that all pbhs have mass roughly equal to the horizon mass at a moment of the formation , independently of the perturbation size .
however , it was shown recently that near the threshold of black hole formation the gravitational collapse behaves as a critical phenomenon @xcite . in this case
the initial mass function will be quite different from the analogous function in the analytic calculations @xcite .
the main feature is that the pbh mass formed depends on the size of the fluctuation @xcite , @xmath4 in this formula @xmath5 is the horizon mass at the time the fluctuation enters the horizon ; @xmath2 , @xmath6 , @xmath1 are parameters of a concrete model of the critical collapse @xcite
. it is seen from eq.([1 ] ) that pbh mass may be arbitrarily small independently of the value of @xmath5 . besides , the value of the critical overdensity , @xmath2 , in such models is typically @xmath7 , i.e. , about a factor of 2 larger than the value found analytically @xcite .
the second important ingredient of a pbh initial mass function calculation is a taking into account the spread of horizon masses at which pbhs are formed .
this problem exists independently of a nature of gravitational collapse .
it is assumed very often that the majority of the black hole formation occurs at the shortest possible scale . in particular , the authors of ref.@xcite determined the pbh initial mass function under the assumption that all pbhs form at the same horizon mass .
the accuracy of such approximation was studied in the work @xcite using the excursion set formalism , and it was found that it is good enough , at least in the case of power - law density perturbation spectra with the spectral index close to @xmath8 ( i.e. , slightly `` blue '' spectra , with the value of @xmath9 satisfying observational constraints ) .
recently , in the works of kim @xcite and present authors @xcite the expression for the pbh mass spectrum ( for the critical collapse case ) was obtained which takes into account the accumulation of pbhs formed at all epochs after the beginning of the ordinary radiation dominated era .
the derivation is based on the press- schechter formalism @xcite . resulting formula for the pbh mass spectrum , taking into account the spread of horizon masses , depends on the horizon crossing amplitude @xmath10 which is a function of horizon mass @xmath5 . in the present paper
we assume a power law of primordial density perturbations for all scales , and in this case one has ( @xmath11 ) @xmath12 @xmath13 in these formulas @xmath14 is the non - smoothing horizon crossing amplitude ( which is time - independent ) , @xmath15 is the present horizon mass , @xmath16 is the horizon mass at the moment of matter - radiation equality , @xmath17 is the numerical coefficient arising as a result of the smoothing procedure ( the top - hat form of the window - function was used in all numerical calculations ) .
the amplitude is normalized on cobe data ( see @xcite for detailes and references ) .
the expression for the pbh mass spectrum is @xcite @xmath18 @xmath19 here , @xmath20 amd @xmath21 are the background density and the horizon mass at @xmath22 ( @xmath22 is the moment of a beginning of growth of the density fluctuations ) . the value of pbh mass @xmath23 is connected with @xmath5 by the relation ( [ 1 ] ) . the upper limit of integration in eq.([4 ] )
is determined by the expression @xmath24.\ ] ] note , that in ref.@xcite the upper limit of integration in the expression for @xmath25 was erroneously put equal to 1 .
all numerical calculations were carried out with the following values of parameters @xcite : @xmath26 the known formula @xcite for the pbh mass spectrum in the standard collapse case can be obtained from eq.([4 ] ) by using the substitutions @xcite @xmath27 and the approximate relation @xmath28 substituting eqs.([6b ] ) and ( [ 6c ] ) in eq.([4 ] ) one obtains @xmath29 in this case the relation between @xmath23 and @xmath30 is independent on @xmath31 : @xmath32 we carried out in this work the calculations of the neutrino background spectra from pbh evaporations using the mass spectrum given by eq.([4 ] ) and , in parallel , the mass spectrum from ref.@xcite , @xmath33 ^ 2}{2\sigma_h^2}}.\ ] ] in this expression the amplitude @xmath10 is determined by the same formula ( [ 2 ] ) , as above , but with the constant value of @xmath5 , @xmath34 , in accordance with the assumption of authors of ref.@xcite that all pbhs form at the smallest scale .
our approach has two free parameters : the spectral index @xmath9 , giving the perturbation amplitude ( normalized on cobe data on large scales ) and @xmath22 , the moment of time just after reheating , from which the process of pbhs formation started .
the value of @xmath22 is connected with a value of the reheating temperature , @xmath35 ( @xmath36 is the number of degrees of freedom in the early universe ) . typical results of pbh mass spectrum calculations are shown on fig.1 .
we use , for convenience , the following abbreviations : kl , nj and bk signify , correspondingly , the pbh mass spectra calculated by formulae of kim and lee @xcite ( eq.([6d ] ) ) , niemeyer and jedamzik @xcite ( eq.([6 ] ) ) and bugaev and konishchev @xcite ( eq.([4 ] ) ) .
one can see from fig.1 that at large values of the spectral index there is a clear difference in a behavior of two mass spectra , nj and bk , based both on a picture of the critical collapse : bk spectrum is less steep at @xmath37 . of course , this difference is not so spectacular in cases when the spectral index @xmath9 is more close to 1 ( it is proved by another method in ref.@xcite ) .
the common feature of nj and bk spectra is the maximum at @xmath38 . besides
, both spectra have same slope in the region @xmath39 .
evolution of a pbh mass spectrum due to the evaporation leads to the approximate expression for this spectrum at any moment of time : @xmath40 where @xmath41 accounts for the degrees of freedom of evaporated particles and , strictly speaking , is a function of a running value of the pbh mass @xmath42 . in all our numerical calculations we use the approximation @xmath43 where @xmath44 is the value of @xmath23 in the initial mass spectrum corresponding to the maximum of this spectrum .
special study shows that errors connected with such approximation are rather small .
the expression for the spectrum of the background radiation is @xcite @xmath45 \cdot f(e\cdot ( 1+z),m)e^{-\tau(e , z)}\nonumber\\ \\ \equiv \int f(e , z)d \log_{10 } ( z+1).\nonumber\end{aligned}\ ] ] in this formula @xmath46 is a total instantaneous spectrum of the background radiation ( neutrinos or photons ) from a black hole evaporation .
it includes the pure hawking term and contributions from fragmentations of evaporated quarks and from decays of pions and muons ( see @xcite for details ) .
the exponential factor in eq.([10 ] ) takes into account an absorption of the radiation during its propagation in space .
the processes of the neutrino absorption are considered , in a given context , in ref.@xcite . in last line of eq.([10 ] ) we changed the variable @xmath47 on @xmath48 using , for simplicity , the flat model with @xmath49 for which @xmath50 several examples of @xmath48-distributions ( the integrands of the integral over @xmath48 in eq.([10 ] ) ) are shown on fig.2 . again one can see the rather strong difference of two cases : the existence of a tail of heavy masses in bk spectrum leads to a relative enhancement of low @xmath48-contributions .
the effect becomes more distinct with a rise of the reheating temperature .
the sharp cut - off of all @xmath48-distributions near @xmath51 is entirely due to the neutrino absorption @xcite .
the shrinkage of @xmath48-distributions at large @xmath52 in nj - case is due to the absence of large masses in the spectrum ( pbhs of small masses evaporated earlier , and their radiation today is more redshifted ) .
some results of calculations of the neutrino background spectra are shown on figs.3 - 5 .
the functional form of these spectra is qualitatively the same as in the standard collapse case @xcite : @xmath53 law at large neutrino energies and a flat part at low energies with the crossover energy depending on @xmath52 .
the only new feature is the sharp steepening of the spectra at high values of @xmath52 in nj case ( fig.4 ) connected with the corresponding shrinkage of the @xmath48-range .
fig.5 shows the sensitivity of background neutrino intensities to a chosen value of the spectral index @xmath9 , at a fixed @xmath52 .
one can see from this figure that the neutrino background flux from pbh evaporations is comparable with the atmospheric neutrino flux at @xmath54 if @xmath55 ( at @xmath56 ) .
such large values of the spectral index are , probably , excluded by the recent large - scale experiments .
next figure shows the constraints on the spectral index following from data of neutrino experiments , as a function of the reheating temperature . as in our previous work , we used for a description of these constraints the data of the kamiokande atmospheric neutrino experiment @xcite and the experiment on a search of an antineutrino flux from the sun @xcite ( see @xcite for details ) . for completeness
, we show on the same figure the constraints following from extragalactic diffuse gamma ray data ; part of them ( nj case ) is in qualitative agreement with the results of ref.@xcite .
it is seen that the constraints are much more weak in nj case ( at @xmath57 ) .
it is again the result of an absence of the tail of large masses in nj spectrum , leading to a shrinkage of the @xmath48-distriburions and to a steepening of the background neutrino spectra .
\1 . the initial concentration of pbh s is larger in the case of the standard picture of the gravitational collapse ( as compared with the critical collapse case ) .
it is due to relatively small value of the critical overdensity ( @xmath58 ) . in the region near the maximum of pbh
mass spectra the ratio of intensities is given by the approximate relation @xmath59 correspondingly , densities of the neutrino and gamma backgrounds produced by pbh s evaporations are smaller and spectral index constraints are weaker in the critical collapse case .
the summation over all epochs of pbh s production ( i.e. , the transition from nj to bk case ) leads to an appearing of the tail of large pbh masses in the pbh mass spectrum .
this tail is especially sufficient if the spectral index @xmath9 is relatively large .
spectral index constraints following from the comparison of neutrino background predictions with existing data of neutrino experiments are rather weak if the purely power law of primordial density fluctuations is assumed for all scales ( and , correspondingly , the normalization on cobe data is used ) .
however , it does not mean that neutrino background intensities from pbh s evaporations can not be noticeable . if , for example , the spectrum of primordial density fluctuations is a combination of two simple power law spectra @xcite , i.e. , @xmath61 then the small value of @xmath62 ( @xmath63 ) which follows , in particular , from cobe data , does not contradict with the possibility that @xmath64 is large and , correspondingly , @xmath10 at small scales is also large . if @xmath65 and @xmath66 are known from somewhere , one can easily obtain from the neutrino data the constraints on the small scale spectral index @xmath67 , in the same manner as it is done above for the all scale spectral index @xmath9 .
several cosmological models of such kind , which predict a large density fluctuation amplitude just on small scales ( while the amplitude on large scales is constrained by the small cmbr anisotropy ) appeared recently ( see , e.g. , @xcite ) .
b. j. carr and s. w. hawking , mon . not .
* 168 * , 399 ( 1974 ) .
b. j. carr , astrophys .
j. * 201 * , 1 ( 1975 ) .
j. c. niemeyer and k. jedamzik , phys .
lett . * 80 * 5481 ( 1998 ) .
j. c. niemeyer and k. jedamzik , phys .
d * 59 * 124013 ( 1999 ) .
a. m. green and a. r. liddle , phys .
d * 60 * , 063509 ( 1999 ) .
h. i. kim , phys .
d * 62 * 063504 ( 2000 ) .
e. v. bugaev and k. v. konishchev , _ astro - ph/_0005295 .
e. v. bugaev and k. v. konishchev , _ astro - ph/_0103265 . w. h. press and p. schechter , astrophys .
j. * 187 * , 425 ( 1974 ) .
h. i. kim and c. h. lee , phys .
d * 54 * , 6001 ( 1996 ) .
e. v. bugaev and v. a. naumov , phys .
b * 232 * , 391 ( 1989 ) .
k. s. hirata _
et al . _ ,
b * 280 * , 146 ( 1992 ) .
m. aglietta _
et al . _ , pisma zh .
. fiz . * 63 * , 753 ( 1996 ) .
g. d. kribs , a. k. leibovich , and i. z. rothstein , phys . rev . * 60 * , 103510 ( 1999 ) .
l. randall , m. soljacic , and a. h. guth , nucl .
b * 472 * , 377 ( 1996 ) .
j. garcia - bellido , a. d. linde , and d. wands , phys .
d * 54 * , 6040 ( 1996 ) . | comparative analysis of different types of initial mass spectra of primordial black holes ( pbhs ) is carried out .
it is assumed that primordial density fluctuations have power law spectrum and the spectral index is constant throughout all scales .
it is shown that densities of background radiations ( @xmath0 , @xmath1 ) from pbh evaporations strongly depend on the type of gravitational collapse and on the way of taking into account the spread of horizon masses at which pbhs can form . constraints on spectral index values based on pbh evaporation process and on atmospheric and solar neutrino data are obtained . |
applications of perturbative qcd to deeply virtual compton scattering and hard exclusive electroproduction processes @xcite require a generalization of usual parton distributions for the case when long - distance information is accumulated in nonforward matrix elements @xmath8 of quark and gluon light - cone operators . as argued in refs .
@xcite , such matrix elements can be parametrized by two basic types of nonperturbative functions . with @xmath9
taken in the lightcone `` minus '' direction , the double distributions ( dds ) @xmath1 specify the light - cone `` plus '' fractions @xmath10 and @xmath11 of the initial hadron momentum @xmath12 and the momentum transfer @xmath13 carried by the initial parton .
though @xmath9 is an integration variable , only one direction on the lightcone ( specified by external momenta ) is important for the lightcone - dominated processes . in other words ,
only the lightcone plus direction of the hadron @xmath12 and @xmath13 momenta are essential for such processes . by definition ,
the dds @xmath1 do not depend on the @xmath14 ratio . on the other hand , treating the proportionality coefficient as an independent parameter : @xmath15 , one can introduce an alternative description in terms of the nonforward parton distributions @xmath2 with @xmath16 being the total fraction of the initial hadron momentum taken by the initial parton .
the shape of the functions @xmath2 explicitly depends on the parameter @xmath5 characterizing the skewedness of the relevant nonforward matrix element .
this parametrization of nonforward matrix elements by @xmath2 is similar to that proposed originally by x. ji @xcite who introduced off - forward parton distributions ( ofpds ) @xmath17 .
the latter are close to functions considered earlier in ref .
the functions @xmath17 have a simple relation to nonforward distributions ( nfpds ) @xmath2 , while the non - diagonal distributions @xmath18 discussed by collins , frankfurt and strikman @xcite essentially coincide with @xmath19 ( see ref .
@xcite for details ) . the basic distinction between our approach and those of refs .
@xcite is that we treat the double distributions @xmath1 as the primary objects of the qcd analysis producing the nonforward distributions @xmath2 ( and other types of distributions ) after an appropriate integration .
the formalism of double distributions provides a rather effective tool for studying some general ( e.g. , spectral ) properties of nfpds and it allows to find analytic solutions of evolution equations @xcite . incorporating symmetries of dds @xcite imposes rather strong restrictions on realistic models of nfpds .
a possible strategy for a self - consistent model building is to use nonperturbative or phenomenological approaches ( mit bag , quark models , qcd sum rule ideas , etc . ) to construct double distributions at low normalization point and then evolve them to higher @xmath20 values .
the evolution equation for the nonsinglet quark double distribution was derived in ref.@xcite , where its analytic solution was also given .
evolution of the gluon distribution in pure gluodynamics was discussed in refs.@xcite . in this paper , we present also a full set of evolution equations for the flavor - singlet case and derive a solution following the method of refs.@xcite .
an independent study of singlet evolution based on our approach was performed in ref.@xcite .
evolution equations for various versions of nonforward distributions can be found in @xcite .
a convenient way to obtain the relevant evolution kernels is to use the universal light - ray evolution kernels @xcite .
the evolution of nonforward distributions was studied numerically in refs.@xcite . in the present paper
, we incorporate the spectral and symmetry properties of double distributions to construct some simple models for dds . using the relations between dds and nfpds / ofpds
, we derive models for the latter and show that using the formalism of double distributions we can easily explain characteristic qualitative and quantitative features of the evolution of nonforward distributions observed in refs.@xcite .
the kinematics of the amplitudes of the dvcs process @xmath21 and hard electroproduction @xmath22 can be specified by the initial nucleon momentum @xmath12 , the momentum transfer @xmath23 and the momentum @xmath24 of the final photon or meson . to get a bjorken - type scaling limit
, one should also keep the invariant momentum transfer @xmath25 small compared to the virtuality @xmath26 of the initial photon and the energy invariant @xmath27 .
the essential features of the hard electroproduction processes ( dvcs included ) can be most easily demonstrated if we set @xmath28 , @xmath29 , @xmath30 and use @xmath12 , @xmath24 as the basic light - cone ( sudakov ) 4-vectors .
it is easy to see that the requirement @xmath31 reduces in this limit to the condition @xmath32 which can be satisfied only if the two lightlike momenta @xmath12 and @xmath13 are proportional to each other : @xmath33 , where @xmath5 coincides with the bjorken variable @xmath34 .
the latter satisfies the constraint @xmath35 . for small but finite @xmath36 and @xmath37 ,
the momentum transfer @xmath13 still must have a non - zero plus component @xmath38 .
it also may have a transverse component @xmath39 . in the pqcd factorization treatment of hard electroproduction processes ,
the nonperturbative information is accumulated in the nonforward matrix element @xmath40 ( we use here @xmath41 as a generic notation for quark ( @xmath42 ) or gluonic ( @xmath43 ) fields ) .
it depends on the relative coordinate @xmath9 through three invariant variables @xmath44 and @xmath45 . in the forward case , when @xmath46 , one gets the usual parton distributions by fourier transforming the light - cone projected ( i.e. , @xmath47 ) matrix element with respect to @xmath48 . in the nonforward case , we can try to start with the general fourier representation @xmath49 with respect to all three @xmath9-dependent invariants .
the fourier transform @xmath50 can be called a _
triple _ distribution .
note that the generous @xmath51 limits for all three variables @xmath52 serve for a most general function of @xmath53 and @xmath45 .
however , incorporating information that the fourier transformation is written for a function @xmath54 given by feynman integrals having specific causality properties , one arrives at more narrow limits : @xmath55 runs from 0 to @xmath56 , @xmath57 is between @xmath58 and 1 while @xmath59 is between @xmath60 and 1 ( this was proven in @xcite for any feynman diagram using the approach of ref .
@xcite ) . to interpret the @xmath57-variable as the fraction of the initial momentum @xmath12 carried by the relevant parton , it makes sense to separate integration over positive and negative @xmath57 components and redefine @xmath61 and @xmath62 for the negative @xmath57 component .
after that , the @xmath57-variable is always positive and @xmath57 and @xmath59 are further constrained by inequality @xmath63 @xcite .
these spectral conditions can be summarized by the following representation @xmath64 \theta(x+y \leq 1 ) \,dx \ , dy \ , , \nonumber \end{aligned}\ ] ] in which @xmath65 and @xmath66 result from positive-@xmath57 and negative-@xmath57 components of @xmath50 respectively . in particular , for quark operators , @xmath67 can be interpreted as the quark distribution while @xmath68 as the antiquark one ( a more detailed discussion is given in the next section ) .
similarly , @xmath59 and @xmath69 can be interpreted as the fractions in which the momentum transfer @xmath13 is shared among the two fields of the composite operator @xmath70 .
finally , the @xmath55 variable characterizes the virtuality of these fields . for a light - cone dominated process ,
the leading term is given by the @xmath71 limit of the nonforward matrix element , i.e. by zeroth moment of @xmath72 with respect to @xmath55 @xmath73 where @xmath74 is the _
double distribution_. for a lightlike interval @xmath47 , one can treat @xmath9 as having only light - cone `` minus '' component , and then the scalar products @xmath44 project out the `` plus '' components of general ( non - lightlike ) momenta @xmath12 and @xmath13 .
this allows to give a parton interpretation of @xmath74 as a probability amplitude for the active parton to carry fractions @xmath10 and @xmath11 of the plus components of the external momenta @xmath13 and @xmath12 . though the momenta @xmath75 and @xmath76 can be treated as proportional to each other @xmath77 , @xmath75 and @xmath76 specify the `` + ' '
-momentum flow in two different channels . for @xmath78 ,
the net `` + ' ' -momentum flows only in the @xmath79-channel and the total `` + ' ' -momentum entering into the composite operator vertex is zero . in this case , the matrix element is analogous to a distribution function . the partons entering the composite vertex then carry the fractions @xmath80 of the initial proton momentum ( @xmath81 ) .
when @xmath82 is negative , we interpret the parton as belonging to the final state to secure that the integral always runs over the segment @xmath83 . in this parton picture , the spectators take the remaining momentum @xmath84 . on the other hand , if the total `` + ' ' -momentum flowing through the composite vertex is @xmath76 , the matrix element has the structure of a distribution amplitude in which the momentum @xmath76 splits into the fractions @xmath11 and @xmath85 carried by the two fields that appear in the vertex . in a combined situation ,
when both @xmath75 and @xmath76 are nonzero , the initial parton takes @xmath86 , while the final one carries the momentum @xmath87 . for @xmath46
, we get the forward matrix element which is parametrized by the usual parton distributions @xmath4 .
this gives reduction relations @xcite connecting double distributions with the usual ones ( see eqs.([34]),([35 ] ) below ) .
for quark operators , the double distributions are defined by the following representation @xcite : @xmath88 for parton helicity - averaged ones and by @xmath89 in the parton helicity - sensitive case . here and in what follows
we adhere to the `` bar '' convention @xmath90 , @xmath91 for momentum fractions and use the `` russian hat '' notation @xmath92 . as usual , @xmath93 are the dirac spinors for the nucleon . in this definition ,
we explicitly separate quark and antiquark components of the double distribution .
note that such a separation is unambiguous : in the fourier representation , it is completely determined by the sign of the @xmath57-term in the exponential . to clarify the physical meaning of separating the dds into two components ,
it is instructive to consider the forward limit @xmath46 in which the matrix element is parametrized by usual parton densities , e.g. , in the helicity averaged case @xmath94 the exponential factors accompanying the quark and antiquark distributions reflect the fact that the field @xmath95 appearing in the operator @xmath96 consists of the quark annihilation operator ( quark with momentum @xmath97 comes into this point ) and the antiquark creation operator ( i.e. , antiquark with momentum @xmath97 goes out of this point ) . to get the relative signs with which quark and antiquark distributions appear in these definitions
, we should take into account that antiquark creation and annihilation operators appear in @xmath96 in the opposite order . comparing the expression ( [ 33 ] ) with the @xmath98 limit of the definitions for dds
, we obtain `` reduction formulas '' relating the two components of the double distributions to the quark and antiquark parton densities , respectively : @xmath99 and similarly for the helicity - sensitive case : @xmath100 the reduction formulas tell us that integrating the double distribution @xmath101 over a vertical line @xmath102const in the @xmath103-plane , one gets the quark density @xmath104 while integrating its counterpart @xmath105 gives the antiquark density @xmath106 .
this is an illustration of our statement that @xmath107 and @xmath108 are independent functions . in particular
, @xmath107 contains the valence component ( reducing to @xmath109 ) absent in @xmath108 .
our definitions ( [ 31 ] ) , ( [ 32 ] ) reflect the results of the @xmath110-representation analysis @xcite that the plus component of the momentum of the particle ( either quark or antiquark ) _ going out _ of the hadronic blob can be written as @xmath111 with both @xmath57 and @xmath59 positive and @xmath112 .
this is in full compliance with the parton model based expectation that the initial hadron splits into an active parton and spectators which both carry positive fractions of its plus momentum . to show the positivity of the plus momentum component for spectators
, we should explicitly take into account that , in the kinematics of dvcs and hard electroproduction processes , the plus component of the momentum transfer @xmath23 is positive @xmath113 . requiring that the plus component of the final hadron momentum is also positive ,
we conclude that @xmath114 .
hence , @xmath115 ( since @xmath114 and @xmath63 ) , i.e. , the plus component of the momentum carried by spectators is also positive . on the other hand , the parton _ `` going back '' _ has the momentum whose plus component @xmath116 may be either positive or negative , depending on the relationship between @xmath57 , @xmath59 and @xmath5 . when @xmath117 is negative , one may wish to interpret such a parton as an antiparton leaving the hadron together with the initial parton .
one should remember , however , that the double distributions @xmath1 `` know nothing '' about the magnitude of the skewedness @xmath5 : they are universal functions describing flux of @xmath75 and @xmath76 independently of what the ratio @xmath14 might be . as we explained above , the quark dds
are unambiguously divided into two separate components @xmath118 and @xmath119 , but there is no further subdivision inside them based on interrelation between the values of @xmath57 and @xmath59 . in a similar way
, we can introduce double distributions for the gluons @xmath120
@xmath121 there are no `` antigluons '' , so the positive-@xmath57 and negative-@xmath57 parts are described by the same function .
note that our definition of the gluon double distributions here differs from that used in our earlier papers@xcite by an extra factor of @xmath57 in its right hand side .
this form is more convenient for applications of the method of refs .
@xcite to solve evolution equations for double distributions in the singlet case .
the choice made above corresponds also to the simplest form of the reduction formulas @xmath122 another ambiguity in the definition of the gluon double distribution is related to the overall factor @xmath123 in the rhs of eqs.([36 ] ) and ( [ 37 ] ) . instead of it , we could take , e.g. , @xmath124 or @xmath125 ( such a choice is utterly inconvenient for taking the forward limit ) or @xmath126 where @xmath127 is a symmetric combination of the initial and final momenta .
the latter choice ( made in ref.@xcite ) is more convenient for the studies of symmetry properties of the gluon dds .
our choice made in eqs .
( [ 36 ] ) and ( [ 37 ] ) simplifies the expressions for off - diagonal ( @xmath128 and @xmath129 ) evolution kernels ( see eq.([420 ] ) below ) .
the flavor - singlet quark operators @xmath130 \label{39}\ ] ] and @xmath131 \ , \label{310}\ ] ] are expressed in terms of double distributions @xmath132 , @xmath133 , etc . specified by @xmath134 @xmath135 they are given by the sum of `` @xmath136 '' distributions : @xmath137
the qcd perturbative expansion for the matrix element in eq.([21 ] ) generates @xmath138 terms . as a result ,
limit @xmath71 is singular and the distributions @xmath1 , etc .
, contain logarithmic ultraviolet divergences which require an additional @xmath139-operation characterized by some subtraction scale @xmath140 : @xmath141 .
the @xmath140-dependence of @xmath142 is governed by the evolution equation @xmath143 where @xmath144 .
a similar set of equations , with kernels denoted by @xmath145 prescribes the evolution of the parton helicity sensitive distributions @xmath146 .
since the evolution kernels do not depend on @xmath36 , from now on we will drop the @xmath36-variable from the arguments of @xmath147 in all cases when this dependence is inessential ( likewise , the @xmath140-variable will be ignored in our notation when it is not important ) .
since integration over @xmath59 converts @xmath148 into the parton distribution function @xmath149 , whose evolution is described by the dglap equations @xcite @xmath150 the kernels @xmath151 must satisfy the reduction relation @xmath152 aternatively , integration over @xmath57 converts @xmath148 into an object similar to a meson distribution amplitude ( da ) , so one may expect that the result of integration of @xmath151 over @xmath57 should be related to the kernels governing the da evolution @xcite . for the diagonal kernels the relations are rather simple : @xmath153 for the quark kernel and a slightly more complicated expression for the gluon kernel : @xmath154
the @xmath155 factor appears because of the extra @xmath57 which was added in the definition of the gluon dd by analogy with the definition for the usual gluon densities .
the nondiagonal kernels @xmath156 and @xmath157 obey more complicated reduction formulas ( see the appendix ) . the reduction properties of the diagonal evolution kernels can be illustrated using the explicit form of the @xmath158-kernel : @xmath159 -2\delta(1-x/\xi ) \delta(y-\eta ) \int_0 ^ 1 \frac{z}{1-z } \ , dz \right \}. \nonumber\end{aligned}\ ] ] here the last ( formally divergent ) term , as usual , provides the regularization for the @xmath160 singularities present in the kernel .
this singularity can be also written as @xmath161 for the term containing @xmath162 and as @xmath163 for the term with @xmath164 .
depending on the chosen form of the singularity , incorporating the @xmath165 term into a plus - type distribution , one should treat @xmath9 as @xmath155 , @xmath166 or @xmath167 .
one can check that integrating @xmath168 over @xmath59 or @xmath57 gives the dglap splitting function @xmath169 and the da evolution kernel @xmath170 , respectively : @xmath171 \theta(y \leq \eta ) + \left ( \frac{\bar y}{\bar \eta } \right ) \left [ 1 + \frac1{y- \eta } \right ]
\theta(y \geq \eta ) \right \}_+ \ , .
\label{48 } \end{aligned}\ ] ] here , `` + '' denotes the standard `` plus '' regularization @xcite .
a convenient way to get explicit expressions for @xmath151 is to extract them from the kernels @xmath172 describing the evolution equations for the light - ray operators @xcite @xmath173 since the definitions of the gluon distributions @xmath174 , @xmath175 contain an extra @xmath48 factor on the right - hand side , which results in the differentiation @xmath176 of the relevant kernel , it is convenient to proceed in two steps
. first , we introduce the auxiliary kernels @xmath177 directly related by @xmath178 to the light - ray evolution kernels @xmath179 .
the second step is to get the @xmath139-kernels using the relations @xmath180 hence , to obtain @xmath181 , we should integrate @xmath182 with respect to @xmath57 .
we fix the integration ambiguity by the requirement that @xmath181 vanishes for @xmath183 .
then @xmath184 this convention guarantees a simple relation ( [ 43 ] ) to the dglap kernels .
explicit expressions for the evolution kernels and discussion of evolution equations in the singlet case is given in the appendix ( see also ref .
the structure of the integrals relating double distributions with the usual ones @xmath185 [ where @xmath186 ) ] has a simple graphical illustration ( see fig.[fig:1]a ) .
the dds @xmath187 live on the triangle defined by @xmath188 . integrating @xmath187 over a line parallel to the @xmath59-axis
, we get @xmath4 . the reduction formulas and the interpretation of the @xmath57-variable of @xmath187 as a fraction of the @xmath75 momentum suggests that the profile of @xmath187 in the @xmath57-direction is basically driven by the shape of @xmath4 . on the other hand ,
the profile in the @xmath59-direction characterizes the spread of momentum induced by the momentum transfer @xmath76 .
hence , the @xmath59-dependence of @xmath187 for fixed @xmath57 should be similar to that of a distribution amplitude @xmath189 . by analogy with ,
e.g. , the pion distribution amplitude @xmath190 , which is symmetric with respect to the change @xmath191 , one may expect that the distribution of the @xmath13-momentum between the two partons described by the same field should also have some symmetry .
however , the symmetry can not be as simple as @xmath191 since the initial @xmath12 and the final @xmath192 momenta are not treated symmetrically in our description : the variable @xmath57 specifies the fraction of the _ initial _ momentum @xmath12 both for the outgoing ( @xmath193 ) and incoming ( @xmath194 ) partons . to treat @xmath12 and @xmath195 symmetrically
, we should interpret @xmath57 for the returning parton as the fraction of the final hadron momentum @xmath196 , i.e. , rewrite its momentum @xmath194 as @xmath197 . hence , the symmetry of a double distribution @xmath187 may be only with respect to the interchange @xmath198 @xcite .
another way to make the symmetry between the initial and final hadrons more explicit is to use @xmath199 and @xmath13 as the basic momenta rather than @xmath12 and @xmath13 ( cf .
@xcite ) writing the momenta of the partons as @xmath200 and @xmath201 .
then the @xmath198 symmetry corresponds to @xmath202 symmetry .
the variable @xmath203 changes in the interval @xmath204 . writing @xmath205 as @xmath206
, we introduce a new variable @xmath110 satisfying a symmetric constraint @xmath207 , where @xmath208 .
the @xmath198 symmetry now converts into @xmath209 symmetry . finally , rescaling @xmath110 as @xmath210 produces the variable @xmath211 with @xmath57-independent limits : @xmath212 . written in terms of @xmath57 and @xmath211
, a modified double distribution @xmath213 obeys the reduction formula @xmath214 it is instructive to study some simple models allowing to satisfy this relation .
namely , let us assume that the profile in @xmath211-direction is a universal function @xmath215 for all @xmath57 , i.e. , take the factorized ansatz @xmath216 with @xmath215 normalized by @xmath217 possible simple choices for @xmath215 may be @xmath218 ( no spread in @xmath211-direction ) , @xmath219 ( characteristic shape for asymptotic limit of quark distribution amplitudes ) , @xmath220 ( asymptotic shape of gluon distribution amplitudes ) , etc . in our original variables
@xmath221 , the factorized ansatz can be written as @xmath222 where @xmath223 is a function symmetric with respect to the interchange @xmath198 .
a trivial observation is that the variable @xmath57 itself is given by a combination @xmath224 $ ] symmetric with respect to the @xmath198 transformation .
the normalization function @xmath225 is specified by @xmath226 for the three simple choices mentioned above , the model ( [ 65 ] ) gives @xmath227 in a similar way , one can construct anstze for functions @xmath1 involving nonzero @xmath36 values .
the nonforward matrix elements accumulate process - independent information and , hence , have a quite general nature .
the coefficient of proportionality between @xmath75 and @xmath76 characterizes the skewedness of matrix elements .
the characteristic feature implied by representations for double distributions ( see , e.g. , eqs.([31 ] ) , ( [ 32 ] ) ) is the absence of the @xmath5-dependence in the dds @xmath187 and @xmath228 .
an alternative way to parametrize nonforward matrix elements of light - cone operators is to use the ratio @xmath229 and the total momentum fraction @xmath230 as independent variables .
taking into account that for a lightcone dominated process only one direction for @xmath9 gives the leading contribution , one can do the change @xmath231 directly in our definitions of double distributions . as a result
, the variable @xmath59 would appear there only in the @xmath232 combination , where @xmath233 can be treated as the _ total _ fraction of the initial hadron momentum @xmath12 carried by the active quark .
if we require that the light - cone plus component of the final hadron momentum ( i.e. , @xmath234 ) is positive , then @xmath235 . using the spectral property @xmath63 of double distributions we obtain that the variable @xmath233 satisfies a similar `` parton '' constraint @xmath236 . integrating each particular double distribution @xmath237 over @xmath59 gives the nonforward parton distributions @xmath238 where @xmath239 .
the two components of nfpds correspond to positive ( @xmath240 ) and negative ( @xmath241 ) values of the fraction @xmath242 associated with the returning parton . as explained in refs .
@xcite , the second component can be interpreted as the probability amplitude for the initial hadron with momentum @xmath12 to split into the final hadron with momentum @xmath243 and the two - parton state with total momentum @xmath244 shared by the partons in fractions @xmath245 and @xmath246 , where @xmath247 ( see fig.[fig : nonfwd ] ) . for the gluon dds
, the @xmath198 symmetry holds only if , instead of @xmath123 , one uses the symmetric overall factor @xmath126 in the definitions ( [ 36]),([37 ] ) .
the use of such a definition of the gluon dds is implied in this section .
furthermore , the nonforward gluon distribution @xmath248 is obtained by integrating @xmath249 . to simplify notations
, it will be also implied below that , for the gluons , @xmath250 in eq.([71 ] ) corresponds to @xmath251 .
the basic distinction between double distributions @xmath187 and nonforward distributions @xmath252 is that nfpds explicitly depend on the skewedness parameter @xmath5 .
they form families of functions @xmath253 whose shape changes when @xmath5 is changed . the fact that the functions @xmath252 corresponding to different @xmath5 s are obtained by integrating the same double distribution @xmath187 imposes essential restrictions on possible shapes of @xmath252 and on how they change with changing @xmath5 .
the relation between nfpds and dds has a simple graphical illustration on the `` dd - life '' triangle defined by @xmath188 ( see fig.[fig:2 ] ) . to get @xmath252
, one should integrate @xmath187 over @xmath59 along a straight line specified by @xmath254 .
fixing some value of @xmath5 , one deals with a set of parallel lines corresponding to different values of @xmath233 . evidently , each such line intersects the @xmath57-axis at @xmath255 .
the upper limit of the @xmath59-integration is determined by intersection of this line either with the line @xmath256 ( this happens if @xmath6 ) or with the @xmath59-axis ( if @xmath257 ) .
the line corresponding to @xmath258 separates the triangle into two parts generating two components of the nonforward parton distribution . in the forward case ,
when @xmath259 , there is only one component , and the usual parton densities @xmath4 are produced by integrating @xmath187 along the vertical lines @xmath102const ( see fig.[fig:1 ] ) . in case when @xmath240 , looking at the integration line for the nonforward parton distribution @xmath260 one can see ( fig.[fig:2]b ) that it is inside the space between the integration lines giving the usual parton densities @xmath261 and @xmath262 corresponding to the momentum fractions @xmath233 , @xmath263 of the initial and final parton . assuming a monotonic decrease of the double distribution @xmath187 in the @xmath57-direction and a universal profile in the @xmath59-direction , one may expect that @xmath260 is larger than @xmath261 but smaller than @xmath262 .
inequalities between forward and nonforward distributions were recently discussed in refs .
they are based on the application of the cauchy - schwartz inequality @xmath264 to the nonforward distributions @xmath252 written generically as @xmath265 where @xmath266 describes the probability amplitude that the hadron with momentum @xmath12 converts into a parton with momentum @xmath267 and spectators @xmath268 .
the forward matrix elements are identified with the usual parton densities @xmath269 notice that the hadron momentum in the second forward matrix element is @xmath270 , hence the argument of the relevant parton density is @xmath271 , and one has @xmath272 as a result , we obtain ( compare with @xcite ) @xmath273 \ , .
\label{basineq}\ ] ] in other words , the functions involved in the bound for @xmath260 are @xmath261 and @xmath274 where the fraction @xmath275 is larger than @xmath276 @xcite .
one can see that @xmath277 is given exactly by the @xmath57-value of the intersection point in which the integration line @xmath278 giving the nonforward distribution @xmath260 crosses the boundary line @xmath256 ( see fig.[fig:2]c ) . for the gluon nonforward distributions , one should take into account extra factors @xmath279 , @xmath280 present in the definitions of forward distributions and the overall factor chosen in the rhs of definitions of the nonforward gluon distributions ( see eqs .
( [ 36 ] ) , ( [ 37 ] ) ) .
if one uses the @xmath281 symmetric combination @xmath126 , then @xmath282 \ , . \label{modineq}\ ] ] it is clear that the whole construction makes sense only if @xmath283 ( or @xmath6 ) . if @xmath284 , the nonforward distribution corresponds to matrix elements @xmath285 which have no obvious relation to the usual parton densities .
furthermore , in our graph of fig.[fig:2]a , the left end of the line @xmath278 in this case corresponds to @xmath286 , where the usual parton densities are infinite , and the inequalities become trivial .
in fact , they are trivial even for the border point @xmath258 .
another deficiency of the cauchy - schwartz - type inequalities is that they do not give the lower bound for nonforward distributions though our graphical interpretation suggests that @xmath260 for @xmath287 is larger than @xmath261 if the @xmath57-dependence of the double distribution @xmath187 along the lines @xmath288 is monotonic . to develop intuition about possible shapes of nonforward distributions ,
it is instructive to derive the nfpds corresponding to three simple models specified in the previous section .
in particular , for the ansatz we get @xmath289 i.e. , nfpds for non - zero @xmath5 are obtained from the forward distribution @xmath290 by a shift and rescaling .
note that the model ( [ model ] ) satisfies the inequalities ( [ basineq ] ) , ( [ modineq ] ) in the region @xmath6 for any function @xmath291 of the type provided that @xmath292 and @xmath293 . using the relations @xmath294 between our nonforward distributions and ji s off - forward parton distributions ( ofpds ) @xmath295 @xcite , one can see that the delta - function ansatz gives the simplest @xmath296-independent model @xmath297 for ofpds at @xmath298 .
is outside the physical region ; hence , the @xmath299 limit should be understood in the sense of analytic continuation . ]
it is worth noting that the mit bag model calculation @xcite did produce a set of ofpds which are almost independent of @xmath296 .
an evident interpretation is that the model constructed in ref.@xcite strongly suppresses the redistribution of the momentum transfer among the constituents which results in a very narrow spread of @xmath187 in the @xmath59-direction .
even if such a picture is physically correct for a low normalization point @xmath300 mev , evolution to higher values @xmath301 gev widens the @xmath59-profile of @xmath187 and evolved ofpds would change their shape with @xmath296 , as was explicitly demonstrated through a numerical calculation by belitsky _
_ @xcite .
the evolution of _ nonforward _ distributions @xmath302 was recently studied in refs .
@xcite . as a starting condition
, the authors assume that , at some low scale @xmath303 , the nonforward distributions @xmath304 for all @xmath5 have the same universal shape coinciding with that of the usual ( forward ) densities @xmath305 .
this assumption corresponds to the ansatz @xmath306 with double distribution being nonzero on the @xmath57-axis only .
this ansatz is not realistic , since it has no symmetry with respect to the @xmath307 interchange .
however , evolution equations are applicable to any distribution and , just due to its asymmetric profile , this unrealistic double distribution has a very distinctive evolution pattern reflecting the restoration of the @xmath198 symmetry .
namely , the asymptotic functions @xmath308 are @xmath307 symmetric .
in particular , both in pure gluodynamics and in qcd , we have @xmath309 ( see ref.@xcite and the appendix ) .
hence , one may expect that the evolition of @xmath310 shifts its crest towards the @xmath311 line and also makes the @xmath59-shape of the double distribution wider . to see whether the results of refs.@xcite reflect this expectation
, we introduce a general model with a narrow @xmath59-dependence : @xmath312 ( in what follows , it will be referred to as the `` @xmath313-delta ansatz '' ) .
this double distribution is concentrated on the @xmath314 line and gives @xmath315 for nonforward distributions . in case of two other models
, simple analytic results can be obtained only if we specify a model for @xmath4 .
for the `` valence quark''-oriented ansatz @xmath316 , the following choice of a normalized distribution @xmath317 is @xmath318 close to phenomenological valence quark distributions and @xmath319 produces a simple expression for the double distribution since the denominator @xmath320 factor in eq.([74 ] ) is canceled . as a result ,
the integral in eq.([71 ] ) is easily performed and we get @xmath321 resulting curves for @xmath322 with @xmath323 and @xmath324 are shown in fig.[fig:3 ] . a characteristic feature of each curve is a maximum located close to the relevant border point @xmath325 and slightly shifted to the left from it .
note that both the functions @xmath322 and their derivatives @xmath326 are continuous at @xmath258 .
the latter property is secured by the fact that @xmath316 vanishes at the upper corner @xmath327 .
the @xmath328 denominator factor for the `` gluon - oriented '' ansatz @xmath329 is canceled if one takes the model @xmath330 which , fortunately , is also consistent with the @xmath331 behavior of the phenomenological gluon distributions .
it is well known @xcite that the values of nonforward distributions @xmath332 taken at the border point @xmath258 determine imaginary parts of dvcs and hard electroproduction amplitudes .
an interesting question is the relation between the usual distributions @xmath333 and the values @xmath334 of nonforward distributions at the border point .
it is easy to calculate that for the @xmath335 delta ansatz @xmath336 this ratio is given by @xmath337 it is larger than 1 for any monotonically descreasing function @xmath4 , i.e. , the nonforward distribution @xmath334 in this case is larger than @xmath333 . in the small-@xmath5 limit , @xmath338 is completely determined by the small-@xmath57 behavior of @xmath4 , and the expression for @xmath339 simplifies to @xmath340 hence , if @xmath4 has a purely powerlike behavior @xmath341 for small @xmath57 , then @xmath342 , i.e. , for small @xmath5 , the ratio of the nonforward distribution @xmath334 and the usual parton density @xmath333 is practically constant , deviating from the @xmath259 limiting value by @xmath343 terms only .
the limiting value in this case is 1.41 for @xmath323 and 1.23 ( 1.15 ) for @xmath344 ( @xmath345 ) . however , if @xmath4 is a sum of two different powerlike terms @xmath346 or if it contains logarithms , e.g. , @xmath347 for small @xmath57 , then the @xmath5-dependence is more pronounced . in the latter case @xmath348 and
there is a visible deviation from the limiting @xmath349 value for all accessible @xmath5 : on the @xmath350 scale , the @xmath5-dependence of the ratio @xmath351 can not be neglected even for @xmath352 . for a general @xmath313-delta model @xmath312 , the ratio @xmath353 for small @xmath5
can be approximated by @xmath354 which again gives a @xmath5-independent constant @xmath355 for a purely powerlike function @xmath341 while the @xmath356-factor would modify the constant by @xmath357 $ ] .
if one uses the `` valence quark''-oriented ansatz @xmath316 with a simple powerlike behavior @xmath341 for small @xmath57 , the ratio is given by @xmath358 just like in the previous example , the nonforward distribution @xmath334 is larger than @xmath333 for all positive @xmath359 . for small @xmath5
, the ratio tends to @xmath360 , e.g. , to @xmath361 for @xmath323 which is the usual choice for valence quark distributions ( for comparison , taking @xmath344 ( @xmath362 ) gives @xmath363 ( 1.44 ) for @xmath364 ) .
for small @xmath359 , this result can be translated into @xmath365 , which coincides with the ratio @xmath366 for the modified narrow ansatz @xmath367 with @xmath368 .
hence , for @xmath369 the widening of the @xmath59-distribution can be approximated by a narrow distribution shifted from @xmath370 upwards to the @xmath371 line .
again , a logarithm @xmath356 in @xmath4 at small @xmath57 would induce a visible @xmath5-dependence for the @xmath372 ratio even for very small @xmath5 .
switching to the `` gluon - oriented '' ansatz @xmath373 with a purely power behavior @xmath341 for small @xmath57 , we obtain a similar expression @xmath374 which is close to @xmath375 for small @xmath359 . to approximate this result by the delta ansatz @xmath376
, one should take @xmath377 .
the effective shift upward is smaller in this case because @xmath373 is more narrow in the @xmath59-direction than @xmath316 .
choosing @xmath359 , we should take into account that the nonforward gluon distribution @xmath378 reduces to @xmath379 in the @xmath349 limit @xcite .
hence , @xmath333 in the above formulas should be understood as @xmath380 .
now , if we make an old - fashioned assumption that @xmath379 tends to a constant as @xmath381 , then @xmath382 and @xmath383 tends to 1 at small @xmath5 , i.e. , the nonforward distribution @xmath384 coincides in the small-@xmath5 limit with its forward counterpart @xmath380 . to get
a more realistic gluon distribution @xmath379 growing at small @xmath233 one should use a positive parameter @xmath359 .
taking @xmath344 , we get @xmath385 , and @xmath386 ( 1.39 ) for @xmath345 ( @xmath362 ) ) . these estimates for the ratio @xmath387 are close to those obtained in refs .
@xcite where the nonforward distributions @xmath388 at high normalization point @xmath389 were constructed by applying evolution equations to an initial low normalization point @xmath303 ansatz @xmath390 which was assumed to have a universal @xmath5-independent shape coinciding with the usual distribution @xmath391 . in particular , martin and ryskin considered the evolution of the gluon nfpd in pure gluodynamics .
they took @xmath392 gev@xmath393 ( two other choices @xmath394 gev@xmath393 and @xmath395 gev@xmath393 were also considered ) and then evolved @xmath388 to higher @xmath20 values @xmath396 , and 100 gev@xmath393 .
they found that @xmath397 for @xmath398 gev@xmath393 , which corresponds to @xmath399 in our @xmath400 model .
this value is close to those used in phenomenological parametrizations of the gluon distributions .
it should be also noted that the results for @xmath401 obtained in ref.@xcite have a nonnegligible @xmath5-dependence .
this feature can be expected since the grv gluon distribution @xcite which they use can be rather well approximated at @xmath402 gev@xmath393 by a simple formula @xmath403 which works with @xmath404 accuracy for @xmath57 ranging from @xmath405 to @xmath406 . in the pure gluodynamics approximation used in ref.@xcite , its shape does not drastically change when evolved either to @xmath407 gev@xmath393 or to @xmath408 and 100 gev@xmath393 . as discussed above ,
the assumption that the nonforward distributions @xmath390 have a universal @xmath5-independent shape corresponds to the ansatz @xmath409 , i.e. , to the @xmath313-delta ansatz with the vanishing slope @xmath410 .
modeling the evolved double distributions by a @xmath313-delta ansatz with nonzero @xmath313 , we expect that , due to the restoration of the @xmath411 symmetry , the effective slope parameter @xmath313 should increase with @xmath20 .
namely , for the @xmath313-delta ansatz , the ratio of the nonforward distribution @xmath412 and the forward parton distribution @xmath413 is given by @xmath414 taking @xmath415 and the @xmath20-dependent slope @xmath416 for @xmath417 and 100 gev@xmath393 , respectively , we were able to reproduce the results of ref.@xcite for a wide range of @xmath5 parameters : @xmath418 and @xmath406 . the relevant curves , coinciding with those of ref .
@xcite within a few per cent accuracy , are shown in fig.[fig : mr ] .
hence , the increase of the ratio @xmath419 with @xmath20 observed in refs.@xcite basically reflects the shift of the gluon double distribution from the @xmath57 axis @xmath420 towards the symmetry line @xmath421 .
this effect , being an artifact of the initial conditions , plays the dominant role up to @xmath422 gev@xmath393 .
as argued above , the @xmath5-dependence of the ratio may be traced to the fact that the gluon distribution @xmath423 differs from a simple power @xmath424 .
since the assumption @xmath425 is equivalent to the ansatz @xmath426 which is not symmetric with respect to the @xmath427 interchange , one should avoid using it as a starting condition for evolution .
as explained earlier , a more realistic set of nonforward distributions @xmath428 is generated by the @xmath429 ansatz for the double distribution corresponding to skewedness - independent set of ji s off - forward distributions .
comparing these two sets , one may be tempted to argue that for extremely small @xmath5 considered in ref.@xcite , @xmath430 terms in eq.([712 ] ) are inessential .
of course , @xmath430 can be neglected when subtracted from 1 . however , for the @xmath233-values close to the border point @xmath431 , the shift by @xmath430 produces visible changes for functions having the @xmath432 behavior with @xmath433 . in the case of the ansatz ( [ 712 ] ) , the ratio @xmath434 differs from 1 for all @xmath389 . for small @xmath5 ,
the difference is significant only for @xmath233 close to @xmath5 .
when a narrow double distribution has its crest on the @xmath435 line from the very start , there are no effects due to the shift of the crest , and the @xmath389-evolution of @xmath436 in the region @xmath6 reflects only the widening of the double distribution in the @xmath59-direction and the change of its profile in the @xmath57-direction . as we have seen , the widening of the double distribution changes the effective slope @xmath313 by a small amount only .
hence , for small @xmath5 one can use the approximate formula @xmath437 for evolved distributions as well .
in other words , the ratio @xmath438 for @xmath240 and small @xmath5 can be estimated from existing results for the usual gluon density @xmath439 . comparing the formula ( [ eq:617 ] ) with the relation ( [ onfpd ] ) between our nonforward and ji s off - forward distributions
, one can conclude that eq.([eq:617 ] ) is equivalent to a statement that at small @xmath296 and @xmath440 one can neglect the @xmath296-dependence of the off - forward distributions @xmath441 .
again , such a statement is only nontrivial if @xmath442 . to analyze the accuracy of eq.([eq:617 ] ) , we will construct an expansion of @xmath441 in powers of @xmath296 . to this end , it is convenient to use the parton picture based on modified double distribution @xmath443 in which the plus component of the parton momenta is measured in units of that of the average hadron momentum @xmath127 .
the parton momenta then are @xmath444 and @xmath445 with @xmath446 changing between @xmath447 and @xmath448 . defining @xmath449 and @xmath450 ,
one obtains the description in terms of the off - forward parton distributions @xmath441 @xcite .
the parton momenta are now @xmath451 and @xmath452 . in the region
@xmath440 , the ofpds are obtained from @xmath453 by the integral @xmath454 using the @xmath455 symmetry of @xmath456 , it is easy to see from this expression that the off - forward parton distributions @xmath441 are even functions of @xmath296 : @xmath457 this result was originally obtained by x. ji @xcite with the help of a different technique . expanding the rhs of eq.([offf ] ) in powers of @xmath296 , we get @xmath458 + \ldots \ , . \label{exxi2}\ ] ] where @xmath459 is the forward distribution . hence , for small @xmath296 ,
the corrections are formally @xmath460 , i.e. , they look very small .
however , if @xmath461 has a singular behavior like @xmath424 , then @xmath462 and the relative suppression of the first correction is @xmath463 i.e. , the corrections are tiny for all @xmath464 except for the region @xmath442 where the correction has no parametric smallness . nevertheless , even in this region it is suppressed numerically , because the @xmath465 moment is rather small for a distribution concentrated in the small-@xmath446 region .
this discussion shows that the formula ( [ eq:617 ] ) is not just an automatic consequence of the @xmath460 nature of the first nonvanishing correction .
it is easy to write expicitly all the terms which are not suppressed in the @xmath466 limit @xmath467 the numerical suppression of higher terms is even stronger , and the series converges rather fast . in terms of the off - forward distributions , the inequality ( [ basineq ] ) reads @xmath468\ , .
\label{hq}\ ] ] for the gluons , one should use the inequality ( [ modineq ] ) , which leads to @xmath469\ , .
\label{hg}\ ] ] again , if one takes the model @xmath470 , the inequalities ( [ hq ] ) and ( [ hg ] ) are valid for any function @xmath4 of @xmath471 type with @xmath472 .
so far we assumed in our models that dds are finite everywhere on the `` life triangle '' .
consider , however , a situation when the partons emerge from a meson - like state ( or glueball / pomeron in the gluon case ) exchanged in the @xmath36 channel . in this case
, the partons just share the plus component of the momentum transfer @xmath13 : information about the magnitude of the initial hadron momentum is lost if the exchanged particle can be described by a pole propagator @xmath473 .
hence , the meson - exchange contribution to a double distribution is proportional to @xmath474 or its derivatives , e.g. : @xmath475 where @xmath476 is the distribution amplitude of the meson @xmath477 .
this contribution to the nonforward distribution is nonzero only in the @xmath478 region : @xmath479 at the beginning , we described the nonforward matrix element of a quark operator by two functions @xmath480 and @xmath481 corresponding to positive-@xmath57 and negative-@xmath57 parts of the general fourier representation . since @xmath286 for a meson - exchange contribution
, it makes sense to treat it as a third independent component , i.e. , to parametrize the nonforward matrix element by the sum @xmath482 .
all three components contribute to the nonforward distributions in the @xmath483 region .
however , the @xmath484 terms do not contribute to the nonforward distributions in the @xmath485 region and to the usual parton densities @xmath4 .
for this reason , the @xmath484 terms , if they exist , would lead to violation of sum rules ( like energy - momentum sum rule ) for the usual parton densities . note that if the meson da @xmath486 does not vanish at the end - points , the nonforward distribution does not vanish at @xmath487 ( the off - forward parton distributions @xmath488 in this case are discontinuous at @xmath489 ) . as explained in ref .
@xcite , pqcd factorization for dvcs and other hard electroproduction processes fails in such a situation , because of the @xmath490 factors ( @xmath491 factors if ofpd formalism is used ) contained in hard amplitudes .
it should be mentioned that a nearly discontinuous behavior of ofpds for @xmath492 was obtained in the chiral soliton model @xcite .
formally , the evolution to sufficiently high @xmath140 results in the functions vanishing at the end - point @xmath487 .
a non - trivial question , however , is whether evolution starts at all in a situation when pqcd factorization fails .
in this paper , we duscussed the formalism of double distributions .
we treated them as the starting objects in parametrization of nonforward matrix elements . an alternative description in terms of nonforward or off - forward parton distributions
was obtained by an appropriate integration of the relevant dds . incorporating spectral and symmetry properties of double distributions , we proposed simple models producing self - consistent sets of non - forward distributions @xmath260 and discussed their @xmath5-dependence and relation to usual ( forward ) parton densities . using a qualitative picture of the evolution of double distributions
, we were able to explain and model the basic features of the evolution pattern of nonforward distributions observed in numerical evolution studies @xcite . in the appendix , we present the set of evolution equations for double distributions in the singlet case and discuss their analytic solution . work on numerical evolution of the nonforward distributions corresponding to realistic anstze ( [ 712 ] ) is in progress @xcite .
another interesting problem for a future investigation is a numerical evolution of double distributions .
i acknowledge stimulating discussions and communication with i. balitsky , a. belitsky , j. blumlein , v. braun , j. collins , c. coriano , l. frankfurt , b. geyer , k. golec - biernat , x. ji , l. mankiewicz , i. musatov , g. piller , m. polyakov , d. robaschik , m. ryskin , m. strikman and o.v .
this work was supported by the us department of energy under contract de - ac05 - 84er40150 .
as described in sec .
iv , the evolution kernels for double distributions can be conveniently obtained from the light - ray evolution kernels @xmath493 . for the parton helicity averaged case ,
the latter were originally obtained in refs .
@xcite . here
we present them in the form given in ref.@xcite : @xmath494_+ + \delta(v ) [ \bar u/ u]_+ - \frac1{2 } \delta ( u)\delta(v ) \right ) \
, , \\ & & b^{gq}(u , v ) = \frac{\alpha_s}{\pi } c_f \biggl ( 2 + \delta ( u)\delta(v ) \biggr ) \ , , \\ & & b^{qg}(u , v ) = \frac{\alpha_s}{\pi } n_f \left ( 1 + 4uv -u -v \right ) \
, , \\ & & b^{gg}(u , v ) = \frac{\alpha_s}{\pi } n_c \biggl ( 4(1 + 3uv -u -v ) + \frac{\beta_0}{2 n_c } \ , \delta(u)\delta(v ) + \left \ { \delta(u ) \biggl [ \frac{\bar v^2}{v } - \delta ( v ) \int_0 ^ 1 \frac{d \tilde v}{\tilde v } \biggr ] + \ { u \leftrightarrow v \ } \right \ } \biggr ) \ , .
\label{414 } \end{aligned}\ ] ] as usual , @xmath495 is the lowest coefficient of the qcd @xmath211-function .
evolution kernels for the parton helicity - sensitive case are given by @xcite @xmath496 at one loop , @xmath497 , and this kernel was already displayed in eq .
( [ 46 ] ) .
other kernels , including the @xmath498 kernel originally obtained in ref.@xcite , are given by @xmath499 -2 \delta(1-x/\xi ) \delta(y-\eta ) \int_0 ^ 1 \frac{d \tilde v } { 1-\tilde v } \ , \biggr \}\ , , \nonumber \\ & & r^{gg}(x , y ; \xi , \eta ; g ) = \delta r^{gg}(x , y;\xi , \eta;g ) + 12 \ , \frac{\alpha_s}{\pi } \ , n_c \ , \frac{1}{x } \ , ( y-\eta x/\xi)(\bar y - \bar \eta
x/\xi ) \ , \theta ( 0 \leq x/\xi \leq { \rm min } \ { y/\eta , \bar y / \bar \eta \ } ) \ , , \\ & & \delta r^{gq}(x , y ; \xi , \eta ; g ) = \frac{\alpha_s}{\pi } c_f \ , \frac1{x } \ , \biggl \
{ -2 \biggl [ \left ( \frac{y}{\eta } - \frac{x } { \xi } \right ) \ , \theta ( x / \xi \leq y/\eta \leq 1 ) + \left \ { { y \to \bar y } \atop { \eta \to \bar \eta } \right \ } \biggr ] + \delta(\eta -y ) \ , \theta ( 0 \leq x \leq \xi ) \biggr \ } \ , , \\ & & r^{gq}(x , y ; \xi , \eta ; g ) = \frac{\alpha_s}{\pi } c_f \ , \frac1{x } \ , \biggl \ { 2 \biggl [ \left ( \frac{y}{\eta } - \frac{x } { \xi } \right ) \ , \theta ( x / \xi \leq y/\eta \leq 1 ) + \left \ { { y \to \bar y } \atop { \eta \to \bar \eta } \right \ }
\biggr ] + \delta(\eta -y ) \ , \theta ( 0 \leq x \leq \xi ) \biggr \ } \ , , \\ & & \delta r^{qg}(x , y ;
\xi , \eta ; g ) = \frac{\alpha_s}{\pi } n_f \frac1{\xi } \biggl \ { \frac{x}{\xi } \biggl ( \delta(x/\xi - y/\eta ) \ , \theta(y \leq \eta ) + \delta(x/\xi - \bar y/ \bar \eta ) \ , \theta(y \geq \eta )
\biggr ) \nonumber \\ & & \hspace{9 cm } - \ , \theta ( 0 \leq x/\xi \leq { \rm min } \ { y/\eta , \bar y/ \bar \eta \ } ) \biggr \ } \ , , \\ & & r^{qg}(x , y ; \xi , \eta ; g ) = \delta r^{qg}(x , y ; \xi
, \eta ; g ) + 4 \ , \frac{\alpha_s}{\pi } n_f \ , \frac1{\xi } \ , \eta \bar \eta \biggl ( \frac { y}{\eta } + \frac { \bar y } { \bar \eta } - 2\ , \frac{x}{\xi } \biggr ) \ , \theta ( 0 \leq x/\xi \leq { \rm min } \ { y/\eta , \bar y/ \bar \eta \ } ) \ , .
\label{428 } \end{aligned}\ ] ] to find a formal solution of the evolution equations for double distributions , we proposed in refs .
@xcite to combine the standard methods used to solve the evolution equations for parton densities and distribution amplitudes .
hence , let us start with taking the moments with respect to @xmath57 . utilizing the property @xmath500 we get @xmath501 where @xmath502 is the @xmath503th @xmath57-moment of @xmath504 @xmath505 the kernels @xmath506 and
analogous kernels @xmath507 governing the evolution of @xmath508 are given by @xmath509 \theta(y \leq \eta ) \right .
\nonumber \\ & & + \left ( \frac{\bar y}{\bar \eta } \right ) ^{n+1 } \left [ \frac1{n+1 } + \frac1{y- \eta } \right ] \theta(y \geq
\eta ) \left .
- \frac1{2 } \ ,
\delta(y-\eta ) - 2 \delta(y-\eta ) \int_0 ^ 1 \frac{z}{1-z } \ , dz \right \}. \label{eq : rnkernelqq } \end{aligned}\ ] ] @xmath510 \right \ } , \label{eq : rnkerneldgg } \end{aligned}\ ] ] @xmath511 @xmath512 @xmath513 @xmath514 \right \ } , \label{58 } \end{aligned}\ ] ] @xmath515 \right \ } \ , .
\label{59 } \end{aligned}\ ] ] from eqs .
( [ 56 ] ) , ( [ 58 ] ) one can derive the following reduction formulas for the nondiagonal kernels : @xmath516 @xmath517 the same relations connect the nondiagonal kernels @xmath518 with the exclusive kernels @xmath519 given in ref .
@xcite . to understand their structure
, one should realize that constructing the nondiagonal @xmath128 and @xmath129 kernels , one faces mismatching @xmath279 factors which in the pure @xmath520 case are converted into derivatives with respect to @xmath59 .
it is straightforward to check that all the kernels @xmath521 ( and @xmath522 ) have the property @xmath523 where @xmath524 .
hence , the eigenfunctions of the evolution equations are orthogonal with the weight @xmath524 , @xmath525 they are proportional to the gegenbauer polynomials @xmath526 , see @xcite and refs.@xcite where the general algorithm was applied to the evolution of flavor - singlet distribution amplitudes .
expanding the moment functions @xmath527 over the gegenbauer polynomials @xmath526 @xmath528 we get the evolution equation for the expansion coefficients @xmath529 where @xmath530 are the eigenvalues of the kernels @xmath531 related to the elements @xmath532 of the usual flavor - singlet anomalous dimension matrix @xmath533 and similarly for the helicity - sensitive quantities @xmath534 .
namely , @xmath535 \ , , \\ & & \gamma_n^{gg}= - 2 n_c \left [ - \frac1{n(n+1 ) } - \frac1{(n+2)(n+3 ) } + \sum_{j=1}^{n+1 } \frac1{j}\right ] + \frac{\beta_0}{2 } \ , , \\ & & \delta \gamma_n^{gg}= - 2 n_c \left [ - \frac2{(n+2)(n+3 ) } + \sum_{j=1}^{n+1 } \frac1{j}\right ] + \frac{\beta_0}{2 } \ , , \\ & & \gamma_n^{gq}= c_f \frac{n^2 + 3n+4}{(n+1)(n+2 ) } \hspace{0.5 cm } , \hspace{0.5 cm } \gamma_n^{qg}= n_f \frac{n^2 + 3n+4}{n(n+1)(n+2)(n+3 ) } \ , , \\ & & \delta \gamma_n^{gq}= c_f \frac{n(n+3)}{(n+1)(n+2 ) } \hspace{0.5 cm } , \hspace{0.5 cm } \delta \gamma_n^{qg}= n_f \frac{1}{(n+2)(n+3 ) } .
\label{eq : ads } \end{aligned}\ ] ] let us consider first two simplified situations . in the quark nonsinglet case , the evolution
is governed ( in helicity - averaged case ) by @xmath536 alone : @xmath537^ { 2\gamma_{n+k}^{qq}/\beta_0 } , \label{eq : fnnons}\ ] ] since @xmath538 while all the anomalous dimensions @xmath539 with @xmath540 are negative , only @xmath541 survives in the asymptotic limit @xmath542 while all the moments @xmath543 with @xmath544 evolve to zero values .
hence , in the formal @xmath542 limit , we have @xmath545 @xmath525 in each of its variables , the limiting function @xmath546 acquires the characteristic asymptotic form dictated by the nature of the variable : @xmath474 is specific for the distribution functions @xcite , while the @xmath547-form is the asymptotic shape for the lowest - twist two - body distribution amplitudes @xcite . for the nonforward distribution of a valence quark @xmath548
this gives @xmath549 where @xmath550 is the number of the valence @xmath548-quarks in the hadron .
another example is the evolution of the gluon distribution in pure gluodynamics which is governed by @xmath551 with @xmath552 .
note that the lowest local operator in this case corresponds to @xmath553 .
furthermore , in pure gluodynamics , @xmath554 vanishes while @xmath555 if @xmath556 .
this means that in the @xmath542 limit we have @xmath557 for the double distribution which results in @xmath558 for the nonforward distribution . in the formulas above , the total momentum carried by the gluons ( in pure gluodynamics ! ) was normalized to unity . in qcd , we should take into account the effects due to quark - gluon mixing .
diagonalizing eq.([a27 ] ) , we obtain two multiplicatively renormalizable combinations @xmath559 where ( omitting the @xmath560 indices ) @xmath561 their evolution is governed by the anomalous dimensions @xmath562 in particular , @xmath563 and @xmath564 which means that @xmath565 does not evolve : the total momentum carried by the partons is conserved .
another multiplicatively renormalizable combination involving @xmath566 and @xmath567 is @xmath568 it vanishes in the @xmath542 limit , and we have @xmath569 since all the combinations @xmath570 with @xmath571 vanish in the @xmath542 limit , we obtain @xmath572 or @xmath573 in terms of nonforward distributions this is equivalent to @xmath574 @xmath575 note that both @xmath576 and @xmath577 vanish in the @xmath542 limit .
x. ji , phys.rev.lett . * 78 * 610 ( 1997 ) .
radyushkin , phys .
b * 380 * ( 1996 ) 417 .
x. ji , phys.rev .
d * 55 * ( 1997 ) 7114 .
radyushkin , phys .
b * 385 * ( 1996 ) 333 .
radyushkin , phys .
d * 56 * ( 1997 ) 5524 .
brodsky , l. frankfurt , j.f .
gunion , a.h .
mueller and m. strikman , phys .
d * 50 * ( 1994 ) 3134 .
collins , l. frankfurt and m. strikman , phys .
d * 56 * ( 1997 ) 2982 .
d. mller , d. robaschik , b. geyer , f .- m .
dittes and j. hoeji , fortschr.phys .
* 42 * ( 1994 ) 101 .
l. mankiewicz , g. piller and t. weigl , eur.phys.j .
c * 5 * ( 1998 ) 119 .
j. blumlein , b. geyer and d. robaschik , desy-97 - 209 ( 1997 ) , hep - ph/9711405 .
l. frankfurt , a. freund , v. guzey and m. strikman , phys .
b * 418 * ( 1998 ) 345 .
j. blumlein , b. geyer and d. robaschik , phys .
b * 406 * ( 1997 ) 161 .
balitsky and a.v .
radyushkin , phys .
b * 413 * ( 1997 ) 114 . th .
braunschweig , b. geyer and d. robaschik , ann .
( leipzig ) * 44 * ( 1987 ) 403 .
balitsky and v.m .
braun , nucl.phys .
b * 311 * ( 1988/89 ) 541 . a.v .
belitsky , b. geyer , d. mller and a. schfer , phys .
b * 421 * ( 1998 ) 312 .
a.d . martin and m.g .
ryskin , phys .
d * 57 * ( 1998 ) 6692 .
k. golec - biernat , j. kwiecinski and a.d .
martin , hep - ph/9803464 .
radyushkin , phys .
b * 131 * ( 1983 ) 179 . v.n .
gribov and l.n .
lipatov , sov .
. phys . * 15 * ( 1972 ) 78 ; + l.n .
lipatov , sov . j. nucl
* 20 * ( 1975 ) 94 .
g. altarelli and g. parisi , nucl .
b * 126 * ( 1977 ) 298 .
l. dokshitser , sov.phys .
jetp , * 46 * ( 1977 ) 641 .
efremov and a.v .
radyushkin , theor .
* 42 * ( 1980 ) 97 .
s.j . brodsky and g.p .
lepage , phys .
b * 87 * ( 1979 ) 359 .
x.ji , j. phys .
g * 24 * ( 1998 ) 1181 .
b. pire , j. soffer and o.v .
teryaev , hep - ph/9804284 .
x. ji , w. melnitchouk and x. song , phys .
d * 56 * ( 1997 ) 5511 .
m. glck , e. reya and a. vogt , z. phys .
c * 67 * ( 1995 ) 433 .
petrov , p.v .
pobylitsa , m.v .
polyakov , i. bornig , k. goeke and c. weiss , phys .
d * 57 * ( 1998 ) 4325 .
musatov and a.v .
radyushkin , work in progress .
s.v . mikhailov and a.v .
radyushkin , nucl . phys .
* b273 * ( 1986 ) 297 .
chase , nucl .
b174 * ( 1980 ) 109 .
shifman and m.i .
vysotsky , nucl.phys .
b * 186 * ( 1981 ) 475 . th .
ohrndorf , nucl .
b * 186 * ( 1981 ) 153 .
baier and a.g .
grozin , nucl . phys .
b * 192 * ( 1981 ) 476 .
d.j . gross and f. wilczek , phys .
d * 9 * ( 1974 ) 980 .
h. georgi and h.d .
politzer , phys .
d * 9 * ( 1974 ) 416 . | applications of perturbative qcd to deeply virtual compton scattering and hard exclusive meson electroproduction processes require a generalization of usual parton distributions for the case when long - distance information is accumulated in nonforward matrix elements @xmath0 of quark and gluon light - cone operators . in our previous papers we used two types of nonperturbative functions parametrizing such matrix elements : double distributions @xmath1 and nonforward distribution functions @xmath2 . here
we discuss in more detail the double distributions ( dds ) and evolution equations which they satisfy .
we propose simple models for @xmath3 dds with correct spectral and symmetry properties which also satisfy the reduction relations connecting them to the usual parton densities @xmath4 . in this way , we obtain self - consistent models for the @xmath5-dependence of nonforward distributions . we show that , for small @xmath5 , one can easily obtain nonforward distributions ( in the @xmath6 region ) from the parton densities : @xmath7 .
pacs number(s ) : 12.38.bx , 13.60.fz , 13.60.le 100000 -48pt -0.5 cm * double distributions and evolution equations * a.v .
radyushkin + _ physics department , old dominion university , _ + _ norfolk , va 23529 , usa _ + _ and _ + _ jefferson lab , _ + _ newport news , va 23606 , usa _ |
of high - quality and real - time image processing algorithms is a key topic in today s context , where humans demand more capabilities and control over their rapidly increasing number of advanced imaging devices such as cameras , smart phones , tablets , etc .
many efforts are being made toward satisfying those needs by exploiting dedicated hardware , such as graphics processing units ( gpus ) @xcite .
some areas where gpus are emerging to improve human computer interaction ( hci ) include cybernetics ( for example , in facial @xcite or object @xcite recognition , classification using support vector machines @xcite and genetic algorithms for clustering @xcite ) , and image processing ( e.g. , unsupervised image segmentation @xcite , optical coherence tomography systems @xcite , efficient surface reconstruction from noisy data @xcite , remote sensing @xcite , real - time background subtraction @xcite , etc . ) . in such hardware - oriented designed algorithms , the computational efficiency of processing tasks is significantly improved by parallelizing the operations .
high level tasks such as ( a ) object / person - tracking or ( b ) video - based gestural interfaces often require a foreground segmentation method as a base building block.,title="fig : " ] + ( a ) high level tasks such as ( a ) object / person - tracking or ( b ) video - based gestural interfaces often require a foreground segmentation method as a base building block.,title="fig : " ] + ( b ) automated video analysis applications such as person tracking or video - based gestural interfaces ( see [ fig : applications ] ) rely on lower - level building blocks like foreground segmentation @xcite , where the design of efficient computational methods for dedicated hardware has a significant impact .
multimodal nonparametric segmentation strategies have drawn a lot of attention @xcite since they are able to provide high - quality results even in complex scenarios ( dynamic background , illumination changes , etc . )
however , their main drawback is their extremely high computational cost ( requiring the evaluation of billions of multidimensional gaussian kernels per second ) , which makes them difficult to integrate in the latest generation of image processing applications @xcite . to overcome this important drawback and achieve real - time performance @xcite , the use of parallel hardware such as gpus helps but may not be enough by itself , depending on the required resolution , hence the need for algorithms capable of evaluating non - linear ( e.g. , gaussian ) functions at high speed within a required error tolerance .
gpu vendors are aware of this recurrent computing problem and provide hardware implementations of common transcendental functions in special function units ( sfus ) @xcite ; indeed , we can find in the literature examples of successful use of these hardware facilities @xcite .
however , their ease of use comes at the price of non - customizable reduced numerical precision @xcite .
we propose a novel , fast , and practical method to evaluate any continuous mathematical function within a known interval .
our contributions include , based on error equalization , a nearly optimal design of two types of piecewise linear approximations ( linear interpolant and orthogonal projection ) in the @xmath0 norm under the constraint of a large budget of evaluation subintervals @xmath1 . moreover , we provide asymptotically tight bounds for the approximation errors of both piecewise linear representations , improving upon existing ones .
specifically , in addition to the @xmath2 convergence rate typical of these approximations , we quantify their interpolation constants : @xmath3 for the linear interpolant and a further @xmath4 improvement factor in case of the orthogonal projection .
the obtained error relations are parameterized by the number of segments used to represent the complex ( nonlinear ) function , hence our approach allows the user to estimate the errors given @xmath1 or , conversely , estimate the required @xmath1 to achieve a target approximation error .
we also propose an efficient implementation of the technique in a modern gpu by exploiting the fixed - function interpolation routines present in its texture units to accelerate the computation while leaving the rest of the gpu ( and , of course , the cpu ) free to perform other tasks ; this technique is even faster than using sfus @xcite .
although the initial motivation of the developed method was to improve the efficiency of nonparametric foreground segmentation strategies , it must be noted that it can be also used in many other scientific fields such as computer vision @xcite or audio signal processing @xcite , where the evaluation of continuous mathematical functions constitutes a significant computational burden .
section [ sec : piecewise - linear - approximations ] summarizes the basic facts about piecewise linear approximation of real - valued univariate functions and reviews related work in this topic ; section [ sec : finding - a - non - uniform ] derives a suboptimal partition of the domain of the approximating function in order to minimize the distance to the original function ( proofs of results are given in the appendixes ) ; section [ sec : computational - analysis - and ] analyzes the algorithmic complexity of the proposed approximation strategies and section [ sec : implementation - in - a - gpu ] gives details about their implementation on modern gpus .
section [ sec : results ] presents the experimental results of the proposed approximation on several functions ( gaussian , lorentzian and bessel s ) , both in terms of quality and computational times , and its use is demonstrated in an image processing application .
finally , section [ sec : conclusions ] concludes the paper .
in many applications , rigorous evaluation of complex mathematical functions is not practical because it takes too much computational power .
consequently , this evaluation is carried out approximating them by simpler functions such as ( piecewise-)polynomial ones .
piecewise linearization has been used as an attractive simplified representation of various complex nonlinear systems @xcite .
the resulting models fit into well established tools for linear systems and reduce the complexity of finding the inverse of nonlinear functions @xcite .
they can also be used to obtain approximate solutions in complex nonlinear systems , for example , in mixed - integer linear programming ( milp ) models @xcite .
some efforts have been devoted as well to the search for canonical representations in one and multiple dimensions @xcite with different goals such as black box system identification , approximation or model reduction .
previous attempts to address the problem considered here ( optimal function piecewise linearization ) include @xcite , the latter two in the context of nonlinear dynamical systems . in @xcite an iterative multi - stage procedure based on dynamical programming
is given to provide a solution to the problem on sequences of progressively finer 2-d grids . in @xcite the piecewise linear approximation
is obtained by using the evolutionary computation approach such as genetic algorithm and evolution strategies ; the resulting model is obtained by minimization of a sampled version of the mean squared error and it may not be continuous . in @xcite the problem
is addressed using a hybrid approach based on curve fitting , clustering and genetic algorithms . here
we address the problem from a less heuristic point of view , using differential calculus to derive a more principled approach @xcite .
our method solely relies on standard numerical integration techniques , which takes few seconds to compute , as opposed to recursive partitioning techniques such as @xcite , which take significantly longer . in this section
we summarize the basic theory behind piecewise linear functions @xcite and two such approximations to real - valued functions : interpolation and projection ( sections [ sub : linear - interpolation ] and [ sub : the - l2-projection ] , respectively ) , pictured in [ fig : pipvsproj ] along with their absolute approximation error with respect to @xmath5 .
top : fifth - degree polynomial @xmath6 and two continuous piecewise linear ( cpwl ) approximations , the orthogonal projection @xmath7 and the linear interpolant @xmath8 ; bottom : corresponding absolute approximation errors ( magnified by a @xmath9 factor).,width=317 ] hat functions constitute a basis of @xmath10 . since all the basis functions are zero outside @xmath11 $ ] , the basis functions associated with boundary nodes are only _ half hats_.,width=317 ] in essence , a piecewise function over an interval @xmath12 $ ] is a partition of @xmath13 into a set of @xmath1 subintervals @xmath14 , where @xmath15 , and a set of @xmath1 functions @xmath16 , one for each subinterval @xmath17 . in particular we are interested in continuous piecewise linear ( cpwl ) functions , which means that all the @xmath16 are linear and @xmath18
. cpwl functions of a given partition @xmath19 are elements of a vector space @xmath10 : the addition of such functions and/or multiplication by a scalar yields another cpwl function defined over the same subintervals . a useful basis for the vector space @xmath10 is formed by the set of _ hat functions _ or _ nodal basis functions _
@xmath20 , pictured in [ fig : hat - functions - constitute ] and defined in general by the formula @xmath21\\ ( x - x_{i+1})/(x_{i}-x_{i+1 } ) , & x\in\left[x_{i},x_{i+1}\right]\\ 0 , & x\notin\left[x_{i-1},x_{i+1}\right ] .
\end{cases}\ ] ] the basis functions @xmath22 and @xmath23 associated to the boundary nodes @xmath24 and @xmath25 are only _ half hats_. these basis functions are convenient since they can represent any function @xmath26 in @xmath10 by just requiring the values of @xmath26 at its nodal points , @xmath27 , in the form @xmath28 the piecewise linear interpolant @xmath29 of a continuous function @xmath5 over the interval @xmath13 can be defined in terms of the basis just introduced : @xmath30 while this cpwl approximation is trivial to construct , and may be suitable for some uses , it is by no means the best possible one .
crucially , @xmath31 for any function @xmath5 that is convex in @xmath13 . depending on the application in which this approximation is used , this property could skew the results .
however , as we will see in section [ sec : finding - a - non - uniform ] , the linear interpolant is useful to analyze other possible approximations .
it is also at the heart of the trapezoidal rule for numerical integration .
let the usual inner product between two square - integrable ( @xmath0 ) functions in the interval @xmath13 be given by @xmath32 then , the vector space @xmath10 can be endowed with the above inner product , yielding an inner product space . as usual , let @xmath33 be the norm induced by the inner product , and let @xmath34 be the distance between two functions @xmath35 .
the orthogonal projection of the function @xmath5 onto @xmath10 is the function @xmath36 such that @xmath37 since @xmath36 , it can be expressed in the nodal basis @xmath38 by @xmath39 where the coefficients @xmath40 solve the linear system of equations @xmath41 .
the @xmath42 matrix @xmath43 has entries @xmath44 , and vector @xmath45 has entries given by @xmath46 .
the gramian matrix @xmath47 is tridiagonal and strictly diagonally dominant .
therefore , the system has exactly one solution @xmath48 , which can be obtained efficiently using the thomas algorithm ( a particular case of gaussian elimination ) @xcite .
@xmath7 is the element in @xmath10 that is closest to @xmath5 in the sense given by the aforementioned @xmath0 distance @xmath49 , as we recall next . for any @xmath50 , @xmath51 where we used property that the vector @xmath52 . next , applying the cauchy - schwarz inequality , @xmath53 and so @xmath54 , i.e. , @xmath55 with equality if @xmath56 .
this makes @xmath7 most suitable as an approximation of @xmath5 under the @xmath0 norm .
as just established , for a given vector space of cpwl functions @xmath10 , @xmath7 is the function in @xmath10 whose @xmath0 distance to @xmath5 is minimal .
however , the approximation error @xmath57 does not take the same value for every possible @xmath10 ; therefore , we would like to find the optimal partition @xmath58 ( or a reasonable approximation thereof ) corresponding to the space @xmath59 in which the approximation error is minimum , @xmath60 this is a difficult optimization problem : in order to properly specify @xmath7 in and measure its distance to @xmath5 , we need to solve the linear system of equations @xmath41 , whose coefficients depend on the partition itself , which makes the problem symbolically intractable .
numerical algorithms could be used but it is still a challenging problem .
let us examine the analogous problem for the interpolant function @xmath61 defined in section [ sub : linear - interpolation ] .
again , we would like to find the optimal partition @xmath58 corresponding to the space @xmath59 in which the approximation error is minimum , @xmath62 albeit not as difficult as the previous problem , because @xmath8 is more straightforward to define , this is still a challenging non - linear optimization problem .
fortunately , as it is shown next , a good approximation can be easily found under some asymptotic analysis . in the rest of this section
we investigate in detail the error incurred when approximating a function @xmath5 by the interpolant @xmath8 and the orthogonal projection @xmath7 defined in section [ sec : piecewise - linear - approximations ] ( see [ fig : pipvsproj ] ) .
then we derive an approximation to the optimal partition ( section [ sub : approximation - to - the - optimal - partition ] ) that serves equally well for @xmath8 and @xmath7 because , as we will show ( section [ sub : error - in - a - single - interval ] ) , their approximation errors are roughly proportional under the assumption of a sufficiently large number of intervals .
first , let us consider the error incurred when approximating a function @xmath5 , twice continuously differentiable , by its linear interpolant in an interval ( [ fig : errorzone ] ) .
[ thm : lininterpolanterrbound ] the @xmath0 error between a given function @xmath63 and its linear interpolant @xmath64 with @xmath65 , in the interval @xmath66 $ ] , of length @xmath67 , is bounded : @xmath68 where @xmath69 . see appendix [ app : proofinterpolant ] .
formula has the following intuitive explanation : the error measures the deviation of @xmath5 from being straight ( linear ) , and this is directly related to the convexity / concavity of the function , thus the presence of the term @xmath70 to bound the amount of bending .
the other interesting part of is the @xmath71 dependence with respect to the interval size or the local density of knots .
other works in the literature use the weaker bound that does not require carrying out integration @xcite : @xmath72 , for some constant @xmath73 .
function @xmath5 in the subinterval @xmath66 $ ] and two linear approximations . on the left , the linear interpolant @xmath8 given by ; on the right , a general linear segment @xmath74 given by
@xmath75 is a signed distance with respect to @xmath76.,width=317 ] let us now characterize the error of the orthogonal projection @xmath7 .
we do so in two steps : first we compute the minimum error of a line segment in the interval @xmath77 , and then we use an asymptotic analysis to approximate the error of the orthogonal projection . stemming from @xmath78 , we can write any ( linear ) segment in @xmath17 as @xmath79 where @xmath80 and @xmath81 are extra degrees of freedom ( pictured in [ fig : errorzone ] ) with respect the interpolant @xmath8 that allow the line segment to better approximate the function @xmath5 in @xmath17 . by computing the optimal values of @xmath80 and @xmath82 we obtain the following result
. [ thm : minerrlinesegment ] the minimum squared @xmath0 error between a given function @xmath63 and a line segment in an interval @xmath66 $ ] , of length @xmath67 , adopts the expression @xmath83 for some @xmath84 in @xmath85 .
see appendix [ app : prooflinesegment ] .
[ thm : minerrsmalllinesegment ] if @xmath17 is sufficiently small so that @xmath86 is approximately constant within it , say @xmath87 , then @xmath88 substitute @xmath89 for @xmath90 in .
the segments in @xmath7 may not strictly satisfy because @xmath7 must also be continuous across segments .
however , as the size of the intervals @xmath91 become smaller ( due to a finer partition @xmath19 of the interval @xmath13 , i.e. , the number of segments @xmath1 in @xmath19 increases ) we may approximate @xmath7 in @xmath77 by the best ( independently - optimized ) segment @xmath74 and , consequently , use corollary [ thm : minerrsmalllinesegment ] to yield the following result .
[ thm : minerrsmallortproj ] the squared @xmath0 error between a given function @xmath63 and its orthogonal projection @xmath7 in a _ small _ interval @xmath66 $ ] , of length @xmath67 , is @xmath92 where @xmath87 is the approximately constant value that @xmath86 takes within the interval @xmath17 . see appendix [ app : convergence ] . in the same asymptotic situation
, the linear interpolant , ( see ) gives a bigger squared error by a factor of six , @xmath93 now we give a procedure to compute a suboptimal partition of the target interval @xmath13 and then derive error estimates for both @xmath94 and @xmath95 on such a partition and the uniform one .
let us consider a partition @xmath19 of @xmath11 $ ] with subintervals @xmath66,\ , i=1,\ldots , n$ ] .
a suboptimal partition for a given @xmath1 is one in which every subinterval has approximately equal contribution to the total approximation error @xcite , which implies that regions of @xmath5 with higher convexity are approximated using more segments than regions with lower convexity .
let us assume @xmath1 is large enough so that @xmath86 is approximately constant in each subinterval and therefore the bound is tight .
consequently , @xmath96 for some constant @xmath73 , and the lengths of the subintervals ( local knot spacing @xcite ) should be chosen @xmath97 , i.e. , smaller intervals as @xmath70 increases .
hence , the local knot distribution or density is @xmath98 so that more knots of the partition are placed in the regions with larger magnitude of the second derivative .
then , the proposed approximation @xmath99 to the optimal partition is as follows : @xmath100 , @xmath101 , and take knots @xmath102 given by @xmath103 where the monotonically increasing function @xmath104\to[0,1]$ ] is @xmath105 in this procedure , pictured in [ fig : optimization ] , the range of @xmath106 is divided into @xmath1 contiguous sub - ranges of equal length , and the values of @xmath107 are given by the abscissas corresponding to the endpoints of the sub - ranges . with this partition , we can further estimate approximation error bounds in the entire interval based on those of the subintervals .
[ thm : errorsuboptpartpip]the approximation error of the linear interpolant @xmath108 in the partition @xmath99 given by in @xmath11 $ ] is @xmath109 the total squared error for any partition @xmath19 is the sum of squared errors over all subintervals @xmath77 , and by , @xmath110 which , under the condition of error equalization for the proposed partition @xmath99 , becomes @xmath111 to compute @xmath112 , let us sum @xmath113 over all intervals @xmath17 and approximate the result using the riemann integral : @xmath114 whose right hand side is independent of @xmath1 . substituting in gives the desired approximate bound for the error in the interval @xmath11 $ ] .
graphical summary of the proposed knot placement technique .
top : local knot density obtained from input function @xmath5 ( example in [ fig : pipvsproj ] ) ; middle : cumulative knot distribution function @xmath115 given by and knots given by the abscissas corresponding to a uniform partition of the range of @xmath115 , as expressed by ; bottom : approximation of @xmath63 by cpwl interpolant @xmath116 with @xmath117 ( 32 knots ) . in this suboptimal partition
, knots are distributed according to the amount of local convexity / concavity of @xmath5 given by the @xmath118 in the middle plot .
hence , fewer knots are placed around the zeros of the @xmath118 , which correspond to the less steep regions of @xmath115 .
, width=317 ] since the approximation error of @xmath7 in each interval is roughly proportional to that of @xmath8 , as shown in and , the partition is also a very good approximation to the optimal partition for @xmath7 as the number of subintervals @xmath1 increases .
this is briefly stated next .
[ thm : errorsuboptpartortproj]the approximation error of the orthogonal projection @xmath119 in the partition @xmath99 given by in @xmath11 $ ] is @xmath120 both cpwl approximations ( @xmath8 and @xmath7 ) converge to the true function @xmath5 at a rate of at least @xmath2 ( and ) .
we use a similar procedure to derive an estimate error bound for the uniform partition @xmath121 that can be compared to that of the optimized one .
[ thm : erroruniformpart]the approximation error of the linear interpolant @xmath122 in the uniform partition @xmath121 of the interval @xmath11 $ ] is @xmath123 for @xmath121 , we may substitute @xmath124 in and approximate the result using the riemann integral , @xmath125 thus , we can estimate how much we can expect to benefit from optimizing a partition by simply dividing by .
since also shows a @xmath2 convergence , the profit will not depend on @xmath1 ( assuming @xmath1 is large enough ) .
1 determine the subinterval @xmath77 that contains @xmath126 : @xmath127 2 find the fractional distance @xmath128 to the left endpoint of @xmath77 : @xmath129 3 return the interpolated value @xmath130 the implementation of the approximations previously discussed is straightforward ( algorithm [ alg : given - as - input ] ) and we only need to distinguish two different cases depending on whether the partition @xmath19 of @xmath13 has all subintervals of equal width or not .
although their values are different , @xmath61 and @xmath131 are qualitatively the same from the computational viewpoint .
let us discuss the general case that @xmath19 is non - uniform .
line 1 in algorithm [ alg : given - as - input ] implies searching in the ordered array @xmath132 for the right index @xmath133 . since the other steps in the algorithm do not depend on the size of the input vectors , line 1
is the dominant step and makes its run - time complexity @xmath134 . in the particular case that @xmath19 is uniform
, no search is needed to determine the index : @xmath135 and the fractional part @xmath136 .
thus , the run - time complexity of the uniform case is @xmath137 .
there is also no need to store the full array @xmath132 but only its endpoints @xmath24 and @xmath25 , roughly halving the memory needed for a non - uniform partition of the same number of intervals .
consequently , approximations based on a uniform partition are expected to perform better in usual cpus or gpus , computationally - wise , than those based on non - uniform partitions . however , optimizing the partition might lead , depending on the specific objective function , to a reduction in the memory requirements for a desired approximation error .
if memory constraints are relevant or hardware - aided search routines become available , optimized partitions could become practical .
the proposed algorithm is simple to implement either in a cpu or in a gpu . however , there are some implementation details that help further optimize the performance in a gpu .
modern gpus implement a single instruction , multiple data ( simd ) execution model in which multiple threads execute exactly the same instructions ( each ) over different data .
if two threads scheduled together in the same execution group or _ warp _ follow different branches in a conditional statement , both branches are not executed in parallel but sequentially , thereby halving throughput . in the case of @xmath19
being non - uniform , the ( binary ) search performed to find the index @xmath133 could lead to divergent paths for different threads evaluating @xmath138 for different values of @xmath139 , thus reducing throughput .
since the purpose of the proposed algorithm is to save computation at run - time , it is reasonable to assume that at least @xmath1 , and possibly @xmath132 and @xmath140 too , have been determined at compile time .
then , we can particularize the search code in compile time , unrolling the loop needed to implement the binary search and , most importantly , replacing conditional statements with a fixed number of invocations of the ternary operator , which is implemented in the gpu with a single instruction and no code divergence @xcite . in general , accesses to memory in a gpu need to be arranged so that neighbouring threads access neighbouring addresses in memory .
this allows the host interface of the gpu to coalesce operations from several threads into fewer transactions bigger in size . in the described algorithm , if each thread is evaluating @xmath138 for different values of @xmath139 , it is not guaranteed that the memory access pattern complies with this restriction . although some devices provide caching and isolate the programmer from this issue , not all do .
however , there is a case in a regular computer graphics pipeline in which a similar problem arises .
texture data need to be accessed many times in a short period and it is generally impossible to ensure that neighbouring pixels in screen need to read neighbouring texels .
consequently , most gpus do cache texture data and even have dedicated cache layers for it .
we therefore store the arrays of samples in texture memory rather than in global memory to benefit from texture caches .
the use of the texture units to access our samples provides another important benefit . in the usual computer graphics pipeline ,
gpus need to constantly filter texture data . in order to efficiently perform that task ,
texture units are equipped with hardware fixed - function interpolation routines that include linear interpolation @xcite .
therefore , we can use them to speed up line 3 in algorithm [ alg : given - as - input ] ; results in the next section confirm that the texture filtering unit is indeed faster than performing interpolation `` manually '' .
in this section we apply the proposed linearization algorithm to several functions of interest in cybernetics , computer vision and other scientific areas .
the section starts with the analysis of the gaussian function and demonstrates the proposed technique on an image processing application .
the section then analyzes two other sample nonlinear functions of interest : the lorentzian function and the bessel function of the first kind .
the gaussian function @xmath141 is widely used in many applications in every field of science , but it is of particular interest to us in the context of foreground segmentation in video sequences using spatio - temporal nonparametric background models @xcite . to estimate these models , the gaussian function needs to be evaluated approximately 1300 times per pixel and input image , or around 2 billion times per second at modest video quality ( cif , @xmath142 pixels at 15 fps ) . therefore it is crucial to lower the computing time of the gaussian function . in the performed experiments it was enough to approximate the function in the interval @xmath143 $ ] to achieve the required quality .
@xmath0 distance for different approximations to the gaussian function @xmath141 over the interval @xmath143$].,width=317 ] [ fig : l2-distance ] shows @xmath0 distances between the gaussian function and the approximations described in previous sections .
it reports actual distances as well as the approximate and tight upper bounds to the distances , i.e. , results [ thm : errorsuboptpartpip ] to [ thm : erroruniformpart ] .
it can be observed that the @xmath0 distances between @xmath8 and @xmath5 using the uniform and optimized partitions agree well with results [ thm : errorsuboptpartpip ] to [ thm : erroruniformpart ] .
all curves in [ fig : l2-distance ] have similar slope to that of , i.e. , their convergence rate is @xmath2 , and the ratio between the distances corresponding to @xmath7 and @xmath8 is approximately the value @xmath144 that stems from result [ thm : minerrsmallortproj ] and . [ cols="<,<,^,^,^,^,^ " , ] sample detections using nonparametric modeling with cpwl - approximated gaussian kernels . subfigures
( a ) and ( b ) show the original scene and results of the segmentation , respectively , for the tracking application ; subfigures ( c ) and ( d ) show the original scene and results of the segmentation , respectively , for the video - based interface application.,title="fig : " ] + ( a ) sample detections using nonparametric modeling with cpwl - approximated gaussian kernels .
subfigures ( a ) and ( b ) show the original scene and results of the segmentation , respectively , for the tracking application ; subfigures ( c ) and ( d ) show the original scene and results of the segmentation , respectively , for the video - based interface application.,title="fig : " ] + ( c ) sample detections using nonparametric modeling with cpwl - approximated gaussian kernels
. subfigures ( a ) and ( b ) show the original scene and results of the segmentation , respectively , for the tracking application ; subfigures ( c ) and ( d ) show the original scene and results of the segmentation , respectively , for the video - based interface application.,title="fig : " ] + ( b ) sample detections using nonparametric modeling with cpwl - approximated gaussian kernels .
subfigures ( a ) and ( b ) show the original scene and results of the segmentation , respectively , for the tracking application ; subfigures ( c ) and ( d ) show the original scene and results of the segmentation , respectively , for the video - based interface application.,title="fig : " ] + ( d ) the performance of the approximation technique has also been tested on other functions . for example , the lorentzian function with peak at @xmath145 and width @xmath146 is @xmath147 as a probability distribution , is known as the cauchy distribution .
it is an important function in physics since it solves the differential equation describing forced resonance . in such case
, @xmath145 is the resonance frequency and @xmath146 depends on the damping of the oscillator ( and is inversely proportional to the @xmath148 factor , a measure of the sharpness of the resonance ) .
@xmath0 distance for different cpwl approximations to the standard cauchy distribution @xmath149 in the interval @xmath150$].,width=317 ] [ fig : graphicalerrorscauchy ] reports the @xmath0 distances between @xmath151 and the cpwl approximations described in previous sections , in the interval @xmath152 $ ] .
the measured errors agree well with the predicted approximate error bounds , showing the expected @xmath2 trend .
we have measured the mean per - evaluation execution times of the exact function both in the cpu ( 576 ps ) and in the gpu , both using regular and reduced - precision accelerated division ( 19.5 ps and 9.2 ps , respectively ) .
the evaluation times of our proposed strategy coincide with those of the gaussian function ( table [ tab : mean - per - evaluation - execution ] ) because the processing time of the cpwl approximation does not depend on the function values . as expected from the lower complexity of this function ( compared to that of sec .
[ sec : gaussian ] ) , which only involves elementary arithmetic operations , the advantage of our proposal vanishes in the cpu because the operations needed to manually perform interpolation , together with the two values that need to be fetched from memory , are actually more than what the direct evaluation requires . however , note that in the gpu our proposal still remains competitive due to these operations being carried out by the dedicated circuitry of the texture unit .
approximating the bessel function of the first kind @xmath153 is more challenging than approximating the gaussian or lorentzian ( bell - shaped ) functions because it combines both nearly flat and oscillatory parts ( see [ fig : besselj0 ] ) .
[ fig : graphicalerrorsbesselj0 ] presents the @xmath0 approximation errors for @xmath153 using cpwl functions in the interval @xmath152 $ ] . for @xmath154
the measured errors agree well with the predicted approximate error bounds , whereas for @xmath155 the measured errors slightly differ from the predicted ones ( specifically in the optimized partition ) because in these cases the scarce number of linear segments does not properly represent the oscillations . a sample optimized partition and the corresponding cpwl interpolant @xmath156 is also represented in [ fig : besselj0 ] .
the knots of the partition are distributed according to the local knot density ( see [ fig : besselj0 ] , top ) , accumulating in the regions of high oscillations ( excluding the places around the zeros of the @xmath118 ) .
this function is more complex to evaluate in general because it does not have a closed form .
however , our approximation algorithm works equally well on it .
we have measured the mean per - evaluation execution times in the cpu ( 39 ns using the posix extensions of the gnu c library and 130 ns only double - precision provided using the gnu scientific library ) and in the gpu ( 78 ps using the cuda @xcite standard library ) .
we have also measured the execution time of the first term in the asymptotic expansion ( * ? ? ?
* eq . 9.57a ) , valid for @xmath157 , @xmath158 .
$ ] in the gpu the evaluation takes 42 ps using reduced - precision accelerated functions from the sfu for the cosine and multiplicative inverse of the square root . despite this approximation having a non - customizable error ( decreasing as @xmath139 increases ) and using the fastest available implementation ( sfu ) of the operations involved , our strategy still outperforms it by a sizable margin .
this clearly illustrates that the more complex the function to be evaluated is , the greater the advantage of our proposal . * * suboptimal partition for the bessel function @xmath159 in the interval @xmath152 $ ] .
top : local knot density ( @xmath118 ) corresponding to @xmath63 ; bottom : cpwl interpolant @xmath156 with @xmath160 ( @xmath161 knots ) overlaid on function @xmath63.,width=317 ] @xmath0 distance for different cpwl approximations to the bessel function of the first kind @xmath159 in the interval @xmath152$].,width=317 ]
we have developed a fast method to numerically evaluate any continuous mathematical function in a given interval using simpler continuous piecewise linear ( cpwl ) functions and the built - in routines present in the texture units of modern gpus .
our technique allows real - time implementation of demanding computer vision and cybernetics applications that use such mathematical functions . for this purpose
, we analyzed the cpwl approximations given by the linear interpolant and the @xmath0 orthogonal projection of a function .
we carried out a detailed error analysis in the @xmath0 distance to seek a nearly optimal design of both approximations . in the practical asymptotic case of a large number of subintervals @xmath1
, we used error equalization to achieve a suboptimal design ( partition @xmath58 ) and derived a tight bound on the approximation error for the linear interpolant , showing a @xmath2 convergence rate that was confirmed by experimental results .
the @xmath0 orthogonal projection can only but improve upon the results of the linear interpolant , resulting in a gain factor of @xmath4 .
we discussed the computational complexity and the implementation in a gpu of the numerical evaluation of both cpwl approximations .
our experimental results show that our technique can outperform both the quality and the computational cost of previous similar approaches . in particular , the fastest strategy consists of using the texture units in the gpu to evaluate either of the cpwl approximations defined over a uniform partition .
this is normally faster than performing exact function evaluations , even when using the reduced - precision accelerated implementation of the sfus in a gpu , or evaluating the proposed linear approximations without the assistance of the texture units . in practice , the number of subintervals @xmath1 to be considered for representing the nonlinear function can be decided on its own or based on other considerations such as a target approximation error , speed or memory constraints .
although mathematically sound , the strategies based on a suboptimal ( non - uniform ) partition are not practical to implement in current cpu / gpu architectures due to the high cost incurred to find the subinterval that contains the given point of evaluation .
nevertheless , this opens future research paths to explore practical implementations of such approaches using specialized hardware .
recall one of the theorems on interpolation errors @xcite . let @xmath5 be a function in @xmath162 $ ] , and
let @xmath163 be a polynomial of degree @xmath164 or less that interpolates the function @xmath5 at @xmath165 distinct points @xmath166 $ ] .
then , for each @xmath167 $ ] there exists a point @xmath168 $ ] for which @xmath169 in the interval @xmath17 , the linear interpolant @xmath8 is given by . since @xmath8 interpolates the function @xmath5 at the endpoints of @xmath17 , we can apply theorem ( with @xmath170 ) ; hence , the approximation error solely depends on @xmath86 and @xmath139 , but not on @xmath5 or @xmath171 : @xmath172 let us integrate the square of over the interval @xmath17 , @xmath173 next , to simplify the previous integral , let us use the first mean value theorem for integration , which states that if @xmath174\rightarrow\r$ ] is a continuous function and @xmath175 is an integrable function that does not change sign on the interval @xmath176 , then there exists a number @xmath177 such that @xmath178 since @xmath179 and @xmath180 for all @xmath181 , let us apply to compute @xmath182 for some @xmath183 .
finally , if @xmath69 , it is straightforward to derive the @xmath0 error bound from the square root of .
the approximation error corresponding to the line segment is @xmath184 where , by analogy with the form @xmath78 of @xmath185 we defined @xmath186 .
the proof proceeds by computing the optimal values of @xmath80 and @xmath82 that minimize the squared @xmath0 error over the interval @xmath17 : @xmath187 the first term is given in .
the second term is @xmath188 and the third term is , applying and the change of variables @xmath189 to evaluate the resulting integrals , @xmath190 for some @xmath191 and @xmath192 in @xmath85 . substituting previous results in ,
@xmath193 we may now find the line segment that minimizes the distance to @xmath5 by taking partial derivatives with respect to @xmath80 and @xmath81 , setting them to zero and solving the corresponding system of equations .
indeed , the previous error is quadratic in @xmath80 and @xmath81 , and attains its minimum at @xmath194 the resulting minimum squared distance is .
by the triangle inequality , the jump discontinuity at @xmath195 between two adjacent independently - optimized segments is @xmath196 where @xmath197 and @xmath198 are the offsets with respect to @xmath199 of the optimized segments at the left and right of @xmath195 , respectively ; @xmath200 and @xmath201 lie in the interval @xmath202 .
since we are dealing with functions twice continuously differentiable in a closed interval , the absolute value terms in are bounded , according to the extreme value theorem ; therefore , if @xmath203 and @xmath204 decrease ( finer partition @xmath19 ) , the discontinuity jumps at the knots of the partition also decrease . in the limit , as @xmath205 , @xmath206 , i.e.
, continuity is satisfied .
therefore the union of the independently - optimized segments @xmath207 , which is the unique piecewise linear function satisfying both continuity ( @xmath208 ) and minimization of the @xmath0 error .
consequently , if @xmath1 is large we may approximate @xmath209 ; moreover , if @xmath17 is sufficiently small so that @xmath86 is approximately constant within it , @xmath87 , then we use corollary [ thm : minerrsmalllinesegment ] to get .
v. borges , m. batista , and c. barcelos , `` a soft unsupervised two - phase image segmentation model based on global probability density functions , '' in _ ieee int .
systems , man and cybernetics ( smc ) _ , oct 2011 , pp . 16871692 .
k. kapinchev , f. barnes , a. bradu , and a. podoleanu , `` approaches to general purpose gpu acceleration of digital signal processing in optical coherence tomography systems , '' in _ ieee int .
systems , man and cybernetics ( smc ) _ , oct 2013 , pp .
25762580 .
j. kim , e. park , x. cui , h. kim , and w. gruver , `` a fast feature extraction in object recognition using parallel processing on cpu and gpu , '' in _ ieee int .
systems , man and cybernetics ( smc ) _ , oct 2009 , pp . 38423847 .
s. balla - arabe , x. gao , and b. wang , `` gpu accelerated edge - region based level set evolution constrained by 2d gray - scale histogram , '' _ ieee trans . image process .
_ , vol . 22 , no . 7 , pp .
26882698 , 2013 .
s. balla - arabe , x. gao , b. wang , f. yang , and v. brost , `` multi - kernel implicit curve evolution for selected texture region segmentation in vhr satellite images , '' _ ieee trans .
remote sens .
_ , vol .
52 , no . 8 , pp .
51835192 , aug 2014 .
l. sigal , m. isard , h. haussecker , and m. black , `` loose - limbed people : estimating 3d human pose and motion using non - parametric belief propagation , '' _ int .
j. comput . vision _ ,
98 , no . 1 ,
pp . 1548 , 2012 .
c. cuevas , r. mohedano , and n. garca , `` adaptable bayesian classifier for spatiotemporal nonparametric moving object detection strategies . ''
_ optics letters _ , vol .
37 , no .
15 , pp . 31593161 , aug .
2012 .
d. berjn , c. cuevas , f. morn , and n. garca , `` gpu - based implementation of an optimized nonparametric background modeling for real - time moving object detection , '' _ ieee trans .
_ , vol .
59 , no . 2 , pp .
361369 , may 2013 .
m. sylwestrzak , d. szlag , m. szkulmowski , i. gorczyska , d. bukowska , m. wojtkowski , and p. targowski , `` real time 3d structural and doppler oct imaging on graphics processing units , '' in _ proc .
85710y.1em plus 0.5em minus 0.4emint . soc .
optics and photonics , mar
. 2013 .
m. sehili , d. istrate , b. dorizzi , and j. boudy , `` daily sound recognition using a combination of gmm and svm for home automation , '' in _ proc .
20th european signal process .
_ , bucharest , 2012 , pp .
16731677 .
r. tanjad and s. wongsa , `` model structure selection strategy for wiener model identification with piecewise linearisation , '' in _ int .
, telecommun . and
( ecti - con ) _ , 2011 , pp . 553556
. s. ghosh , a. ray , d. yadav , and b. m. karan , `` a genetic algorithm based clustering approach for piecewise linearization of nonlinear functions , '' in _ int .
devices and communications _ , 2011 , pp
g. gallego , d. berjn , and n. garca , `` optimal polygonal @xmath210 linearization and fast interpolation of nonlinear systems , '' _ ieee trans .
circuits syst .
i _ , vol .
61 , no .
11 , pp . 32253234 , nov 2014 .
v. k. pallipuram , n. raut , x. ren , m. c. smith , and s. naik , `` a multi - node gpgpu implementation of non - linear anisotropic diffusion filter , '' in _ symp .
application accelerators in high performance computing _ , jul .
2012 , pp . 1118 . c. cuevas , n. garca , and l. salgado , `` a new strategy based on adaptive mixture of gaussians for real - time moving objects segmentation , '' in _ proc
6811.1em plus 0.5em minus 0.4emint . soc .
optics and photonics , 2008 . | many computer vision and human - computer interaction applications developed in recent years need evaluating complex and continuous mathematical functions as an essential step toward proper operation .
however , rigorous evaluation of this kind of functions often implies a very high computational cost , unacceptable in real - time applications . to alleviate this problem ,
functions are commonly approximated by simpler piecewise - polynomial representations .
following this idea , we propose a novel , efficient , and practical technique to evaluate complex and continuous functions using a nearly optimal design of two types of piecewise linear approximations in the case of a large budget of evaluation subintervals . to this end
, we develop a thorough error analysis that yields asymptotically tight bounds to accurately quantify the approximation performance of both representations .
it provides an improvement upon previous error estimates and allows the user to control the trade - off between the approximation error and the number of evaluation subintervals . to guarantee real - time operation , the method is suitable for , but not limited to , an efficient implementation in modern graphics processing units ( gpus ) , where it outperforms previous alternative approaches by exploiting the fixed - function interpolation routines present in their texture units . the proposed technique is a perfect match for any application requiring the evaluation of continuous functions ; we have measured in detail its quality and efficiency on several functions , and , in particular , the gaussian function because it is extensively used in many areas of computer vision and cybernetics , and it is expensive to evaluate .
v # 1#2#1,#2 # 1y_#1 # 1y_#1 computer vision , image processing , numerical approximation and analysis , parallel processing , piecewise linearization , gaussian , lorentzian , bessel . |
lattice simulations of technicolor inspired models are plagued by known systematic uncertainties @xcite .
although the models under consideration are qcd - like in that they are four dimensional non - abelian gauge theories coupled to dynamical fermions the systematic effects of the interesting models ( those that are either conformal or walking ) are much more difficult to control than in actual qcd . as a result currently there are disagreements between various approaches , discretizations , etc , and universality is not immediately evident @xcite .
clearly the general expectation is that once all systematic effects are controlled and taken into account the results from different approaches and regularizations will agree as they should . in this paper a toy model
is proposed which mimics many of the features of non - abelian gauge theories in the hope that systematic effects can be fully explored .
hopefully these will help controlling the corresponding effects in the much more complicated gauge theories .
the proposed model is the two dimensional @xmath0 non - linear sigma model with a @xmath1 term . at @xmath6
the model served as a toy model of qcd for a long time since it is asymptotically free , features instantons , confinement and dimensional transmutation @xcite .
it is exactly solvable @xcite even at finite volume @xcite .
since the topological term is invisible in perturbation theory the model is asymptotically free for arbitrary @xmath1 .
the dynamics in the infra red is however expected to be very sensitive to @xmath1 . at @xmath2
the model is conjectured @xcite to have a non - trivial infra red fixed point governed by the @xmath7 wznw model at level @xmath8 and , if the conjecture holds , is also exactly solvable . some numerical evidence in support of the conjecture has been presented in @xcite and a recent very detailed study confirming it in @xcite .
the infra red fixed point implies a zero of the @xmath9-function .
this situation is analogous to gauge theories in the conformal window .
for @xmath10 exact solvability is lost but based on continuity one expects that for @xmath1 not much below @xmath3 the @xmath9-function develops a near zero and the renormalized coupling will walk .
this arrangement is analogous to gauge theories just below the conformal window .
hence dialing @xmath1 corresponds to dialing the number of flavors @xmath11 in the gauge theory . in all three scenarios ( confining , walking , conformal ) one
may also introduce an external magnetic field to mimic the effect of a finite quark mass . before exploring the analogies further and investigating the origins of the severe systematic effects
the first task is to establish non - perturbatively that the @xmath1-term is actually a relevant operator and also what the singularity structure of the theory is for @xmath5 .
this is not immediately obvious largely because of the unusual scaling properties of the topological susceptibility and a class of similar observables .
it is well known that small size instantons render the topological susceptibility @xmath12 ill defined in the semi - classical approximation @xcite .
going beyond the semi - classical approximation fully non - perturbative lattice studies have shown that regardless how one improves the details of the lattice implementation a logarithmically divergent susceptibility is obtained at finite physical volume in the continuum limit . moreover ,
all even moments of the total topological charge distribution @xmath13 have the same property . however , the model at @xmath6 is exactly solvable and both the exact solution and the continuum limit of lattice simulations agree that correlators of the topological charge density , e.g. @xmath14 are finite .
the above two observations , namely that certain statistical properties of the total charge distribution @xmath15 are ill defined while at the same time correlators of @xmath16 are finite , might make one wonder whether the total charge operator @xmath17 is an irrelevant operator while @xmath16 is not .
if so , the only consistent continuum value of @xmath18 would be zero and the apparent divergences in the lattice calculations would be regarded as artifacts .
this scenario would imply that the theory defined on the lattice at non - zero @xmath1 leads to an identical continuum theory as the one defined at @xmath6 .
equivalently , the total charge operator inserted into any correlation function would be zero in the continuum theory @xmath19 , while correlation functions of the type @xmath20 are finite .
this scenario would of course invalidate haldane s conjecture about the equivalence of the @xmath2 theory with a non - trivial interacting conformal field theory . in this work
it is shown that there exist quantities built out of the total topological charge operator @xmath17 which have well defined continuum limits and are non - zero .
these observables are differences of connected correlation functions of the topological charge , in other words the cumulants .
each term is logarithmically divergent but the divergence cancels in the difference and moreover they scale correctly in the continuum limit to non - zero values .
showing correct scaling towards the continuum limit in itself would not be sufficient to prove that the @xmath1-term is a relevant operator because the continuum limit value could be zero .
since all cumulant differences are finite there is only a single uv - divergent parameter in the partition function @xmath4 but otherwise it is finite .
while preparing this manuscript the preprint @xcite appeared also with the conclusion that @xmath1 is a relevant coupling .
the method was different though , in @xcite it was shown to high precision that a well defined observable is different in the continuum limit for three different values of @xmath1 implying that @xmath1 can not be irrelevant . in the current work all simulations are carried out at @xmath6 and the same conclusion is reached by showing that certain combinations of the topological charge operator are non - zero in the continuum .
the model in euclidean continuum notation is defined by the action s = d^2x _ s_a _
s_a for the unit 3-vectors @xmath21 , @xmath22 , where @xmath23 is the bare coupling .
only a torus geometry will be considered corresponding to a box of finite linear size @xmath24 which will be regularized by a symmetric lattice .
the corresponding partition function , free energy per unit volume and topological charge distribution of the model at non - zero @xmath1 and volume @xmath25 is given by z ( ) = e^iq = e^-vf ( ) = _ q p(q ) e^iq , with the normalization @xmath26 . since physics is periodic with period @xmath27 in @xmath1 and @xmath28 is a symmetry the free energy per unit volume can be fourier expanded f ( ) = _ n=1^ ( 1 - ( n ) ) f_n .
it has been pointed out in @xcite that in the semi - classical or dilute gas approximation all @xmath29 coefficients vanish except for @xmath30 which is uv divergent due to instantons of size @xmath31 where @xmath32 is the lattice cut - off and @xmath33 is the physical correlation length .
the remaining coefficients come from interactions between instantons .
semi - classical arguments also suggest that for instantons causing the uv divergence in @xmath30 the ratio between their size and their average separation goes to zero in the continuum limit .
this would imply that the interactions responsible for the @xmath34 coefficients are small in the continuum limit hence will not cause them to diverge . to summarize , the semi - classical approximation accounts for a uv divergent @xmath30 and finite @xmath34 coefficients
. a suitable way of addressing whether this statement is true beyond the semi - classical approximation is to consider observables that can be expressed by the @xmath34 coefficients only and calculating them fully non - perturbatively .
the simplest choice is to take the connected correlation functions of the topological charge , _ 2 m = ( -1)^m+1 . |_=0 and consider their differences , _ 2 m = _ 2 m - _ 2m+2 = _ n=2^f_n n^2 m ( 1 - n^2 ) from which @xmath30 drops out .
the first few such correlation functions are _ 2 & = & + _ 4 & = & + _ 6 & = & .all of these are expected to diverge in the continuum limit but their differences are expected to be finite . some numerical evidence has been presented in @xcite in favor of correct scaling behavior for @xmath35 but whether the continuum value is zero or non - zero has not been discussed . in the following
it will be shown to high precision that the expectations from the semi - classical analysis indeed hold non - perturbatively and all moments @xmath18 and all cumulants @xmath36 are logarithmically divergent but the differences @xmath37 are finite .
this implies that there is a single ill - defined constant in @xmath38 namely @xmath30 but otherwise it is finite .
the constant @xmath30 can be removed by an appropriate renormalization condition leading to a finite and universal free energy and partition function for arbitrary @xmath1 .
it is convenient to take the continuum limit on a symmetric periodic lattice @xmath39 of fixed physical volume .
physical length and mass is defined by the second moment correlation length @xmath40 @xcite , = ( ) ^2 ( 2 - ) where m_2n = ( ) ^2n _ t ( 2 ) ^2n c(t ) is given in terms of the zero spatial momentum projection of the 2-point correlation function @xmath41 of the field @xmath21 .
let us introduce @xmath42 .
note that in this notation @xmath43 is not the mass gap in finite volume but rather is simply defined as the inverse of @xmath40 ( which for @xmath44 agrees with the mass gap but not for finite @xmath24 ) .
the physical volume is fixed to @xmath45 .
a novel @xcite topological lattice action is used for the simulations , s = _ i , j s(s_i , s_j ) where the sum is over all neighboring sites and s(s_i , s_j ) = \ { l 0 if s_i s_j > + . in other words
the action is zero for two neighboring vectors if their relative angle is smaller than @xmath46 and infinite otherwise .
the continuum limit is taken by tuning the bare coupling @xmath46 towards zero .
this action is topological because small perturbations of the field @xmath21 do not change the action nevertheless it has been shown that it is in the right universality class @xcite . if @xmath47 powerful improvements exist for the measurement of the topological charge distribution @xcite based on a generalization of the usual cluster algorithms @xcite .
the topological charge operator from @xcite is used assigning an integer charge to each configuration even at finite lattice spacing .
the continuum extrapolation of the cumulant differences will be done through 12 lattice spacings using the parameter values from @xcite listed in table [ tab : parameters ] .
the measured correlation lengths and topological susceptibilities are in agreement with those in @xcite . in the present work
@xmath48 configurations were generated at each volume and every 10@xmath49 was measured for the topological charge distribution and correlation length .
the large number of configurations was necessary because there are huge cancellations between the various terms in the difference of cumulants , especially for @xmath50 .
the third difference , @xmath51 , was already impossible to obtain with the current statistics .
.results for the first few cumulants and their differences for fixed physical volume @xmath45 .
the bare parameters @xmath46 are taken from @xcite . [
cols="^,^,^,^,^,^,^,^",options="header " , ]
it has been known for a long time that the topological susceptibility in the two dimensional @xmath0 model is ill - defined in the continuum . consequently the topological charge distribution @xmath15 does not have a finite continuum limit .
the semi - classical analysis predicts precisely what part of @xmath15 is actually divergent and what part of it is finite . in this work non - perturbative evidence
has been presented supporting the semi - classical result .
the only divergent quantity is the first fourier coefficient of the free energy density , f_1 = - _ 0^f ( ) ( ) , while the remaining part @xmath52 is finite and non - zero .
hence the quantity f_r ( ) = f ( ) - ( 1 - ( ) ) f_1 is finite and universal and one may consider the subtraction an additive renormalization . similarly the renormalized partition function @xmath53 is finite and universal and related to the bare partition function by a multiplicative renormalization . instead of subtracting @xmath30
it is sufficient to subtract only its divergent piece .
the logarithmic singularity is expected to be volume independent .
let us then denote this singular quantity by @xmath54 . since @xmath55 a suitable definition of @xmath54 is the logarithmic singularity in the topological susceptibility which can directly be measured in lattice calculations .
a natural renormalization procedure is then the following : one defines the theory for non - zero @xmath1 by the action s ( ) = s(= 0 ) - iq - ( 1-())vf_1s and all resulting correlation functions related to topology ( i.e. derivatives with respect to @xmath1 ) become finite .
the last term in the full action above is a non - perturbatively generated counter term .
it is important to note that the above renormalization does not mean that @xmath1 itself gets renormalized , the bare @xmath1 is still a physical quantity which does not require renormalization .
it would of course be very interesting to check the volume independence of @xmath54 in lattice simulations .
the finite quantities @xmath34 and @xmath56 are not volume independent and are non - trivial functions of @xmath57 .
since the model is exactly solvable at @xmath58 it would be interesting to derive the first few cumulant differences @xmath59 from the exact solution or at least their value in the infinite volume limit . in any case
the finite and non - zero cumulant differences naturally lead to the conclusion that @xmath1 is a relevant coupling of the theory and the total topological charge operator @xmath17 is a relevant operator despite the ill - defined nature of the moments @xmath60 .
the original motivation was the study of a toy model mimicking confining , walking and conformal behavior in four dimensional gauge theories in order to study the severe systematic effects of the latter .
it was proposed that increasing @xmath1 is analogous to increasing the number of flavors @xmath11 because as @xmath1 goes from zero to @xmath3 the model goes from confining to walking and to conformal . in the toy model a suitable renormalized coupling
is @xmath61 which would then run with the finite volume @xmath24 .
a necessary condition for this analogy to hold was establishing precisely the divergence structure of the partition function at non - zero @xmath1 .
a particular difficulty of the gauge theory calculation can also be studied in the toy model .
it is very difficult to distinguish numerically the following two cases : the theory with zero quark mass just below the conformal window and the theory with a small but non - zero quark mass just inside the conformal window .
both theories walk , the former for the usual reason of being just below the conformal window while the latter because even though it would be conformal for zero quark mass , the non - zero mass drives the coupling away from the would - be fixed point as soon as the running scale goes below the massive fermionic states .
this phenomenon can be mimicked in the toy model by considering it at zero external magnetic field and @xmath62 and also at a small but non - zero external magnetic field and @xmath2 .
both theories are expected to walk and it would be interesting to explore in the toy model what intrinsic features are different despite the similar behavior of the walking coupling constant .
there are a couple of differences between the toy model and gauge theory though .
less important is the fact that while @xmath1 does not enter the perturbative @xmath9-function , @xmath11 does .
more significant is the fact that due to the @xmath63 symmetry and periodicity by @xmath27 the two values @xmath64 and @xmath65 lead to the same continuum theory and it does not have an infra red fixed point ( for non - zero @xmath66 the coupling walks ) .
this means that the zero of the @xmath9-function at @xmath67 is eliminated by arbitrary perturbations of @xmath1 meaning that this zero is a second order zero , unlike in the gauge theory where generically the zero is expected to be first order and is preserved by small perturbations .
hence the @xmath67 model is really analogous to a gauge theory which is exactly at the lower edge of the conformal window
. it would be interesting to find a simple toy model which possesses all essential features and in addition the infra red fixed point is a first order zero of the @xmath9-function and disappears by joining with a non - trivial uv fixed point as expected in gauge theory @xcite .
i am grateful for very helpful discussions with jnos balog , rpd hegeds , sndor katz , julius kuti , martin luscher and ferenc niedermayer .
i am especially grateful to ferenc niedermayer for sharing his cluster algorithm source code .
this work is supported by the eu framework programme 7 grant ( fp7/2007 - 2013)/erc no 208740 and in part by the national science foundation under grant no .
nsf phy05 - 51164 .
i thank the kitp santa barbara for hospitality while preparing the manuscript .
l. del debbio , b. lucini , a. patella , c. pica and a. rago , arxiv:1111.4672 [ hep - lat ] .
s. catterall , l. del debbio , j. giedt and l. keegan , pos lattice * 2011 * , 068 ( 2011 ) [ arxiv:1110.1660 [ hep - ph ] ]
. f. bursa , l. del debbio , d. henty , e. kerrane , b. lucini , a. patella , c. pica and t. pickup _ et al .
_ , phys .
d * 84 * , 034506 ( 2011 ) [ arxiv:1104.4301 [ hep - lat ] ] .
s. sint and p. vilaseca , arxiv:1111.2227 [ hep - lat ] .
z. fodor , k. holland , j. kuti , d. nogradi and c. schroeder , phys .
b * 681 * , 353 ( 2009 ) [ arxiv:0907.4562 [ hep - lat ] ] . t. appelquist , g. t. fleming and e. t. neil , phys
. rev .
d * 79 * , 076010 ( 2009 ) [ arxiv:0901.3766 [ hep - ph ] ] .
x. -y .
jin and r. d. mawhinney , pos lat * 2009 * , 049 ( 2009 ) [ arxiv:0910.3216 [ hep - lat ] ] .
a. deuzeman , m. p. lombardo and e. pallante , phys .
d * 82 * , 074503 ( 2010 ) [ arxiv:0904.4662 [ hep - ph ] ] .
z. fodor , k. holland , j. kuti , d. nogradi , c. schroeder , phys .
b * 703 * , 348 ( 2011 ) [ arxiv:1104.3124 [ hep - lat ] ] .
a. hasenfratz , arxiv:1106.5293 [ hep - lat ] .
v. a. novikov , m. a. shifman , a. i. vainshtein and v. i. zakharov , phys .
* 116 * , 103 ( 1984 ) [ sov .
j. part .
* 17 * , 204 ( 1986 ) ] [ fiz .
chast . atom .
yadra * 17 * , 472 ( 1986 ) ] .
a. b. zamolodchikov and a. b. zamolodchikov , annals phys . * 120 * , 253 ( 1979 ) .
j. balog and a. hegedus , j. phys .
a a * 37 * , 1881 ( 2004 ) [ hep - th/0309009 ] . j. balog and a. hegedus , nucl .
b * 725 * , 531 ( 2005 ) [ hep - th/0504186 ] .
j. balog and a. hegedus , nucl
. phys .
b * 829 * , 425 ( 2010 ) [ arxiv:0907.1759 [ hep - th ] ] .
f. d. m. haldane , phys
. lett .
a * 93 * , 464 ( 1983 ) .
f. d. m. haldane , phys .
lett . * 50 * , 1153 ( 1983 ) .
w. bietenholz , a. pochinsky and u. j. wiese , phys .
* 75 * , 4524 ( 1995 ) [ hep - lat/9505019 ] .
m. bogli , f. niedermayer , m. pepe and u. -j .
wiese , arxiv:1112.1873 [ hep - lat ] .
m. luscher , nucl .
phys . b * 200 * , 61 ( 1982 ) .
g. bhanot , r. f. dashen , n. seiberg and h. levine , phys .
lett . * 53 * , 519 ( 1984 ) .
s. caracciolo , r. g. edwards , a. pelissetto and a. d. sokal , nucl .
b * 403 * , 475 ( 1993 ) [ hep - lat/9205005 ] .
a. patrascioiu and e. seiler , nucl .
suppl . * 30 * , 184 ( 1993 ) .
m. hasenbusch , phys .
d * 53 * , 3445 ( 1996 ) [ hep - lat/9507008 ] .
w. bietenholz , u. gerber , m. pepe and u. -j .
wiese , jhep * 1012 * , 020 ( 2010 ) [ arxiv:1009.2146 [ hep - lat ] ] .
f. niedermayer , phys .
lett . * 61 * , 2026 ( 1988 ) .
u. wolff , phys .
lett . * 62 * , 361 ( 1989 ) . b. berg and m. luscher , nucl .
b * 190 * , 412 ( 1981 )
. j. balog , f. niedermayer and p. weisz , nucl .
b * 824 * , 563 ( 2010 ) [ arxiv:0905.1730 [ hep - lat ] ] .
j. balog , f. niedermayer and p. weisz , phys .
b * 676 * , 188 ( 2009 ) [ arxiv:0901.4033 [ hep - lat ] ] .
h. gies and j. jaeckel , eur .
j. c * 46 * , 433 ( 2006 ) [ hep - ph/0507171 ] .
d. b. kaplan , j. -w .
lee , d. t. son and m. a. stephanov , phys .
d * 80 * , 125005 ( 2009 ) [ arxiv:0905.4752 [ hep - th ] ] .
j. braun , c. s. fischer and h. gies , phys .
d * 84 * , 034045 ( 2011 ) [ arxiv:1012.4279 [ hep - ph ] ] . | a toy model is proposed for four dimensional non - abelian gauge theories coupled to a large number of fermionic degrees of freedom .
as the number of flavors is varied the gauge theory may be confining , walking or conformal .
the toy model mimicking this feature is the two dimensional @xmath0 sigma model with a @xmath1-term . for all @xmath1
the model is asymptotically free . for small
@xmath1 the model is confining in the infra red , for @xmath2 the model has a non - trivial infra red fixed point and consequently for @xmath1 slightly below @xmath3 the coupling walks .
the first step in investigating the notoriously difficult systematic effects of the gauge theory in the toy model is to establish non - perturbatively that the @xmath1 parameter is actually a relevant coupling .
this is done by showing that there exist quantities that are entirely given by the total topological charge and are well defined in the continuum limit and are non - zero , despite the fact that the topological susceptibility is divergent .
more precisely it is established that the differences of connected correlation functions of the topological charge ( the cumulants ) are finite and non - zero and consequently there is only a single divergent parameter in @xmath4 but otherwise it is finite .
this divergent constant can be removed by an appropriate counter term rendering the theory completely finite even at @xmath5 . * an ideal toy model for confining , * * walking and conformal gauge theories : * * the @xmath0 sigma model with @xmath1-term * dniel ngrdi _ etvs university , department for theoretical physics _ _ pzmny pter stny 1/a , budapest 1117 , hungary _ [email protected] |
post - asymptotic giant branch ( post - agb ) stars are short - lived objects ( typically 10@xmath1 years ) , of initial mass between 0.8 8 m@xmath2 . they have evolved from the agb , undergone severe mass loss , and will become planetary nebulae @xcite . this evolutionary stage has been reviewed by , for example , @xcite and @xcite , and is important as it provides an insight into the nucleosynthesis and mixing processes during the later stages of stellar evolution .
most post - agb stars have been identified via their infrared excess , arising from a circumstellar dust shell .
this shell is most readily detected at the earliest stages of post - agb evolution , when it is relatively hot and close to the star . as a result
, the majority of post - agbs found to date have just left the agb , and are of a , f or g spectral type @xcite .
a number of abundance studies have been performed for post - agb stars , with the majority being metal - poor @xcite . however , there are many features of post - agb evolution which are not yet fully understood . for example , several stars show abundance patterns reminiscent of interstellar gas , with si , ca etc following the very low fe abundance , while cno and s are near solar @xcite .
@xcite proposed a mechanism for the formation of such a ` cleaned - up ' photosphere in a post - agb star , where grain formation occurs in the circumstellar shell produced in the agb phase of the star , and subsequent slow accretion of the gas from this material results in the peculiar abundance pattern .
this is most likely to happen during mass transfer between the components of a binary system , or accretion triggered by a binary companion .
indeed , @xcite showed that all the extremely fe deficient post - agb stars known at that time are binaries , providing support for this hypothesis .
although most of the known post - agbs are cool objects , several b - type stars originally classified as being populationi are now believed to be hot post - agbs , on the basis of their derived atmospheric parameters and chemical compositions ( see , for example , @xcite and references therein ) . however , although the b - type stars should be evolving from the cooler post - agbs , there are significant differences .
the most striking is the large c underabundance found for several hot objects , which may imply that these left the agb before the third dredge - up @xcite .
these apparent abundance differences between the hot and cool post - agb candidates indicate that further investigations are required , so that this evolutionary stage may be more fully understood . as a first step , it is clearly important to study hot post - agb objects of known initial metallicity , rather than field stars of unknown origin .
fortunately , a sample of these is available : the well - known uv - bright stars found in some globular clusters , whose post - agb nature has been unambiguously established from their position in the cluster colour - magnitude diagram @xcite .
a review of uv - bright stars in globular clusters can be found in @xcite , while searches for globular cluster post - agb stars include @xcite , @xcite and @xcite . of the b - type post - agb stars studied to date ,
only three have well determined iron abundance estimates , namely zng-1 in m10 @xcite , barnard29 in m13 and roa5701 in @xmath0cen @xcite .
@xcite examined optical spectra of zng-1 , finding an iron abundance larger than the metallicity of the cluster , m10 . barnard29 and roa5701 have been studied several times , primarily due to their relative brightness , using both optical and ultraviolet spectra ( see , for example , ) .
however the iron abundances have previously been estimated using only the uv spectra .
in particular , @xcite have analysed hst observations of these stars and derived fe abundances which are lower than the cluster metallicities , which they explained by the grain formation process discussed above . if barnard29 and roa5701 have such ` cleaned - up ' photospheres , then one would expect them to show similar depletions in mg and si , which are also subject to grain formation .
however , previous studies of barnard29 and roa5701 indicate no depletions in mg nor si . clearly , the abundances of barnard29 and roa5701 need revisiting , in particular those for fe . in this paper , high resolution
optical and ultraviolet spectra of barnard29 and roa5701 are analysed , using model atmosphere techniques to estimate atmospheric parameters and chemical compositions .
the observations and data reduction are discussed in section 2 , while section 3 details the analysis methods and results . in section 4
we present a discussion of our results , and in particular we assess the accuracy of fe abundance determinations for b - type stars using ultraviolet spectra .
optical spectra of barnard29 and roa5701 were obtained during two observing runs , the former with the 2.7-m harlan j. smith telescope ( mcdonald observatory ) on three nights between 2005 april 30 and may 5 , and the latter with the mpg / eso 2.2-m telescope ( la silla , chile ) on 2006 april 10 .
ultraviolet spectra of both stars were taken from the hubble space telescope ( hst ) archive .
details of the observations are provided below .
barnard 29 was observed with the 2d - coud spectrograph @xcite , using a 2.4 arcsec slit to achieve a spectral resolution ( @xmath3=@xmath4/@xmath5 ) of @xmath635000 and a wavelength coverage of @xmath6370010000 .
due to cloud , data were taken only on three nights , resulting in 30 half hour exposures , or 15 hours on - source .
a th - ar hollow cathode lamp was observed at the beginning and end of each night , with the centres of the lines stable to better than @xmath60.1 pixel .
test exposures were taken at the start of the run , to ensure that optical ghosts ( ` the picket fence ' ; @xcite ) and bad columns did not overlap with wavelength regions of interest , in particular the feiii lines around 4420 .
also , in order to ensure that the weak absorption line features detected were real , the central wavelength was changed , by @xmath60.4 , after the first night and remained in the new position for the other two clear nights .
the data were reduced using standard methods within the image reduction and analysis facility ( iraf ; @xcite ) .
this reduction included bias subtraction and flatfielding , with stellar extraction and cosmic ray removal performed using doecslit @xcite and scombine .
the signal to noise ratio ( s / n ) per pixel in the co - added spectrum was @xmath6130 per pixel at @xmath64420 (corresponding to @xmath6180 per resolution element ) , and this was read into the starlink package dipso @xcite for further analysis ( see section [ sec_dip ] ) .
the optical spectra of roa5701 were obtained using the high - resolution fibre - fed extended range optical spectrograph ( feros ; @xcite ) , with the ev 2k@xmath74k ccd detector and a 79 lines mm@xmath8 chelle grating . in total , roa5701 was observed for 9 hours , resulting in a spectral resolution of @xmath3 @xmath646000 and a wavelength coverage of @xmath635609200 .
the data were reduced on - line , using the standard feros data reduction system ( drs ) , implemented with eso midas software .
this reduction includes the extraction and de - blazing of the chelle orders , wavelength calibration and merging of the orders .
the reduced spectral exposures were combined into a single spectrum using scombine within iraf , and this was imported into dipso for further analysis ( see section [ sec_dip ] ) .
an average s / n ratio of @xmath685 per pixel at @xmath64420 was obtained , corresponding to @xmath6120 per resolution element .
initially , the continua of the optical spectra for both barnard29 and roa5701 were normalised using low - order polynomials fitted to spectral regions free from absorption features .
the line fitting program elf within dipso was then employed to measure absorption line equivalent widths , by fitting gaussian profiles to the lines ( see @xcite for further details ) .
balmer lines and diffuse hei line equivalent widths were not measured , as instead theoretical profiles were fitted to the normalized data ( see , for example , fig .
[ fig_hga ] ) . at 4341 in the spectra of barnard29 ( upper panel ) and roa5701 ( lower panel ) .
the theoretical profiles have been generated for surface gravities of 2.75 , 3.0 and 3.25 dex ( upper , middle and lower smooth curves , respectively ) for barnard29 , implying a surface gravity of 2.95 dex .
for roa5701 , results have been calculated for surface gravities of 3.0 , 3.25 and 3.5 dex ( upper , middle and lower smooth curves , respectively ) , implying log _ g _ = 3.3 dex.,scaledwidth=50.0% ] the ultraviolet spectra of barnard29 and roa5701 were obtained directly from the hst multimission archive ( http://archive.stsci.edu ; z3ec0204 t and z3ec0104 m respectively ) .
they were observed with the goddard high resolution spectrograph ( ghrs ) using the g200 m grating ( @xmath3 = 25000 ) and the large science aperture ( lsr ) , giving a wavelength coverage of @xmath618601906 .
fp splits were not used for these observations @xcite .
barnard29 was observed on 1996 november 30 for 4570s , and roa5701 on 1996 august 3 for 4711s .
following the standard pipeline reduction , the spectra were co - aligned within iraf , using the stsdas package ( version 3.4 ) .
the combined spectrum for each star was then read into dipso and cross - correlated with theoretical spectra to obtain the stellar lsr radial velocities of 228 and + 258 km s@xmath8 for barnard29 and roa5701 , respectively .
these are consistent with the radial velocities of m13 ( 227.2 km s@xmath8 , @xcite ) and @xmath0cen ( + 240 km s@xmath8 , @xcite ) .
an interstellar reddening correction , of e(b v ) = 0.02 @xcite for m13 and e(b v ) = 0.11 @xcite for @xmath0cen , was applied to the appropriate spectrum .
the spectra were then binned , with a wavelength interval of 0.05 , resulting in a wavelength coverage of 18681906 for barnard29 and 18651903 for roa5701 .
non - local thermodynamic equilibrium ( non - lte ) model atmosphere grids , calculated using tlusty and synspec , have been used to derive atmospheric parameters and chemical abundances .
a summary of the procedures involved are given in @xcite , while a more detailed discussion of the grids and methods may be found in @xcite and @xcite , and also at http://star.pst.qub.ac.uk .
briefly , four grids have been produced with metallicities corresponding to [ fe / h ] = 7.5 dex for the galaxy , 7.2 dex and 6.8 dex for the large ( lmc ) and small ( smc ) magellanic clouds , respectively , and 6.4 dex for low metallicity regimes .
for each metallicity grid , approximately 3000 models were calculated for a range of effective temperatures from 12000 to 35000 k , logarithmic gravities ( log _ g _ = 4.5 dex down to the eddington limit ) and microturbulences of 0 to 30 km s@xmath8 .
the iron abundance was fixed at each grid , while the light element abundances ( c , n , o , mg , si , s ) were allowed to vary from 0.8 to + 0.8 dex around their base ( metallicity ) value .
iron is assumed to dominate the metal line blanketing .
however , the exclusion of nickel in our models should not be a significant source of error in the atmospheric parameters and chemical composition @xcite .
the atmospheric structure is therefore defined by the iron abundance ( metallicity ) , effective temperature , gravity and microturbulence .
it is also assumed that light element abundances can vary with negligible affect on the atmospheric structure , as shown in tests performed by @xcite .
theoretical profiles and equivalent widths of light element transitions were calculated using the models .
a gui interface in idl allows the user to access the theoretical equivalent widths and interpolate these to calculate equivalent widths and abundance estimates for @xmath6200 metal lines , for any set of atmospheric parameters .
abundance increments of 0.4 dex used in the grids were found to be sufficiently small that no significant errors were incurred due to the interpolation procedures @xcite .
the atmospheric parameters of each star were determined from the optical spectra . as the parameters are interrelated , an iterative process is used ( see , for example , ) .
studies of giants in m13 yield a cluster metallicity of [ fe / h ] = 1.6 @xcite , while for @xmath0cen , which contains at least three stellar populations , the main cluster giant population indicates [ fe / h ] = 1.6 @xcite . hence , for both stars the grid corresponding to low metallicity regions ( 6.4 dex ; 1.1 dex lower than galactic ) was used , as this was the lowest metallicity model available and thus the most appropriate based on the metallicities of the respective clusters .
@xcite found that using grids of different iron abundances changed the effective temperature estimate by typically @xmath9 500 k , while the gravity and microturbulence estimates were unaffected , so the use of this grid should not be a major source of error .
the adopted atmospheric parameters and associated errors for both stars are listed in table [ tab_ap ] .
@lccccccc star & @xmath10 & log _ g _ & @xmath11 & @xmath12sin_i _ + & @xmath13 & dex & km s@xmath8 & km s@xmath8 + + barnard29 & 20 @xmath14 1 & 2.95 @xmath14 0.1 & 4 @xmath14 2 & 0 + roa5701 & 25 @xmath14 1 & 3.3 @xmath14 0.1 & 1 @xmath15 & 3 + effective temperature estimates , @xmath10 , were determined using the silicon ionization balance , with siii / siiii employed for barnard29 and siiii / siiv for roa5701 .
surface gravities , log _ g _ , were derived by over - plotting the observed spectrum in the regions of the balmer lines with theoretical profiles , with associated errors assessed from the uncertainty in the fitting ( fig .
[ fig_hga ] ) .
the microturbulence , @xmath11 , was determined using the observed oii and siiii lines , by removing the dependance of abundance on line strength , i.e. a plot of abundance against line strength having a zero gradient @xcite .
individual multiplets were employed in this procedure , to reduce errors which may arise from using many different multiplets @xcite .
initial atmospheric parameters for barnard29 were taken from @xcite , and are @xmath10 = 20000 k , log _ g _ = 3.0 and @xmath11 = 10 km s@xmath8 .
the temperature estimate remained unchanged at 20000 k with an associated error of @xmath141000 k , the gravity was constrained to log _ g _ = 2.95 @xmath14 0.10 , while a determination of the microturbulence led to a value of @xmath11 = 4 @xmath14 2 km s@xmath8 . for roa5701 ,
initial parameters were taken from @xcite , namely @xmath10 = 25000 k , log _ g _ = 3.25 and @xmath11 = 5 km s@xmath8 .
the present analysis led to no change in the effective temperature .
as no absorption line was detected for heii 4541 , assuming a normal helium abundance also allowed us to place an independent upper limit of @xmath10 @xmath16 26000 k. therefore , the associated error for @xmath10 is @xmath141000 k. the gravity was determined to be log _ g _ = 3.3 @xmath14 0.1 , and the microturbulence @xmath11 = 1 @xmath15 km s@xmath8 .
in addition to the microturbulent broadening , some rotational broadening may occur .
the effect of this was investigated by fitting instrumentally broadened non - lte theoretical profiles for the siiii multiplet at 4560 , calculated using the appropriate @xmath10 , log _ g _ and @xmath11 values , to the observed spectra .
the theoretical profiles were rotationally broadened and overlain in order to determine the most appropriate @xmath12sin_i _ ( see @xcite for further details ) .
for barnard29 , the profile fitting resulted in a value of @xmath12sin_i _
= 0 km s@xmath8 , while in the case of roa5701 , a @xmath12sin_i _ = 3 km s@xmath8 was obtained .
@xcite discuss the difficulty in differentiating between the effects of the two types of broadening when such low values are obtained . in order to place an upper limit on the @xmath12sin_i _ values ,
a microturbulence of 0 km s@xmath8 was assumed for both stars , as a non - zero microturbulence would lead to lower @xmath12sin_i _ estimates @xcite .
this resulted in , for barnard29 @xmath12sin_i _
@xmath9 6 km s@xmath8 and for roa5701 , @xmath12sin_i _ @xmath9 3 km s@xmath8 .
abundances for both stars were determined using the adopted atmospheric parameters ; atomic data were taken from http://star.pst.qub.ac.uk .
table [ tab_all ] contains the equivalent width measurements for the observed metal lines , along with the associated abundance estimates .
if more than one line contributed to an equivalent width , the lines were treated as blends within tlusty .
there are currently no non - lte grids available for aliii , siii or feiii , although sulphur is included in our non - lte model atmosphere calculations using a relatively simple model . using tlusty ,
models were calculated , adopting the appropriate atmospheric parameters .
these were then employed to generate theoretical lte equivalent widths ( see @xcite ) , which were used to derive the abundance estimates shown in table [ tab_all ] , with errors calculated by utilising the same method as for other elements .
@llcccccccc & & + rest @xmath4 & species & observed @xmath4 & w & quality@xmath17 & abundance@xmath18 & observed @xmath4 & w & quality@xmath17 & abundance@xmath18 + ( ) & & ( ) & ( m ) & & & ( ) & ( m ) + + 4267.00 & cii@xmath19 & @xmath20 & @xmath1610 & @xmath20 & @xmath166.13 & @xmath20 & @xmath1610 & @xmath20 & @xmath166.27 + 6578.05 & cii & @xmath20 & @xmath1610 & @xmath20 & @xmath166.64 & @xmath20 & @xmath1610 & @xmath20 & @xmath166.54 + 3995.00 & nii@xmath19 & 3995.00 & 65 & a & 7.22 & 3995.00 & 45 & a & 6.99 + 4035.08 & nii & 4035.09 & 12 & c & 7.62 & @xmath20 & @xmath20 & @xmath20&@xmath20 + 4041.31 & nii & 4041.32 & 25 & a & 8.30 & @xmath20 & @xmath20 & @xmath20&@xmath20 + 4043.53 & nii & 4043.52 & 14 & b & 7.47 & @xmath20 & @xmath20 & @xmath20&@xmath20 + 4227.74 & nii & 4227.71 & 8 & c & 7.36 & @xmath20 & @xmath20 & @xmath20&@xmath20 + 4241.78 & nii@xmath19 & 4241.79 & 21 & a & 7.62 & 4241.82 & 15 & c & 7.67 + 4432.73 & nii@xmath19 & 4432.72 & 12 & c & 7.39 & 4432.75 & 11 & c & 7.35 + 4447.03 & nii & 4446.99 & 30 & a & 7.28 & 4447.03 & 18 & b & 6.95 + 4601.48 & nii & 4601.47 & 27 & a & 7.34 & @xmath20 & @xmath20 & @xmath20&@xmath20 + 4607.16 & nii & 4607.14 & 21 & b & 7.29 & @xmath20 & @xmath20 & @xmath20&@xmath20 + 4613.86 & nii & 4613.85 & 18 & a & 7.34 & @xmath20 & @xmath20 & @xmath20&@xmath20 + 4621.29 & nii & 4621.38 & 16 & a & 7.12 & @xmath20 & @xmath20 & @xmath20&@xmath20 + 4630.54 & nii & 4630.56 & 50 & a & 7.28 & 4630.55 & 26 & b & 6.93 + 4643.09 & nii & 4643.10 & 27 & a & 7.33 & 4643.10 & 11 & c & 6.93 + 5001.10 & nii@xmath19@xmath21 & 5001.12 & 30 & b & 7.36 & 5001.16 & 20 & a & 6.92 + 5001.50 & nii@xmath19@xmath21 & 5001.47 & 37 & b & @xmath20 & 5001.50 & 25 & a & @xmath20 + 5002.70 & nii@xmath19@xmath21 & 5002.68 & 8 & c & @xmath20 & @xmath20 & @xmath20 & @xmath20&@xmath20 + 5005.15 & nii & 5005.15 & 47 & b & 7.34 & 5005.17 & 30 & a & 6.88 + 5007.33 & nii & 5007.23 & 30 & c & 7.68 & 5007.34 & 11 & b & 6.96 + 3911.96 & oii@xmath19@xmath21 & @xmath20 & @xmath20&@xmath20 & @xmath20 & 3911.98 & 27 & a & 7.64 + 3912.09 & oii@xmath19@xmath21 & @xmath20 & @xmath20&@xmath20 & @xmath20 & @xmath20 & @xmath20 & @xmath20&@xmath20 + 3919.28 & oii & @xmath20 & @xmath20&@xmath20 & @xmath20 & 3919.30 & 19 & b & 7.74 + 3945.04 & oii & @xmath20 & @xmath20&@xmath20 & @xmath20 & 3945.04 & 15 & b & 7.62 + 3954.37 & oii & @xmath20 & @xmath20&@xmath20 & @xmath20 & 3954.38 & 29 & a & 7.65 + 3982.72 & oii & @xmath20 & @xmath20&@xmath20 & @xmath20 & 3982.72 & 20 & b & 7.78 + 4069.62 & oii@xmath19@xmath21 & 4069.61 & 13 & b & 7.55 & 4069.64 & 36 & a & 7.76 + 4069.89 & oii@xmath19@xmath21 & 4069.87 & 17 & b & @xmath20 & 4069.90 & 47 & a & @xmath20 + 4072.16 & oii & 4072.13 & 23 & a & 7.52 & 4072.17 & 57 & a & 7.72 + 4075.86 & oii & 4075.83 & 30 & a & 7.54 & 4075.87 & 68 & a & 7.71 + 4078.84 & oii & @xmath20 & @xmath20&@xmath20 & @xmath20 & 4078.85 & 19 & b & 7.79 + 4089.29 & oii & @xmath20 & @xmath20&@xmath20 & @xmath20 & 4089.29 & 33 & a & 8.02 + 4132.80 & oii & @xmath20 & @xmath20&@xmath20 & @xmath20 & 4132.81 & 24 & b & 7.76 + 4185.46 & oii & @xmath20 & @xmath20&@xmath20 & @xmath20 & 4185.47 & 24 & a & 7.67 + 4294.79 & oii & @xmath20 & @xmath20&@xmath20 & @xmath20 & 4294.77 & 11 & c & 7.56 + 4303.84 & oii@xmath19 & @xmath20 & @xmath20&@xmath20 & @xmath20 & 4303.85 & 15 & c & 7.39 + 4317.14 & oii@xmath19 & 4317.12 & 10 & a & 7.35 & 4317.15 & 37 & a & 7.78 + 4319.63 & oii & 4319.61 & 13 & a & 7.52 & 4319.64 & 46 & a & 7.95 + 4349.43 & oii & 4349.43 & 25 & a & 7.59 & 4349.43 & 71 & a & 7.96 + 4351.26 & oii@xmath19 & 4351.25 & 9 & c & 7.29 & 4351.27 & 38 & a & 7.54 + 4366.90 & oii@xmath19 & 4366.86 & 12 & b & 7.40 & 4366.90 & 37 & a & 7.72 + 4369.27 & oii & @xmath20 & @xmath20&@xmath20 & @xmath20 & 4369.30 & 10 & c & 7.80 + 4395.94 & oii & @xmath20 & @xmath20&@xmath20 & @xmath20 & 4395.95 & 11 & c & 7.69 + 4414.91 & oii & 4414.90 & 26 & a & 7.53 & 4414.91 & 70 & a & 7.86 + 4416.98 & oii & 4416.98 & 20 & a & 7.64 & 4416.98 & 52 & a & 7.89 + 4452.38 & oii & @xmath20 & @xmath20&@xmath20 & @xmath20 & 4452.37 & 16 & b & 7.90 + 4590.97 & oii & 4590.98 & 11 & b & 7.41 & 4590.99 & 43 & a & 7.66 + 4596.17 & oii & 4596.17 & 8 & c & 7.38 & 4596.19 & 33 & a & 7.61 + 4609.44 & oii@xmath19 & @xmath20 & @xmath20&@xmath20 & @xmath20 & 4609.44 & 12 & c & 7.36 + 4638.86 & oii & 4638.87 & 14 & a & 7.54 & 4638.87 & 50 & a & 7.87 + 4641.81 & oii & 4641.83 & 27 & a & 7.52 & 4641.83 & 79 & a & 7.85 + 4649.14 & oii@xmath19@xmath21 & 4649.14 & 39 & a & 7.53 & 4649.16 & 102 & a & 7.86 + 4650.84 & oii@xmath19@xmath21 & 4650.84 & 14 & a & @xmath20 & 4650.86 & 48 & a & @xmath20 + 4661.64 & oii & 4661.63 & 18 & a & 7.61 & 4661.65 & 50 & a & 7.81 + 4673.74 & oii & @xmath20 & @xmath20&@xmath20 & @xmath20 & 4673.75 & 13 & b & 7.90 + 4676.24 & oii & 4676.21 & 16 & b & 7.62 & 4676.25 & 44 & a & 7.83 + 4699.00 & oii@xmath19@xmath21 & @xmath20 & @xmath20&@xmath20 & @xmath20 & 4699.03 & 34 & c & 7.63 + 4699.22 & oii@xmath19@xmath21 & @xmath20 & @xmath20&@xmath20 & @xmath20 & 4699.24 & @xmath20 & b & @xmath20 + 4705.35 & oii & @xmath20 & @xmath20&@xmath20 & @xmath20 & 4705.37 & 37 & a & 7.70 + 4710.01 & oii@xmath19 & @xmath20 & @xmath20&@xmath20 & @xmath20 & 4710.04 & 11 & c & 7.93 + @llccccccccc & & + rest @xmath4 & species & observed @xmath4 & w & quality@xmath17 & abundance@xmath18 & observed @xmath4 & w & quality@xmath17 & abundance@xmath18 + ( ) & & ( ) & ( m ) & & & ( ) & ( m ) + + 4890.93 & oii & @xmath20 & @xmath20&@xmath20 & @xmath20 & 4890.87 & 11 & a & 7.79 + 4906.83 & oii & @xmath20 & @xmath20&@xmath20 & @xmath20 & 4906.84 & 15 & c & 7.60 + 4924.53 & oii & @xmath20 & @xmath20&@xmath20 & @xmath20 & 4924.53 & 28 & a & 7.91 + 4941.07 & oii & @xmath20 & @xmath20&@xmath20 & @xmath20 & 4941.10 & 12 & c & 7.90 + 4943.00 & oii@xmath19 & @xmath20 & @xmath20&@xmath20 & @xmath20 & 4943.01 & 21 & a & 7.94 + 4481.13 & mgii@xmath19@xmath21 & 4481.12 & 19 & a & 6.10 & 4481.16 & 10 & b & 6.15 + 4481.33 & mgii@xmath19@xmath21 & 4481.32 & 11 & b & @xmath20 & 4481.34 & 7 & b & @xmath20 + 4128.05 & siii & 4128.04 & 11 & c & 6.45 & @xmath20 & @xmath20 & @xmath20&@xmath20 + 4552.62&siiii & 4552.60 & 71 & a & 6.45 & 4552.63 & 67 & a & 6.18 + 4567.82&siiii & 4567.86 & 53 & a & 6.43 & 4567.86 & 50 & a & 6.18 + 4574.76&siiii & 4574.77 & 30 & a & 6.48 & 4574.77 & 23 & a & 6.16 + 5739.74&siiii & 5739.76 & 26 & b & 6.59 & 5739.78 & 23 & b & 5.96 + 4088.85 & siiv & @xmath20 & @xmath20&@xmath20 & @xmath20 & 4088.89 & 33 & a & 6.15 + 4116.10 & siiv & @xmath20 & @xmath20&@xmath20 & @xmath20 & 4116.12 & 21 & b & 6.17 + 4512.54&aliii & 4512.58 & 7 & b & 5.13 & @xmath20 & @xmath20 & @xmath20&@xmath20 + 4529.18&aliii & 4529.18 & 11 & b & 5.15 & @xmath20 & @xmath165 & @xmath20 & @xmath164.87 + 4253.59 & siii & 4253.57 & 7 & a & 5.72 & 4253.60 & 14 & b & 5.62 + 4419.60 & feiii & 4419.60 & 3 & c & 6.03 & @xmath20 & @xmath20 & @xmath20&@xmath20 + 5156.12 & feiii &
5156.09 & 5 & c & 6.08 & @xmath20 & @xmath20 & @xmath20&@xmath20 + 5833.93 & feiii & 5833.94 & 4 & c & 6.21 & 5833.99 & @xmath166 & c & @xmath166.10 + + @xmath17 a indicates that the equivalent width measurement should be accurate to better than @xmath1410 per cent , b to better than @xmath1420 per cent , and c to less than @xmath1420 per cent .
+ @xmath18 logarithmic abundance [ m / h ] on the scale log[h ] = 12.00 .
+ @xmath19 treated as blends in tlusty .
+ @xmath21 equivalent widths of adjacent lines were combined to produce the corresponding abundances shown .
table [ tab_ab ] contains a summary of the abundances for each species , along with the associated error estimates and the number of lines observed .
also included are the abundances previously derived for the objects , and those of normal b - type stars , both in the galactic cluster ngc6611 @xcite and in the local field .
the former were determined using the same tlusty code as employed in the present paper . also included
are typical interstellar medium ( ism ) abundances for the elements concerned , as listed in @xcite .
the c , n and o abundances are taken from @xcite and respectively , with all other elements given by @xcite .
the al abundance , however , is taken as 70 per cent of the adopted solar abundance of @xcite , as ism abundances appear to be reduced by 20 30 per cent with respect to solar abundances .
the errors associated with the absolute abundances ( see table [ tab_ab ] ) arise from systematic and random errors ( full details can be found in @xcite ) .
random errors are connected with the data analysis , for example : line fitting errors , oscillator strengths and observational uncertainties .
the random uncertainty is taken as the standard deviation in the abundances of a given species , divided by the square root of the number of lines observed for that species .
where only one line was detected , the standard deviation is that of the best observed species , e.g. oii .
systematic errors arise from the uncertainties in the atmospheric parameters , and are estimated by varying each parameter by its associated errors .
the total uncertainty is the square root of the sum of the squares of the random and systematic errors .
we note that the scatter of individual abundances for each species is reasonably small , with any discrepancies probably arising from poorly determined lines and possible errors in atomic data .
@l cccccccccccc species & & & & ism + & & con94 & dix98 & moe98@xmath22 & & moe98@xmath22&tho06 & kil92/94 & hun06 & + + cii & @xmath166.64 & ( 1 ) & @xmath166.70 & 6.15 & @xmath20 & @xmath166.54 & ( 1 ) & @xmath165.85 & @xmath166.51 & 8.20 & 7.95 & 8.38 + nii & 7.43 @xmath14 0.12 & ( 17 ) & 7.30 & @xmath20 & @xmath20 & 7.06 @xmath14 0.12 & ( 9 ) & 6.86 & 7.05 & 7.69 & 7.59 & 7.88 + oii & 7.50 @xmath14 0.24 & ( 17 ) & 7.60 & @xmath20 & @xmath20 & 7.76 @xmath14 0.10 & ( 41 ) & 7.97 & 7.75 & 8.55 & 8.55 & 8.69 + mgii & 6.10 @xmath14 0.15 & ( 1 ) & 6.00 & @xmath20 & @xmath20 & 6.15 @xmath14 0.21 & ( 1 ) & @xmath20 & 6.27 & 7.38 & 7.32 & 7.40 + siii & 6.45 @xmath14 0.20 & ( 1 ) & @xmath20 & @xmath20 & @xmath20 & @xmath20 & & @xmath20&@xmath20 & @xmath20 & @xmath20 & @xmath20 + siiii & 6.45 @xmath14 0.26 & ( 3 ) & 6.30 & @xmath20 & @xmath20 & 6.17 @xmath14 0.20 & ( 3 ) & 6.14 & 5.94 & 7.28 & 7.41 & 7.27 + siiv & @xmath20 & & @xmath20 & @xmath20 & @xmath20 & 6.16 @xmath14 0.40 & ( 2 ) & @xmath20 & 5.95 & @xmath20 & @xmath20 & @xmath20 + aliii & 5.14 @xmath14 0.10 & ( 2 ) & 5.19 & @xmath20 & @xmath20 & @xmath164.87 & ( 1 ) & @xmath20 & @xmath165.21 & 6.22 & @xmath20 & 6.33 + siii & 5.72 @xmath14 0.26 & ( 1 ) & @xmath166.29 & 5.34 & @xmath20 & 5.62 @xmath14 0.23 & ( 1 ) & @xmath20 & 5.71 & 7.25 & @xmath20 & 7.09 + feiii & 6.07 @xmath14 0.12 & ( 3 ) & @xmath166.70 & 5.28 & 5.30 & @xmath166.10 & ( 1 ) & 4.79 & @xmath20 & @xmath20 & 7.50 & 7.43 + @xmath22 average of curve - of - growth and spectrum synthesis values . + a normal he abundance has been assumed throughout the analysis of both stars , which was tested by over - plotting theoretical spectra on the observations in regions containing hei absorption features .
there is good agreement to within the uncertainties of the atmospheric parameters , suggesting this is a valid assumption .
when lines are badly blended with hydrogen or helium lines , they are not included in the non - lte calculations . for both stars ,
the best observed species are nii and oii , with the majority of equivalent widths believed to be accurate to better than 20 per cent .
if the lines of poorer quality are removed , the abundance estimates remain unchanged ( within the error bars ) , indicating that data of poorer quality do not significantly affect the abundances obtained .
the c abundances for both barnard29 and roa5701 in table [ tab_ab ] are quoted as upper limits . due to the quality of the spectra , it is was only possible to place upper limits on the equivalent widths of the cii lines at 4267 and 6578 .
@xcite finds that the 4267 multiplet gives a limiting precision of @xmath6@xmath14 0.2 dex , over a @xmath10 range of 15000 31000 k ( as reproduced by @xcite ) .
however , for @xmath10 @xmath16 25000 k , the line at 6578 is somewhat more reliable than that at 4267 . therefore , the upper limits given in table [ tab_ab ] are based on the 6578 line alone .
the mgii doublet at 4481 was observed and treated as a blend within tlusty , and for both stars the derived abundances appear reliable .
we note that the si abundance estimates are sensitive to the adopted atmospheric parameters .
for example , for barnard29 a change in @xmath10 of @xmath141000 k leads to a variation of @xmath140.15 dex in the abundance of siii and @xmath140.21 for siiii . in the case of roa5701
, the same temperature change produces an abundance variation of @xmath140.10 dex for siiii and @xmath140.34 dex for siiv .
the siiii line at 5739 was observed in both stars . however , as there are no non - lte grids available for this spectral region , theoretical lte equivalent widths were generated , as for siii , aliii and feiii .
the resulting lte abundances are 6.59 dex for barnard29 and 5.96 dex for roa5701 , which compare well to the siiii results in table [ tab_ab ] .
@xcite and @xcite found that lte abundances agree to 0.2 dex with corresponding non - lte abundances for b - type stars , consistent with the results here .
however , as the 5739 abundances are in lte , they are not included in the relevant non - lte siiii results listed in table [ tab_ab ] .
the siii 4254 line was used to derive the s abundance in both barnard29 and roa5701 . as previously noted ,
these abundances are based on theoretical lte siii equivalent widths .
@xcite find that non - lte effects appear to be negligible for this species .
additionaly , we have carried out test calculations and find little difference between the siii spectrum predicted using lte or non - lte .
hence our use of an lte approach is probably valid .
two aliii lines , at 4512 and 4259 , were observed for barnard29 .
no al features were detected for roa5701 , but an upper limit on the al abundance was set using the 4529 line .
the al abundances are treated in lte , but should be reliable as @xcite state that when using the dominant ion stage , lte abundance calculation will give good agreement with non - lte abundances .
three feiii lines were observed at 4419 , 5156 and 5833 in barnard29 , with one in roa5701 at 5833 . the s / n of the local continuum for the lines are @xmath6130 , 155 , 150 and 110 per pixel , respectively .
lte calculations were used to determine the fe abundance , and fig .
[ fig_fe_op ] shows theoretical profiles corresponding to the measured abundances , over - plotted onto the observed spectra of barnard29 ( a , b and c ) and roa5701 ( d ) .
as can be seen from the figure , some of the features are difficult to distinguish from the noise . indeed , fig .
[ fig_fe_op](a ) appears to show the 4419 line in barnard29 , but numerous noise features of similar size to the 4419 line can also be seen , which may be due to observational effects or the reduction process .
unless these additional features are real ( which is unlikely ) , then the feiii line may also not be real and hence the abundance derived should be used with care .
in addition , the 4419 line appears to be narrower than other weak stellar absorption features in the spectrum , as may be seen from the profile fit in fig .
[ fig_fe_op](a ) , where the theoretical feature is wider than that observed .
however , 4419 gives a similar abundance to the apparently more reliable 5156 line , and so all three observed feiii transitions are used in determining the iron abundance for barnard29 .
also , these lines have the largest gf - values in their respective multiplets , and are among the strongest unblended feiii lines in the optical spectral region . for roa5701 , the sole feiii line observed at 5833 does not appear reliable ( see fig . [ fig_fe_op](d ) ) and hence the abundance is taken as an upper limit .
as the observed optical iron lines are of low quality , ultraviolet spectra were also employed to determine fe abundances for both barnard29 and roa5701 .
this is discussed below . to determine an iron abundance from the ghrs / hst spectrum ,
a goodness - of - fit method was used ( , dufton ( in preparation ) ) .
theoretical spectra were generated for the regions observed by the ghrs / hst , using non - lte tlusty model atmospheres and appropriate stellar atmospheric parameters .
the model grid for low metallicity regions was again used , and the underabundance of elements ( heavier than he , up to and including zn ) , was varied in each spectrum ( in steps of 0.1 dex from 0.0 to 2.5 dex ) relative to the solar estimates adopted in tlusty @xcite .
light elements ( c , n , o , mg , si ) were treated in non - lte while all other elements were in lte .
dufton ( in preparation ) found that @xmath680 per cent of absorption in the ultraviolet spectral region covered by our ghrs / hst data are due to fe , with more than 95 per cent from fe - group elements , therefore the calculations are taken as being in lte .
iron dominates the absorption , therefore any variation in the relative element abundances should not significantly affect the theoretical spectra . due to the numerous absorption features in the ultraviolet spectral region , continuum placement was difficult , as may be seen from the ghrs / hst observations of barnard29 and roa5701 in fig .
[ fig_all ] .
therefore , the continuum was not normalized , instead the flux calibrated spectrum was examined .
this was achieved by smoothing the theoretical spectra with a gaussian function ( to allow for instrumental broadening ) , followed by binning to the exact wavelength range and interval as the observed spectra , and then scaling to the same mean flux as that observed .
a goodness - of - fit was calculated by finding the sum of the squares of the difference between the observed and theoretical spectra for a range of metallicities .
a plot of this produced a parabolic shape , with the model with the best goodness - of - fit situated at the minimum of the parabola .
this method removes the need to normalize the observed spectrum , thus reducing errors associated with fitting the continuum . to test if different spectral intervals would affect the abundance estimate , smaller wavelength ranges
were selected and analysed .
no significant change was found in the goodness - of - fit and associated metallicity value .
results of the minimization procedure are shown in table [ tab_uv ] , along with iron abundances derived by @xcite and @xcite .
errors were calculated using a similar method as for the optical spectra .
systematic errors were obtained by varying the atmospheric parameters by their respective errors , while random errors were estimated by over - plotting theoretical spectra around the minimum and observing where the spectrum no longer matched the strongest observed features .
the total uncertainty was then calculated and is shown in table [ tab_uv ] .
the theoretical spectra that produced a minimum are shown , together with the observed spectra , in fig .
[ fig_all ] . for barnard29 , the goodness - of - fit indicates an [ fe / h ] = 2.1 .
in general , the theoretical spectrum matched many of the individual absorption features in the observational data , supporting the abundance extimates for this star .
roa5701 was less convincing in the fitting of the observed spectrum , with a minimum produced for [ fe / h ] = 2.4 dex .
in particular , the spectrum has fewer identifiable absorption features compared to barnard29 , making the estimation of the metallicity unreliable . in fig .
[ fig_all ] , there is an emission feature at @xmath61883 in the spectrum of barnard29 , which appears real and resembles part of a p cygni profile .
however , the apparent line does not have a strong associated absorption feature and is therefore thought to be an artefact of the reduction process .
if it is removed from the goodness - of - fit analysis , there is no change in the abundance obtained .
the feature was not observed in roa5701 .
@lcccccccccc star & this paper & @xcite & @xcite + + barnard29 & 2.1 @xmath14 0.1 & 2.2 & 2.2 + roa5701 & 2.4 @xmath14 0.3 & 2.7 & @xmath20 +
barnard29 and roa5701 have been the subject of numerous studies , several of which have included determinations of atmospheric parameters and chemical compositions .
we discuss these below in the context of the present results .
previous studies of barnard29 have produced similar atmospheric parameters to those found in this paper .
for example , @xcite used non - lte models with optical spectra to derive @xmath10 = 22500 @xmath14 1200 k and log _ g _ = 3.0 @xmath14 0.15 dex , while @xcite employed ultraviolet spectra from the _ iue _ satellite and ultraviolet colours , finding @xmath10 = 19000 @xmath14 1000 k and log _ g _ = 3.2 dex .
these spectra were again employed by @xcite , along with optical spectra of h@xmath23 , to determine lte atmospheric parameters of @xmath10 = 20250 k and log _ g _ = 3.15 dex .
@xcite used lte models to analyse _ iue _ low - resolution data and wht / ues optical data , resulting in @xmath10 = 20000 @xmath14 1000 k , log _ g _ = 3.0 @xmath14 0.1 dex and @xmath11 = 10 @xmath14 2 km s@xmath8 .
@xcite took a sample of the _ iue _ spectra used by @xcite , along with _
orfeus - spasii _ spectra , and with lte techniques obtained the same surface gravity as @xcite , but an increased temperature estimate of @xmath10 = 21000 @xmath14 1000 k. the previously derived effective temperatures and surface gravities are in good agreement with those found in this paper , within the errors , with the exception of the @xcite temperature , which may be due to using different non - lte models .
@xcite derived a larger microturbulence than in the present paper , presumably due to using lte models rather than our non - lte approach .
also @xcite only employed oii lines in their determination of @xmath11 , compared to our use of both the siiii multiplet and a variety of different multiplets of oii .
given the high quality of our observational data combined with the complexity of our non - lte models , we believe that our atmospheric parameters for barnard29 are probably the most reliable estimates to date .
roa5701 was studied recently by @xcite using the same non - lte models and methods as in the present paper , resulting in atmospheric parameters of @xmath10 = 25000 @xmath14 1000 k , log _ g _ = 3.2 @xmath14 0.2 dex and @xmath11 = 5 @xmath14 5 km s@xmath8 .
@xcite employed eso caspec optical spectra , along with the ghrs / hst data used here , and found @xmath10 = 2200024500 k , log _ g _
= 3.23.4 dex .
in addition , they derived @xmath11 = 23 or 20 km s@xmath8 , the former using feiii transitions and the latter those of oii .
their use of lte models , compared to the non - lte approach in the present paper , may be the reason for the differences in the derived microturbulence values .
the present microturbulence estimate is smaller than that of @xcite , probably due to improved equivalent width measurements and surface gravity estimate , although it is still within the uncertainty quoted .
other studies of roa5701 include @xcite , who analysed narrow - band and ubv photometry to obtain @xmath10 = 26600 k and log _ g _ = 3.5 dex , and @xcite who used _ iue _ spectra to derive @xmath10 = 24000 k. these studies are in agreement with the parameters derived here , with the exception of the @xmath10 estimate of @xcite .
@xcite suggested that the different values of ( b - v ) used by @xcite are responsible for their larger temperature estimate .
the results derived here are hence generally consistent with previous values , but once again are believed to be more reliable due to the improved observational data and non - lte models employed in our analysis .
table [ tab_ab ] shows the absolute abundances derived for barnard29 and roa5701 , along with those previously determined by and @xcite for barnard29 , and @xcite and @xcite for roa5701 .
however , absolute abundances may contain systematic errors , due for example to inaccuracies in the atmospheric parameters and atomic data , in particular the f - values . in order to minimize these errors , a detailed line - by - line differential abundance analysis
can be performed , with respect to a young b - type star of known chemical composition ( populationi ) which has similar atmospheric parameters .
therefore , in table [ tab_diff ] the abundances of barnard29 and roa5701 are compared to those of young b - type stars in the galactic cluster ngc6611 @xcite and in the local field . for the former
, results have been determined using the same non - lte model atmosphere calculations as in the present paper .
the latter uses different model atmosphere calculations and hence differential abundances using this source may not be as reliable .
the standard error in the differential abundance estimates for barnard29 and roa5701 , as compared with young b - type stars , are typically @xmath140.2 dex .
these errors do not include any systematic errors due to the atmospheric parameters as these are assumed to cancel due to the use of the same methods for both samples .
note that where we did not have access to results for individual absorption lines , differential abundance values are based on the mean abundance estimates listed , with a minimum error of @xmath140.2 dex .
it is not possible to perform a differential analysis for feiii ultraviolet lines using young b - type stars , as these are affected by line crowding and are often saturated .
the strengths of the feiii absorption lines are only weakly dependent on the effective temperature and surface gravity , so they should not be a significant source of error . in order to test the dependence on microturbulence for young b - type stars , the galactic star hr1886 was examined using stis spectra taken from the hst archive .
this star was chosen as its @xmath10 = 24000 k and log _ g _ = 4.1 dex @xcite are similar to those of roa5701 . for a microturbulence of 0 km
s@xmath8 , synthetic spectra were generated and scaled to the same mean flux as the observed spectrum , using the methods described in section [ sec_uv2 ] .
the microturbulence was then increased to 10 km s@xmath8 , and the process repeated .
the change in microturbulence resulted in a change of metallicity of @xmath61.4 dex , implying that feiii ultraviolet lines in galactic b - type stars are indeed very sensitive to changes in microturbulence , and are not appropriate for a differential abundance analysis .
@xcite found a similar dependence on microturbulence for the standard star hd149438 ( @xmath24scorpii ) , with a change of @xmath11 from 0 to 10 km s@xmath8 yielding a metallicity change of @xmath60.8 dex .
for barnard29 and roa5701 , a @xmath11 change from 0 to 10 km s@xmath8 resulted in a metallicity change of @xmath60.4 and 0.2 dex , respectively .
@l cccccccccccccccc species & & + & & & barnard29 & roa5701 + & this paper&con94 & dix98 & moe98 & this paper&moe98 & tho06 & + + c & @xmath161.3 & @xmath161.3 & 1.8 & @xmath20 & @xmath161.4 & @xmath162.1 & @xmath161.4 & @xmath161.7 & @xmath161.8 + n & 0.3 & 0.3 & @xmath20 & @xmath20 & 0.6 & 0.7 & 0.6 & 0.5 & 0.8 + o & 1.1 & 1.0 & @xmath20 & @xmath20 & 0.8 & 0.6 & 0.8 & 1.2 & 0.9 + mg & 1.2 & 1.3 & @xmath20 & @xmath20 & 1.2 & @xmath20 & 1.1 & 1.3 & 1.3 + si & 1.0 & 1.1 & @xmath20 & @xmath20 & 1.2 & 1.3 & 1.5 & 0.8 & 1.1 + al@xmath17 & 1.1 & 1.0 & @xmath20 & @xmath20 & @xmath161.4 & @xmath20 & @xmath161.0 & 1.2 & @xmath161.5 + s @xmath17 & 1.5 & @xmath161.0 & 1.9 & @xmath20 & 1.6 & @xmath20 & 1.5 & 1.4 & 1.5 + fe ( optical ) & 1.4 & @xmath160.8 & @xmath20 & @xmath20 & @xmath161.4 & @xmath20 & @xmath20 & 1.4 & @xmath161.4 + fe ( uv ) & 2.1 & @xmath20 & 2.2 & 2.2 & 2.4 & 2.7 & @xmath20 & 2.1 & 2.1 + cluster [ fe / h ] & & + note : con94 = @xcite , dix98 = @xcite , moe98 = @xcite , tho06 = @xcite . differential abundances for barnard29 , relative to young b - type stars , are given in table [ tab_diff ] , along with the results of and @xcite .
as stated previously , @xcite used _
iue _ and wht / ues data with lte models in their analysis .
their upper limit for the c abundance agrees with that found in this paper , while n , o , mg , si and al are @xmath60.1 dex lower . reasons for the differences in abundance could include different atmospheric parameters , improved equivalent width measurements ( as a consequence of our higher s / n observations ) , and our use of non - lte model atmosphere codes . to investigate if our non - lte treatment may be responsible for the differences , we have calculated lte abundances for the elements , with the exception of al which is already in lte .
this results in abundances of @xmath165.93 for c , 7.40 for n , 7.75 for o , 5.93 for mg , 6.78 dex for si .
the n , o and si abundances are still larger than the @xcite abundances , while c and mg are lower .
the lower microturbulence used in the present analysis may be a further source of the differences . adopting a microturbulence of 10 km s@xmath8 from @xcite , lte abundances of 7.30 for n ,
7.62 for o and 6.47 dex for si are obtained , while the other elements abundances do not significantly change .
the values are now in good agreement with @xcite , with the exception of c , suggesting that the microturbulence is a factor .
the discrepancy for c may be due to the use of different lines .
however , the results in the present paper should be more reliable due to our use of non - lte techniques and improved observational data .
@xcite used a far - ultraviolet spectrum from _ orfeus - spasii _ , fitting lte synthetic models to the data , similar to the method used here .
they determined abundances for the elements he , n , o , al and s , finding results consistent with @xcite but with larger error bars , suggesting the use of lower resolution data as the reason for this .
the c abundance of @xcite was found using two ciii lines , and is 0.5 dex lower than the upper limit found here .
their s abundance was derived from lines of siii and siv , with some scatter due to blending with nearby features .
these authors also observed sii lines , but these were not included in the analysis as they appeared to be affected by non - lte .
the s abundance of dixon & hurwitz is @xmath60.4 dex lower than our value , which may be due to the use of different spectral wavelength regions .
an iron abundance of [ fe / h ] = 2.2 was determined by both @xcite and @xcite .
@xcite found it difficult to determine an fe abundance , as the spectrum was contaminated by interstellar lines and many of the expected fe transitions were in regions of poorly - fitted continuum .
they fitted a band of fe lines rather than individual features , similar to the method used in this paper for the uv observations , noting that their abundance was more than 0.5 dex below the model atmosphere used to derive the synthetic spectra ( i.e. [ m / h ] = 1.5 ) , with tests showing that the model choice had little affect on the abundance .
@xcite determined an fe abundance using the same ghrs / hst spectrum employed here .
they adopted the atmospheric parameters of @xcite , and fitted a global continuum to the normalized data , noting that an uncertainty of 5 per cent in the continuum definition of their spectrum would lead to an error of 0.2 dex in the derived abundance . in the present paper , an feiii abundance of [ fe / h ] = 2.1
is obtained from the uv spectrum and [ fe / h ] = 1.4 from the optical .
the uv value is well below the metallicity of the cluster ( [ fe / h ] = 1.6 , @xcite ) , but is in slightly better agreement compared to the previous uv studies .
however , our optical result is consistent with the cluster metallicity , to within the errors .
this discrepancy is discussed further in section [ sec_evo ] . for roa5701 ,
there is good agreement between the present abundances and those of @xcite , with c , n and o giving the same differential values .
the si abundance estimates differ by 0.2 dex , and this discrepancy appears to be due to differences in the atmospheric parameters .
this was tested by calculating non - lte abundances using the equivalent widths from this paper and the atmospheric parameters from @xcite .
the resulting values are 5.93 dex for siiii and 5.88 dex for siiv , suggesting the changed atmospheric parameters are a cause for the differences .
the upper limit to the c abundance of @xcite is lower than that derived in this paper , despite using the same upper limit value for the equivalent width .
this difference could be due to large non - lte effects in cii , as proposed by @xcite and hence , the upper limit derived here may be more reliable .
although the n and o abundances of @xcite differ by @xmath60.2 dex , and the si abundance by @xmath60.1 dex , from those presented here , again this may be due to non - lte effects . to test this ,
lte abundances have been derived , using the tlusty grid , resulting in @xmath166.23 for c , 7.99 for o , 7.10 dex for n and 6.50 dex for si .
the o abundance is in very good agreement with @xcite , but the c , n and si values are still higher .
the changes could be due to the use of different equivalent width measurements and atmospheric parameters .
however , the present results should be more reliable due to our non - lte analysis .
no abundances for mg , al or s were derived by @xcite , while @xcite determined values for mg and s , which differ from those found in this paper by @xmath60.1 dex .
this is probably due to the improved s / n observations , which allow better equivalent width measurements ; changing the atmospheric parameters had little affect on these abundances .
the listed al abundance is still an upper limit , but due to the improved s / n of the spectrum it was possible to place a revised lower limit on the equivalent width . in general , the abundances quoted in the present paper should be more accurate due to the improved quality of the observational data .
@xcite observed two unidentified features at 4512.41 and 4352.22 in their data , which were obtained using ucles at the aat .
however , no features are observed at these wavelengths in the feros spectrum , which is of higher resolution and s / n , indicating that the lines in the ucles data may not be real .
@xcite obtained a differential feiii abundance of [ fe / h ] = 2.7 from the ghrs / hst ultraviolet spectrum , compared to [ fe / h ] = 2.4 derived here .
both values are lower than the metallicity of @xmath0cen , namely [ fe / h ] = 1.6 ( @xcite ) , although our result is in better agreement with the metallicity of the cluster .
an optical value [ fe / h ] @xmath16 1.4 is obtained in the present paper .
the [ fe / h ] abundance of roa5701 is discussed further in section [ sec_evo ] .
@l cccccccccccc species & & + & barnard29 & giants&roa5701&giants + + c & @xmath166.64 & @xmath20 & @xmath166.54 & 6.86 + n & 7.45 & @xmath20 & 7.06 & @xmath20 + o & 7.50 & 7.53 & 7.76 & 7.45 + mg & 6.10 & 6.14 & 6.15 & 6.26 + si & 6.45 & 6.38 & 6.17 & 6.26 + al & 5.14 & 5.75@xmath19 & @xmath164.87 & 4.87@xmath21 + s & 5.72 & @xmath20 & 5.62 & @xmath20 + fe & 6.07 & 5.90 & @xmath166.02 & 5.92 + the absolute abundances of barnard29 and roa5701 are given in table [ tab_giant ] , along with those of giants from their respective clusters , namely m13 and @xmath0cen .
in general the abundances compare well , to within the errors .
a notable exception is the al abundance of barnard29 , which is lower than that of the giants .
however , @xcite and @xcite have both shown that the aluminium abundances in the giants range from [ al / fe ] @xmath6 0 1.2 dex . the giants with [ al / fe ] @xmath6 0.3 have similar o abundances to those found here , suggesting that the barnard29 abundance compares well with a selection of the giant population .
@xcite previously studied roa5701 , suggesting that the observed abundance patterns implied that the star had left the agb prior to the onset of the third dredge - up , as have @xcite to similarly explain the abundance patterns in barnard29 .
dredge - up occurs in stars during their agb evolution , as products of hydrogen and helium burning are moved from the core to the surface , altering the surface composition @xcite .
briefly , during the first dredge - up , the surface o sees little change while c decreases as n increases .
the second dredge - up sees c and o mostly converted into n. at the third dredge - up , triple@xmath25 products are brought to the surface and c is enhanced relative to o and n. both barnard29 and roa5701 exhibit abundance patterns compatible with no third dredge - up .
the n abundances are strongly enhanced relative to both c and o , especially for barnard29 , while o is larger than c. this is consistent with the previous studies of both stars , with the exception of the lower n abundance of @xcite for roa5701 .
thus the abundances of cno appear compatible with the suggestion that the third dredge - up may not have occurred .
the low c abundance could be explained in terms of hot bottom burning .
@xcite found v435oph to have a low metallicity and no enhancement of c , suggesting hot bottom burning as an explanation of the observed abundances .
in hot bottom burning ( hbb ; @xcite ) , c is converted to n at the base of the convective envelope , when temperatures become large enough for cno processing .
@xcite indicates that hbb may have an important role at extremely low metallicities .
however , @xcite suggest that population i stars of initial mass m @xmath26 5 m@xmath2 , and population ii stars of m @xmath26 3.5 m@xmath2 experience hbb .
as the initial mass of the stars is unknown , no conclusions can be made regarding this process .
although the lack of a third dredge - up appears a likely explanation for the abundance patterns of the light elements in both stars , @xcite and @xcite explain the abundances of the heavier elements in terms of the gas - dust separation process .
the basic premise involves grain formation that occurs in the circumstellar shells of stars during the agb evolution .
dust particles are removed by radiation pressure and the ` cleaned ' gas is re - accreted onto the surface .
this results in abundances similar to the ism , i.e. cno and s near solar , while other elements such as mg , al , si and ca follow the low fe abundance @xcite and see for example the rvtauri stars uycma , hp lyr and bz sct @xcite .
@xcite found that , in the atmospheres of post - agb stars , metals with a high condensation temperature , e.g. fe , are more likely to condense into dust grains , compared to elements such as c , n , o and s with low condensation temperatures which stay in the gas phase with no significant depletions observed .
@xcite suggested that this process occurred as they found a low fe value , with n and o virtually undepleted , while @xcite also cite the low fe abundance as evidence for the process .
for this scenario , similar depletions of mg , si and fe would therefore be expected .
examining the differential ism abundances ( table [ tab_diff ] ) it is possible to determine if the stars follow the abundance trends seen in gas - dust separation .
for barnard29 , we find differential abundances of [ mg / h ] = 1.3 , [ al / h ] = 1.2 , [ si / h ] = 0.8 , [ s / h ] = 1.4 and [ fe / h](optical ) = 1.4 ( relative to the ism ) .
in addition , for roa5701 the differential abundances are [ mg / h ] = 1.3 , [ si / h ] = 1.1 and [ s / h ] = 1.5 .
the mg , al and fe ( optical ) abundances of barnard29 agree with the value for s and the metallicity and giant abundances of m13 ( [ fe / h ] = 1.6 , @xcite ) , although the si abundance estimate is somewhat larger , possibly due to its sensitivity to the atmospheric parameters . for roa5701 ,
the s abundance agrees ( to within the errors ) with the mg and si values and is consistent with the metallicity and giants of @xmath0cen ( [ fe / h ] = 1.6 , @xcite ) .
similar results would be obtained if the abundances were compared with young b - type stars instead of the ism . for gas - dust separation , cno and s
would be expected to be relatively undepleted and not in agreement with mg , si , al or fe .
however , we find that the s abundance is generally in agreement with the mg , si , al and fe(optical ) abundances , suggesting that gas - dust separation has not occurred . also this process is generally confined to binary systems .
the possible binarity of our two stars was investigated using radial velocity measurements .
however a literature search produced sparse results , indicating no conclusions can be made .
additionally , the position of the stars on log _ g_/@xmath10 or logl/@xmath10 diagrams suggest core masses for both stars of @xmath6 0.55 m@xmath2 , using log(l / l@xmath2 ) = 3.2 @xcite for roa5701 and log(l / l@xmath2 ) = 3.23 @xcite for barnard29 .
@xcite found that stars of this average mass do not show an infrared excess , which is supported by the lack of iras detection for both stars , and no emission features observed in h@xmath25 for roa5701 .
these results suggest there is no evidence of a circumstellar dust shell present for either star .
hence the gas - dust separation process appears unlikely to be responsible for the observed abundance patterns . in this paper
, we find a difference of 0.7 dex and @xmath161.0 dex for barnard29 and roa5701 respectively , between the optical and ultraviolet derived iron abundances .
analyses of the feiii lines in other hot post - agb stars , such as bd+33@xmath272642 @xcite and hd177566 , show similar fe abundance trends to those found here .
@xcite analysed optical and _ iue _ spectra , and their results led them to suggest the occurrence of gas - dust separation , as he was near solar , c , n , o , mg and si depleted by @xmath61.0 dex , and fe depleted by @xmath62.0 .
@xcite set upper limits for the fe abundance from feiii lines in both optical and ultraviolet spectra , finding a difference of @xmath60.5 dex between the two sets of data .
similarly , @xcite derived an fe abundance of [ fe / h ] = 6.95 dex from the feiii lines in the _ iue _ spectra of the main - sequence b - type star @xmath23peg , compared to [ fe / h ] = 7.56 dex from optical feiii transitions .
in addition , @xcite in their analysis of @xmath28her found that the ultraviolet fe abundance was @xmath260.5 dex below the near solar value derived from optical lines . @xcite and @xcite studied the b - type supergiant hd51309 . while the former state two different iron abundance estimates depending on the temperature structure used , there is a difference of @xmath260.75 dex found between their optical results ( from feii and feiii lines ) and the _ iue _ iron abundance of @xcite .
dufton ( in preparation ) have observed similar abundance discrepancies for young b - type stars in the small and large magellanic clouds ( smc , lmc ) , as well as in the magellanic bridge ( mb ) .
they determined fe abundances from ghrs / hst and stis / hst spectra , using the same methods and feiii transitions as described here , and found that these were consistently lower ( more than 0.5 dex ) than the estimated metallicities of the smc , lmc and mb regions .
it is possible to derive the abundance of other elements using ultraviolet spectra , but that is not possible here as the ghrs data employed do not cover a large enough spectral range .
spectra have been used by a number of authors to derive the abundances of several light elements .
@xcite obtained a c abundance using both ultraviolet and optical spectra , finding the abundances to be compatible .
@xcite found similar differential abundances for c , n and si between their optical and _ iue _ spectra , with differences being less than 0.15 dex . also , the mg abundance obtained by @xcite showed a difference of 0.05 dex between their optical and ultraviolet values .
however , the ultraviolet abundances obtained by @xcite for c and s are lower by 0.5 and 0.4 dex , respectively , than the optical abundances determined here ( see section [ sec_b29chem ] ) . despite this result
, the use of ultraviolet spectra to determine the abundances of light elements generally appears to be reliable and do not suffer the same problem observed with iron .
the above studies , in conjunction with the present work , strongly indicate that the ultraviolet feiii lines yield systematically low fe abundances , possibly as a result of errors in the adopted atomic data , in particular f - values .
currently , atomic data from kurucz models ( http://nova.astro.umd.edu , ) are used in the tlusty grids .
we have therefore performed a literature search of f - values for feiii , to investigate if more recent calculations or measurements would produce different fe abundances .
table [ tab_os ] lists the six strongest transitions observed in the ghrs / hst spectra , along with the oscillator strengths from kurucz , and the more recent results of @xcite and @xcite .
there are little differences between the sets of atomic data , so these will not solve the fe abundance problem
. however we note that further calculations and measurements for these feiii transitions are planned in the future . despite numerous studies of the ultraviolet feiii transitions ,
the line list for the spectral region employed in this paper may not be complete .
however , it is assumed that , if this is the case , any lines excluded during the fitting procedure would be compensated by making other transitions too strong .
this would potentially lead to an over estimate in the derived abundance , which would not solve the abundance discrepancy observed .
cccccccccccccc transition@xmath19 & g@xmath29@xmath19 & g@xmath30@xmath19 & wavelength & + & & & ( ) & kurucz & @xcite & @xcite + + a@xmath31g@xmath32z@xmath31f@xmath33 & 5 & 5 & 1869.841 & -0.666 & -0.672 & @xmath20 + a@xmath31g@xmath32z@xmath31f@xmath33 & 5 & 3 & 1871.152 & -0.121 & -0.116 & @xmath20 + b@xmath31d@xmath32y@xmath31d@xmath33 & 9 & 9 & 1877.990 & 0.238 & 0.273 & @xmath20 + a@xmath31g@xmath32z@xmath31f@xmath33 & 13 & 11 & 1890.664 & 0.434 & 0.407 & @xmath20 + a@xmath34s@xmath32z@xmath34p@xmath33 & 7 & 9 & 1895.456 & 0.461 & 0.415 & 0.436 + a@xmath35i@xmath32y@xmath35h@xmath33 & 13 & 11 & 1896.814 & 0.480 & 0.398 & @xmath20 + to investigate the fe abundance discrepancy further will require very high s / n data for barnard29 and roa5701 , sufficient to ascertain the reliability of the observed optical feiii lines and observe further feiii lines in both stars . unlike the ultraviolet transitions , the optical lines are sufficiently weak , even in main - sequence populationi b - type stars , for a detailed line - by - line differential abundance analysis to be performed .
this will hence allow the removal of systematic effects such as the dependence on f - value , and hence the determination of accurate fe abundances , for comparison with the ultraviolet measurements .
we plan such observations in the near future .
the abundance patterns observed for both barnard29 and roa5701 indicate that the stars have not undergone the third dredge - up .
in addition , the gas - dust separation process suggested by previous authors appears unlikely .
the optical spectra imply abundances for the fe and other elements , which are compatible with those found in the late - type giants , while the feiii lines in the ultraviolet produce lower abundances than expected , as found in several other studies .
this strongly suggests that the ultraviolet feiii do not provide reliable abundance indicators .
future studies require high s / n optical data , in order to obtain accurate fe abundance estimates . a re - evaluation of the feiii f - values ,
either theoretically or experimentally , would also be highly desirable .
some of the data presented in this paper were obtained from the multimission archive at the space telescope science institute ( mast ) .
stsci is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas5 - 26555 .
support for mast for non - hst data is provided by the nasa office of space science via grant nag5 - 7584 and by other grants and contracts .
iraf is distributed by the national optical astronomy observatories , u.s.a .
we would like to thank ian howarth for his continued help .
hmat acknowledges financial support from the northern ireland department of education and learning ( del ) .
fpk is grateful to awe aldermaston for the award of a william penney fellowship .
jvs thanks the staff of the mcdonald observatory for their expert assistance in taking these observations and the particle physics and astronomy council for financial support .
dll thanks the robert a. welch foundation of houston , texas for their support .
aaz thanks the saao for hospitality during a sabbatical visit .
landsman w. , moehler s. , napiwotzki r. , heber u. , sweigart a. , catelan m. , stecher t. , 2000 , in leige , noels a. , magain p. , caro d. , jehin e. , parmentier g. , thoul a. a. , eds , 35th leige int .
colloq . , the galactic halo : from globular cluster to field stars .
et de geop . ,
belgium , p. 515 | high resolution optical and ultraviolet spectra of two b - type post - asymptotic giant branch ( post - agb ) stars in globular clusters , barnard29 in m13 and roa5701 in @xmath0cen , have been analysed using model atmosphere techniques .
the optical spectra have been obtained with feros on the eso 2.2-m telescope and the 2d - coud spectrograph on the 2.7-m mcdonald telescope , while the ultraviolet observations are from the ghrs on the hst .
abundances of light elements ( c , n , o , mg , al and s ) plus fe have been determined from the optical spectra , while the ultraviolet data provide additional fe abundance estimates from feiii absorption lines in the 18751900 wavelength region .
a general metal underabundance relative to young b - type stars is found for both barnard29 and roa5701 .
these results are consistent with the metallicities of the respective clusters , as well as with previous studies of the objects .
the derived abundance patterns suggest that the stars have not undergone a gas - dust separation , contrary to previous suggestions , although they may have evolved from the agb before the onset of the third dredge - up
. however , the fe abundances derived from the hst spectra are lower than those expected from the metallicities of the respective clusters , by 0.5 dex for barnard29 and 0.8 dex for roa5701 .
a similar systematic underabundance is also found for other b - type stars in environments of known metallicity , such as the magellanic clouds .
these results indicate that the feiii ultraviolet lines may yield abundance values which are systematically too low by typically 0.6 dex and hence such estimates should be treated with caution .
[ firstpage ] stars : abundances stars : agb and post - agb stars : early - type stars : individual : ngc6205222 stars : individual : ngc51395701 ultraviolet : stars |
let @xmath1 denote either the closed genus @xmath2 surface @xmath3 or that surface @xmath4 with a marked point .
let @xmath5 be the mapping class group of @xmath1 .
harvey s @xcite complex of curves @xmath6 ( see definition [ def : cc ] below ) plays a major role in the study of @xmath5 in part because it serves as an analog of the rational tits building for an arithmetic group in a semisimple @xmath0-group .
( see borel - serre s celebrated paper @xcite and the fundamental work of harer on homological aspects of the mapping class group , surveyed in @xcite , for background . ) inspiration for the current study is derived from work by ash - rudolph @xcite and others ( _ cf .
_ @xcite , @xcite , @xcite , @xcite , @xcite ) concerning the dualizing modules of these arithmetic groups .
harer ( * ? ? ?
* theorem 4.1 ) and ivanov @xcite ( in the case of the closed surface ) have shown that the mapping class group @xmath5 is a virtual duality group in the sense of bieri - eckmann @xcite ( see also @xcite , @xcite , @xcite ) , and that the dualizing module for any torsion - free , finite index subgroup of @xmath5 is the reduced homology of the curve complex .
the curve complex has the homotopy type of a wedge of infinitely many @xmath7-spheres ( see ( * ? ? ?
* theorem 3.5 ) and ( * ? ? ? * theorem 1.4 ) ) so its reduced homology is concentrated in dimension @xmath8 .
the left action of the mapping class group on the complex of curves makes the reduced homology of @xmath6 into a @xmath5-module .
following harer @xcite , we will call this @xmath5-module the _ steinberg module , @xmath9 _ , of the mapping class group ( see definition [ def : st_def ] below ) .
the purpose of this paper is to initiate an investigation of the ( left ) @xmath5-module structure of @xmath9 .
for all groups @xmath10 , we will consider modules ( resp .
ideals ) over the group ring @xmath11 to be left @xmath10-modules ( resp . left ideals ) .
our results begin in [ sec : st_pres ] with a resolution of @xmath12 ( see definition [ def : st_def ] below ) . * proposition [ prop : stein_res ] . * _ let @xmath13 be the chain complex defined in equation ( [ eq : chain_def ] ) .
the exact sequence @xmath14 is a finite @xmath15-module resolution for the steinberg module @xmath16 . _ from this resolution we have a @xmath15-module isomorphism @xmath17 which yields the presentation in proposition [ prop : st_pres ] .
we use this presentation to prove our main result ( see [ sec : notation]-[sec : cyclic ] below for an explanation of the notation )
. * theorem [ thm : single_gen ] .
* _ let @xmath18 , and let @xmath19 be the @xmath20-filling system in figure [ fig : st_gen ] .
let @xmath21 $ ] be the class of @xmath22 in @xmath23 .
then @xmath16 is generated as a @xmath15-module by @xmath21 $ ] .
for the closed surface let @xmath21 $ ] be the class of @xmath22 in @xmath24 ( see the proof of proposition [ prop : st_pres ] ) .
similarly , @xmath25 is generated as a @xmath26-module by @xmath21 $ ] .
_ theorem [ thm : single_gen ] may be compared to a result of ash - rudolph ( * ? ? ?
* theorem 4.1 ) which implies that the reduced homology of the tits building for @xmath27 is a cyclic @xmath28-module .
theorem [ thm : single_gen ] states that the steinberg module for the mapping class group is a cyclic @xmath5-module and gives a specific single generator .
it follows immediately that this generator must be nontrivial in @xmath29 .
* corollary [ cor : notriv ] . * _ @xmath21 \in \st(\surface_g^1)$ ] ( resp .
@xmath21 \in \st(\surface_g)$ ] ) is nontrivial for @xmath18 .
_ finally in [ sec : g1andg2 ] we consider two examples : the steinberg modules for the mapping class groups of the surfaces of genus @xmath30 and @xmath31 . for these low genus surfaces ,
propositions [ prop : g1_pres ] and [ prop : g2_pres ] convert the multi - generator presentation given in proposition [ prop : st_pres ] to a presentation based on the single generator from theorem [ thm : single_gen ] .
the importance of the reduced homology of the curve complex ( which we will denote here by @xmath32 ) for the homological structure of the mapping class group comes from fact that it is the dualizing module for every torsion - free , finite index subgroup @xmath33 .
this dualizing module connects the homology and cohomology of @xmath10 via the formula @xmath34 where @xmath35 ( resp .
@xmath36 ) is the virtual cohomological dimension of the mapping class group @xmath26 ( resp .
@xmath15 ) , and @xmath10 acts on @xmath37 with the diagonal action : @xmath38 for all @xmath39 , @xmath40 , and @xmath41 .
we briefly overview the reason that the reduced homology of the curve complex is the dualizing module for any torsion - free , finite index subgroup of @xmath5 .
the mapping class group acts properly discontinuously on teichmller space @xmath42 , which is diffeomorphic to @xmath43 , where @xmath44 is the number of marked points on @xmath1 .
for any torsion - free , finite index subgroup @xmath33 the quotient of @xmath42 by the properly discontinuous , fixed - point - free action of @xmath10 is a manifold .
this manifold is not compact but can be compactified either by adding certain `` points at infinity '' following ivanov @xcite or by removing a certain open neighborhood of infinity following harer @xcite . in either case , the universal cover @xmath45 of this compactified manifold is called a _ bordification _ of teichmller space . one
must then show that @xmath45 is contractible and that @xmath46 has the homotopy type of a wedge of spheres of dimension @xmath8 .
it then follows @xcite that the cohomological dimension of @xmath10 is @xmath47 and the dualizing module of @xmath10 is @xmath48 .
harer establishes these results for surfaces with boundary or punctures in @xcite .
in addition , he shows ( * ? ? ? * lemma 3.2 ) that @xmath46 is @xmath5-equivariantly homotopy equivalent to the curve complex .
ivanov @xcite establishes the same for closed surfaces . throughout this paper
we will maintain the viewpoint that the homology of the curve complex should be viewed as an analog for the mapping class group of the homology of the rational tits building for @xmath28 .
briefly , the rational tits building @xmath49 is the simplicial complex of flags of nontrivial proper subspaces of @xmath50 ( see ( * ? ? ?
* example 1b ) ) .
the action of @xmath27 on @xmath50 induces a simplicial action of @xmath27 on @xmath49 . for any basis @xmath51 for @xmath50 the union of all simplices of @xmath49 whose vertices are subspaces spanned by nonempty proper subsets of @xmath51
gives an _ apartment _ of @xmath49 .
apartments have the homeomorphism type of an @xmath52-sphere . by the solomon - tits theorem ( _ cf .
_ @xcite , ( * ? ? ? * theorem 2 ) ) the tits building @xmath49 has the homotopy type of wedge of infinitely many @xmath52-spheres , and @xmath53 is spanned by the homology classes of all apartments . the action of @xmath27 is transitive on apartments of @xmath49 so one sees immediately that @xmath53 is a cyclic @xmath27-module
. the steinberg module for @xmath28 is defined ( * ? ? ?
437 ) to be the infinitely generated free abelian group @xmath54 borel and serre ( * ? ? ?
* theorem 11.4.2 ) show that the dualizing module of any torsion - free finite index subgroup of @xmath28 is @xmath55 providing inspiration for harer s later work @xcite on the mapping class group .
the group @xmath28 is no longer transitive on apartments of @xmath49 ; nevertheless , ash and rudolph @xcite are able to give a reduction argument to show that the homology class of the sphere for any apartment can be written as a linear combination of classes of `` integral unimodular '' apartments where an integral unimodular apartment is one coming from an integral basis @xmath51 for @xmath56 .
of course @xmath28 _ is _ transitive on integral unimodular apartments .
it follows that @xmath57 is a cyclic @xmath28-module .
one should view theorem [ thm : single_gen ] below as an analog for the mapping class group of ash and rudolph s result .
* acknowledgements .
* i owe a huge debt to juan souto who pointed out that the best way to find the spheres in the curve complex is to look for balls they bound in teichmller space .
i am also indebted to karen vogtmann for elucidating the work of harer .
benson farb provided immeasurable support and feedback for this project .
i would also like to thank mladen bestvina , matthew day , justin malestein , and robert penner for some very helpful discussions .
let @xmath3 be the surface of genus @xmath2 , and let @xmath4 be the surface of genus @xmath2 with @xmath30 marked point @xmath58 . the _ mapping class group _ @xmath26 ( resp .
@xmath15 ) is the group of orientation - preserving self diffeomorphisms of @xmath3 ( resp .
orientation - preserving self diffeomorphisms of @xmath4 fixing the marked point ) modulo diffeomorphisms isotopic to the identity ( see @xcite for a survey ) . an _ ( essential ) curve _ in @xmath3 is an isotopy class of the image of an embedding of the circle @xmath59 in @xmath3 not bounding a disk in @xmath3
( essential ) curve _ in @xmath4 is an isotopy class of the image of an embedding of @xmath59 in @xmath60 not bounding a disk in @xmath4 .
( the disk may contain the marked point . )
note that for the surface with a marked point , curves may not be isotoped past the marked point . in either surface , a _ curve system _
is a set of curves ( with isotopy class representatives ) which can be made to be disjoint .
since curves are isotopy classes , a curve system can not have parallel curves .
we can partially order curve systems by inclusion .
[ def : cc ] for @xmath18 the _ curve complex _ @xmath61 ( resp .
@xmath62 ) for the surface @xmath3 ( resp .
@xmath4 ) is the simplicial complex with @xmath63-simplices corresponding to curve systems with @xmath64 curves and face relation given by inclusion .
both @xmath61 and @xmath62 have the homotopy type of an infinite wedge of spheres of dimension @xmath8 ( see ( * ? ? ?
* theorem 3.5 ) and ( * ? ? ? * theorem 1.4 ) ) .
as stated above , the reduced homology of the curve complex ( which is concentrated in dimension @xmath8 ) is fundamental to the structure of the mapping class group as a virtual duality group with virtual cohomological dimension @xmath65 for the closed surface and dimension @xmath66 for the surface with one marked point ( * ? ? ?
* theorem 4.1 ) .
[ def : st_def ] for @xmath18 the _ steinberg module _ for the mapping class group @xmath15 is the @xmath15-module @xmath67 and the _ steinberg module _ for the mapping class group @xmath26 is the @xmath26-module @xmath68 our aim is to investigate the module structure of the steinberg modules @xmath16 and @xmath25 .
a result of harer cuts our work in half .
the mapping class group @xmath15 acts on the curve complex @xmath61 by forgetting the marked point .
hence @xmath25 is naturally a @xmath15-module .
harer ( * ? ? ?
* lemma 3.6 ) has shown that forgetting the marked point gives a @xmath15-equivariant homotopy equivalence @xmath69 .
hence we have the following lemma .
[ lem : same_module ] as @xmath15-modules @xmath70 .
this in turn allows an immediate conclusion about the @xmath15-module structure of @xmath16 which is not at all apparent from direct observation of the action of @xmath15 on @xmath62 .
[ cor : factors ] the action of @xmath15 on @xmath16 factors through its quotient @xmath26 . in light of lemma
[ lem : same_module ] we will focus exclusively on the @xmath15-module structure of @xmath16 . instead of calculating the homology of the curve complex @xmath62 directly , we will work in the _ arc complex _ for @xmath4 ( see definition [ def : ac ] below ) .
lemma [ lem : same_module ] is especially fortunate since no structure analogous to the arc complex is readily available for the surface @xmath3 .
an _ ( essential ) arc _ in @xmath4 is an isotopy class of the image of an embedded loop based at the marked point @xmath58 which does not bound a disk in @xmath4 .
an _ arc system _ is a set of arcs ( with isotopy class representatives ) which intersect only at the marked point .
arc systems may not have parallel arcs .
henceforth we will make no distinction between an arc system and a set of representatives of each arc intersecting only at the marked point .
[ def : ac ] the _ arc complex _
@xmath71 is the simplicial complex with @xmath63-simplices corresponding to arc systems with @xmath64 arcs and face relation given by inclusion .
an arc system @xmath72 _ fills _
@xmath4 if the connected components of @xmath73 are all disks .
the minimum number of arcs needed to fill @xmath4 is @xmath74 . in what follows
we will want to keep careful track of the number of arcs in a filling system , so we will say that a filling arc system _ @xmath75-fills _ @xmath4 if the arc system has @xmath76 arcs .
notice that a @xmath75-filling system cuts the surface into @xmath77 disks .
the _ arc complex at infinity _
@xmath78 is the simplicial subcomplex of @xmath79 which is the union of the simplices of @xmath79 whose vertex sets do not fill @xmath4 .
any arc system with fewer than @xmath74 arcs can not fill @xmath4 so @xmath80 contains the entire @xmath7-skeleton of @xmath79 .
the name `` arc complex at infinity '' calls for some explanation . a very nice discussion is given in @xcite .
the idea is this . using a fundamental construction of jenkins and strebel @xcite it is possible to associate to each point in teichmller space @xmath81 ( the space of finite area marked complete hyperbolic metrics on the surface @xmath60 ) an embedded metric graph in @xmath60 .
one can then homeomorphicly identify @xmath81 with @xmath82 using the arc system in @xmath4 dual to this graph ( see @xcite for details ) .
thus one may think of @xmath80 as extra `` points at infinity '' attached to teichmller space .
harer ( * ? ? ?
* theorem 3.4 ) defined a continuous map @xmath83 and showed that it is a homotopy equivalence ( see [ sec : curv_sphere ] below for more on this map ) .
hence , from definition [ def : st_def ] we get another characterization of the steinberg module @xmath84
in this section we will give a resolution of the steinberg module @xmath16 as a @xmath15-module .
the approach here is analogous to ash s simplification @xcite of lee and szczarba s resolution of the steinberg module for @xmath28 given in @xcite ( see @xcite for a nice summary ) .
we will need the following two results of harer .
[ thm : har_cont ] the arc complex @xmath85 is contractible .
[ thm : har_spheres ] the arc complex at infinity @xmath80 is homotopy equivalent to a wedge of spheres of dimension @xmath8 .
theorem [ thm : har_cont ] was established in ( * ? ? ?
* theorem 1.5 ) ( see @xcite for a concise proof ) and theorem [ thm : har_spheres ] was shown in ( * ? ? ?
* theorem 3.3 ) ( hatcher - vogtmann have simplified the arguments for the key steps in this proof in unpublished work @xcite ) .
consider this portion of the long exact sequence of reduced homology groups for the pair of spaces @xmath86 .
@xmath87 by theorem [ thm : har_cont ] the first and last groups in this sequence are trivial for @xmath88 ; consequently , @xmath89 now combining equations ( [ eq : inf_def ] ) and ( [ eq : a_infinity ] ) we arrive at a very useful description of the steinberg module , @xmath90 the chain complex @xmath91 for cellular homology of the space @xmath92 will provide us with a resolution for @xmath16 as a @xmath15-module .
we define the chain complex @xmath13 to be the shifted complex with @xmath93 and the same boundary maps as @xmath91 .
[ prop : stein_res ] let @xmath13 be the chain complex defined in equation ( [ eq : chain_def ] ) .
the exact sequence @xmath14 is a finite @xmath15-module resolution given in @xcite . ] for the steinberg module @xmath16 .
by theorem [ thm : har_spheres ] the arc complex at infinity @xmath94 has the homotopy type of a wedge of @xmath7-dimensional spheres .
therefore @xmath95 in other words , the chain complex @xmath96 gives a resolution of @xmath16 .
the maximum number of arcs in an arc system occurs when the arcs comprise the one - skeleton of a one - vertex triangulation of @xmath4 .
using euler characteristic one may then conclude that an arc system has at most @xmath97 arcs so @xmath98 and hence @xmath92 is @xmath99 dimensional .
it follows that @xmath100 when @xmath101 .
a @xmath75-filling system corresponds to a unique @xmath102-cell in @xmath103 with the cell decomposition inherited from @xmath98 .
when the orientation of this @xmath102-cell is important we will specify it via an orientation on the unique @xmath102-simplex in @xmath98 mapping to it .
an _ oriented _ @xmath75-filling system will be a @xmath75-filling system together with an order on the set of arcs in the system up to alternating permutations . by definition @xmath104
is the set of finite @xmath105-linear combinations of oriented @xmath75-filling systems - filling systems play a similar role to that of _ modular symbols _ for @xmath28 . ] .
there are a finite number of topologically distinct ways to glue the sides of polygons to get a one - vertex cell decomposition of a surface of genus @xmath2 , and @xmath15 is transitive on cell decompositions of the same topological type .
consequently , as a @xmath15-module , @xmath106 is generated by a finite number of oriented @xmath75-filling systems .
note that in general @xmath106 is not quite a free @xmath15-module since some filling arc systems have nontrivial ( but always finite cyclic ) stabilizers in @xmath15 .
there are two modifications either of which makes the resolution in proposition [ prop : stein_res ] _ projective_. firstly , if coefficients are taken in @xmath107 instead of @xmath108 then proposition [ prop : stein_res ] does give a projective resolution of the rational homology of the curve complex ( see ( * ? ? ?
30 exercise 4 ) ) .
alternatively , if one restricts to some torsion - free subgroup @xmath109 then proposition [ prop : stein_res ] gives a free @xmath10-module resolution of @xmath16 . for the purposes of this paper we will make do with the resolution as is .
note that the tail of this resolution almost provides us with a @xmath15-module presentation for @xmath16 .
it is not exactly a presentation because @xmath110 is not a free @xmath15-module .
however , it can easily be augmented to give a @xmath15-module presentation by artificially adding stabilizer relations which account for the accidental symmetries of certain @xmath20-filling systems . [
prop : st_pres ] let @xmath18 .
choose a set of oriented representatives @xmath111 of each @xmath15-orbit of the set of @xmath20-filling systems .
for each @xmath112 let @xmath113 be a generator of the stabilizer of the arc system for @xmath114 , and let @xmath115 be the sign of the permutation that the mapping class @xmath116 induces on this set of arcs . also choose a set of oriented representatives @xmath117 of each @xmath15-orbit of the set of @xmath30-filling systems .
@xmath16 has a presentation @xmath118 forgetting the marked point gives a surjective ring homomorphism @xmath119 if coefficients in the above presentation are sent to their images under this ring homomorphism then we get a @xmath26-module presentation @xmath120 by proposition [ prop : stein_res ] we have the @xmath15-module isomorphism @xmath121 by definition @xmath110 is spanned by oriented @xmath20-filling systems as a @xmath105-module .
every oriented @xmath20-filling system is of the form @xmath122 for some @xmath112 and @xmath123 .
hence @xmath124 spans @xmath110 as a @xmath15-module .
the only @xmath108-linear dependencies in the set @xmath124 arise from stabilizers of @xmath20-filling systems .
the arc system for a @xmath20-filling system cuts the surface into a single @xmath125-gon so its stabilizer must be a subgroup of the rotational symmetries of the regular @xmath125-gon .
hence , we have the presentation @xmath126 and the presentation for @xmath16 then follows . for the closed surface @xmath3 we do not have a resolution for @xmath25 as a @xmath26-module
; however , we do get a @xmath26-module presentation for @xmath25 by taking co - invariants ( defined below ) in the presentation for @xmath16 .
let @xmath127 be the _ point pushing subgroup _ of @xmath15 ; that is , the kernel of @xmath128 in the birman exact sequence @xcite @xmath129 or for @xmath130 the trivial kernel of @xmath128 in the exact sequence @xmath131 the @xmath132-co - invariants @xmath133 of a @xmath15-module @xmath134 are the quotient of @xmath134 by the @xmath132-submodule generated by the set @xmath135 .
the @xmath132-co - invariants @xmath133 have a @xmath26-module structure ( * ? ? ?
* problem 3 ) .
the exact sequence @xmath136 remains exact @xcite after taking @xmath132-co - invariants to get @xmath137 by corollary [ cor : factors ] and lemma [ lem : same_module ] we have @xmath26-module isomorphisms @xmath138 hence , as a @xmath26-module @xmath25 satisfies @xmath139 all that is left is to observe that taking the @xmath132-co - invariants of @xmath110 achieves the same effect as sending the coefficients in the presentation in ( [ eq : fill0_pres ] ) to their images in @xmath140 .
solomon - tits proved two theorems @xcite .
first the tits building @xmath49 for @xmath27 has the homotopy type of a wedge of spheres of dimension @xmath141 .
second , the @xmath27-module @xmath57 is a cyclic @xmath27-module generated by the @xmath52-sphere coming from a single apartment . in analogy with the first solomon - tits theorem , harer has shown that the curve complex has the homotopy type of a wedge of @xmath7-spheres . in this section
we will prove the analog of the second in theorem [ thm : single_gen ] below which states that as a @xmath5-module the steinberg module @xmath9 is generated by a single element .
actually theorem [ thm : single_gen ] more closely resembles a result of ash - rudolph ( * ? ? ?
* theorem 4.1 ) which implies that the reduced homology of the tits building @xmath49 for @xmath27 is a cyclic @xmath28-module .
the action of @xmath28 is no longer transitive on apartments of @xmath49 , so one must rely on a reduction process to rewrite the homology class of the sphere for an arbitrary apartment as a sum of homology classes of spheres of `` integral unimodular '' apartments . the arc complex at infinity does not come with any apartment structure , but @xmath20-filling systems come in a finite number of types based on their @xmath15-orbits .
proposition [ prop : salient ] below gives a reduction algorithm to write the class of any oriented @xmath20-filling system in the steinberg module for the mapping class group as a linear combination of classes of oriented @xmath20-filling systems of a single type . in order to state and prove theorem [ thm : single_gen ] we will first discuss a notational convenience which will allow us to keep track of complicated filling arc systems .
already in genus two it is difficult to keep track of filling arc systems and their symmetries when drawn on surfaces .
chord diagrams provide a convenient notation for this .
( see ( * ? ? ?
* figure 9 ) and the related discussion for an introduction . ) an ( unlabelled ) _ chord diagram _ ( see figure [ fig : fat_graph ] , left ) will be a regular @xmath142-gon ( which will always be depicted as a circle ) with vertices paired off so that no two adjacent vertices are paired .
the pairing will be indicated by @xmath63 _ chords _ in the chord diagram .
we will call an edge of the @xmath142-gon an _ outer edge _ of the chord diagram .
a _ labelled chord diagram _ ( see figure [ fig : chord_diagram ] , right ) will be a chord diagram in which each chord is labelled on one side by an element of @xmath143 and on the other side by its inverse .
in practice we will only label one side of each chord with the assumption that the other side is labelled with the inverse element in @xmath143 .
two labelled or unlabelled chord diagrams are the same if they can be made to agree after rotation of the @xmath142-gon .
a labelling of a chord diagram will be called _ proper _ if it comes from a filling arc system in the surface @xmath4 .
we implicitly require all labellings of chord diagrams to be proper . in an arc system on the surface @xmath4
, arcs leave and return to a neighborhood of the marked point @xmath58 in a certain cyclic order . to each filling arc system in @xmath4 with @xmath63 arcs ,
associate the labelled chord diagram got by placing a @xmath142-gon in a neighborhood of the marked point so that each arc enters and leaves the @xmath142-gon at a vertex .
for each arc in the arc system connect the two vertices of the @xmath142-gon on that arc with a chord .
the two sides of the chord correspond to the two sides of the arc .
label each side of the chord by the element of @xmath144 got by following the arc on the corresponding side in a counter - clockwise direction ( see figure [ fig : chord_diagram ] ) .
we will do most of our calculations using chord diagrams , so it will be convenient to be able to recover certain characteristics of the surface and embedded filling arc system corresponding to a given chord diagram .
firstly we briefly explain how to reconstruct the surface . from an @xmath63-chord diagram
we may construct a surface with a certain embedded , one - vertex graph as follows .
fat graph _ is a regular neighborhood of a graph embedded in a surface together with the embedding of the graph , or equivalently , a graph together with for each vertex of the graph a cyclic order on the termini of the edges incident with that vertex .
a chord diagram specifies a one - vertex fat graph with one edge for each chord where the cyclic order on the edge termini corresponds to the cyclic order on endpoints of the chords in the diagram ( see figure [ fig : fat_graph ] ) .
now we may glue disks to each of the @xmath145 boundary components of the fat graph to get a closed surface in which the edges of the fat graph form a filling arc system .
note that one generally considers the fat graph dual to an arc system @xcite @xcite , but here our arc system and fat graph agree
. it will be useful to know the genus @xmath2 of this closed surface , which is easily calculated using the number @xmath63 of edges in the fat graph and the number @xmath145 of boundary components of the fat graph , to be @xmath146 in particular , this shows that a chord diagram with @xmath74 chords corresponds to a @xmath20-filling system in a surface of genus @xmath2 precisely when the corresponding fat graph has one boundary component . in light of equation ( [ eq : euler ] )
we would like to be able to quickly read off the number @xmath145 of boundary components in the fat graph corresponding to a given chord diagram .
in fact it is easy to see that @xmath145 is the number of _ cycles _ in the chord diagram .
a cycle in a chord diagram ( see figure [ fig : cycles ] ) is an alternating sequence of chords and outer edges of the chord diagram got by starting at a point just inside an outer edge of the diagram and walking along in a clockwise direction keeping the outer edge on one s left until a chord is encountered , turning right and then following the chord keeping it on the left until an outer edge is encountered , turning right and repeating until one returns to the starting point .
notice that for each cycle in a labelled chord diagram the products of the labels in the cycle must be @xmath147 , because the corresponding loop in the surface bounds a disk .
two chords are _ parallel _ if along with two outer edges they bound a rectangular cycle .
( the left chord diagram in figure [ fig : cycles ] has two parallel chords . ) since we will only be concerned with arc systems with no parallel arcs , we will not consider chord diagrams with parallel chords .
we may identify @xmath75-filling systems for the surface @xmath4 with ( properly ) labelled chord diagrams with @xmath76 chords , @xmath148 cycles and no parallel chords .
when the choice of orientation for a @xmath75-filling system expressed as a chord diagram is important , we will indicate it by marking a starting point on the edge of the diagram with a `` @xmath149 '' and then ordering the chords as they are first encountered traveling clockwise around the edge of the diagram .
this convention has the disadvantage of sometimes only allowing one of the two possible orientations of a @xmath75-filling system to be specified .
however , we will use it due to its notational economy . as an oriented @xmath75-filling system gives an oriented cell in @xmath92
, we may take its boundary in the chain complex @xmath150 by taking the boundary of the corresponding oriented simplex in @xmath98 and projecting that boundary back to @xmath92 .
the boundary of a simplex is an alternating sum of the codimension-1 faces .
thus the boundary of a @xmath75-filling system @xmath151 will be a linear combination of all the @xmath152-filling systems got from @xmath151 by removing one arc .
removing certain arcs from @xmath151 may leave an arc system which no longer fills @xmath4 .
the simplex of @xmath98 for such an arc system is contained in @xmath94 and hence is trivial in @xmath153 . in the language of chord diagrams we will be able to tell when this has happened by counting cycles and applying equation ( [ eq : euler ] ) . by the dehn - nielsen - baer theorem @xcite
, the mapping class group @xmath15 acts as the index @xmath31 subgroup @xmath154 contained in the full automorphism group of the fundamental group of the closed surface of genus @xmath2 .
this gives a left action of @xmath15 on labelled chord diagrams by modifying labels .
this action of @xmath15 preserves the underlying unlabelled chord diagram for a @xmath75-filling system .
conversely , any homeomorphism of fat graphs extends to a homeomorphism of the surface so the mapping class group is transitive on proper labellings for a fixed unlabelled chord diagram .
thus the @xmath15-orbits of @xmath75-filling systems correspond precisely to unlabelled chord diagrams with @xmath76 chords and @xmath148 cycles .
figure [ fig : gen_2 ] depicts the @xmath155 orbits of the action of @xmath156 on the set of @xmath20-filling systems for the surface @xmath157 .
these are all @xmath155-chord diagrams with one cycle and ( redundantly ) no parallel chords . as shown in
[ sec : st_pres ] we have the @xmath15-module isomorphism from proposition [ prop : st_pres ] @xmath121 although @xmath110 and @xmath158 are finitely generated @xmath15-modules , the number of @xmath15-orbits of @xmath20-filling systems grows very quickly as a function of @xmath2 ( see @xcite ) .
it is therefore somewhat surprising that the steinberg module is generated by the class of a single @xmath20-filling system .
[ thm : single_gen ] let @xmath18 , and let @xmath19 be the @xmath20-filling system in figure [ fig : st_gen ] .
let @xmath21 $ ] be the class of @xmath22 in @xmath23 .
then @xmath16 is generated as a @xmath15-module by @xmath21 $ ] .
for the closed surface let @xmath21 $ ] be the class of @xmath22 in @xmath24 ( see the proof of proposition [ prop : st_pres ] ) .
similarly , @xmath25 is generated as a @xmath26-module by @xmath21 $ ] .
theorem [ thm : single_gen ] will be proved using propositions [ prop : salient ] and [ prop : connected ] below .
[ cor : notriv ] @xmath21 \in \st(\surface_g^1)$ ] ( resp .
@xmath21 \in \st(\surface_g)$ ] ) is nontrivial for @xmath18 .
if @xmath21 $ ] were trivial then by theorem [ thm : single_gen ] the steinberg module @xmath16 would be trivial .
but then @xmath15 would have a finite index subgroup @xmath10 whose dualizing module is trivial .
this would imply that @xmath10 must be the trivial group and hence that @xmath15 is finite ( see @xcite for more ) . [ fig : st_gen],title="fig : " ] ( -62,82)@xmath159 ( -52,66)@xmath160 ( -44,46)@xmath161 ( -50,28)@xmath162 ( -80,82)@xmath163 ( -92,66)@xmath164 ( -99,46)@xmath165 ( -92,28)@xmath166 ( -79,0)@xmath167 ( -71,-24)@xmath22 a natural first guess at a single generator for @xmath16 is the arc system coming from the standard identification of the the @xmath125-gon which corresponds to the labelled chord diagram in figure [ fig : guess ] , but in fact by proposition [ prop : connected ] below , the class of this @xmath20-filling system is trivial in @xmath16 .
[ fig : guess],title="fig : " ] ( -62,82)@xmath168 ( -44,66)@xmath169 ( -36,44)@xmath170 ( -48,23)@xmath171 ( -80,82)@xmath172 ( -99,64)@xmath173 ( -106,42)@xmath174 ( -92,22)@xmath175 ( -79,0)@xmath167 a filling arc system describes a decomposition of the surface @xmath4 into polygons . if any of the polygons has more than @xmath176 vertices , one may add an arc connecting two non - adjacent vertices of that polygon to get a filling arc system with one more arc .
this same process can be done from the point of view of chord diagrams . given
a @xmath75-filling system @xmath151 in @xmath4 represented as a labelled chord diagram , the outer edges of @xmath151 are partitioned into @xmath77 disjoint sets ( corresponding to the polygons mentioned above ) according on which of the @xmath77 cycles they belong to .
a @xmath177-filling system can be created by adding a new chord ( not parallel to any of the chords of @xmath151 ) connecting any two outer edges of @xmath151 in the same cycle . note that
since a @xmath20-filling system has a single cycle , _ any _ new ( non - parallel ) chord gives a @xmath30-filling system .
two chords in a chord diagram _ cross _ if the cyclic order on their endpoints forces them to
. the finest partition of the set of chords in which each pair of crossing chords is in the same set will give the _ connected components _ of the chord diagram .
for example the chord diagram in figure [ fig : st_gen ] is connected , while the chord diagram in figure [ fig : guess ] has @xmath2 connected components and so is disconnected for @xmath178 .
[ prop : connected ] if @xmath151 is a disconnected @xmath20-filling system then @xmath179 $ ] is trivial in @xmath23 .
add a chord @xmath180 to @xmath151 which does not cross any of the chords of @xmath151 and such there are nonempty connected components of @xmath151 on both sides of @xmath180 .
let @xmath181 be the resulting chord diagram .
firstly , we claim that @xmath181 is a @xmath30-filling system
. this will be the case unless the chord @xmath180 is parallel to some chord @xmath182 of @xmath151 .
suppose that such a chord @xmath182 exists .
then @xmath182 crosses the same set of chords as @xmath180
so @xmath182 also divides the chord diagram @xmath151 into two halves with no communicating chords .
it follows that the two sides of @xmath182 must be in different cycles of @xmath151 contradicting the fact that @xmath20-filling systems have exactly one cycle . now since @xmath181 is a @xmath30-filling system we may orient it and take its boundary .
next we claim @xmath183 by definition @xmath184 is a linear combination of the @xmath20-filling systems that one can get by removing a single chord of @xmath181 .
the chord @xmath180 cuts the chord diagram into two halves with no communicating chords , so by the argument above , any diagram with a subset of the chords of @xmath181 which contains @xmath180 has at least two cycles .
equation ( [ eq : connection_relation ] ) follows establishing the proposition .
now we will introduce a subset of the @xmath20-filling systems whose classes generate @xmath16 and for which , by proposition [ prop : connected ] , the class of each element , save one , is trivial .
, title="fig : " ] ( -92,4)@xmath167 a chord diagram will be called _ salient _ if each of its connected components is of the form given in figure [ fig : salient ] . of course @xmath22 from figure [ fig : st_gen ] is the unique connected , salient @xmath20-filling system .
[ prop : salient ] @xmath16 is generated by salient @xmath20-filling systems .
for @xmath185 a chord diagram will be said to have a _ salient tail of length @xmath63 _ if a neighborhood of some segment in the boundary of the diagram is of the form pictured in figure [ fig : salient_tail ] . in a chord diagram[fig : salient_tail],title="fig : " ] ( -120,22)@xmath186 ( -71,4)@xmath167 given a @xmath20-filling system @xmath151 with a salient tail of length @xmath63 , we can add a chord @xmath180 to the far right of the salient tail to get a @xmath30-filling system @xmath181 with a salient tail of length @xmath64 .
consider the relation @xmath187 notice that if any single chord in the salient tail of @xmath181 other than @xmath180 is removed from @xmath181 then the resulting chord diagram is disconnected , and so by proposition [ prop : connected ] the class of that @xmath20-filling system is trivial in @xmath16 .
thus for every @xmath188 with @xmath189 the class of the @xmath20-filling system @xmath190 is either trivial in @xmath16 or the chord diagram for @xmath190 has a salient tail of length @xmath64 . repeating this process no more than @xmath74 times we may write the class of any @xmath20-filling system as a linear combination of classes of salient @xmath20-filling systems . with propositions [ prop : salient ] and
[ prop : connected ] established , the proof of theorem [ thm : single_gen ] is immediate . by proposition [ prop : salient ] , @xmath16 is generated by salient @xmath20-filling systems .
the element @xmath19 is a representative of the unique @xmath15-orbit of connected , salient @xmath20-filling systems .
thus by proposition [ prop : connected ] , @xmath21 $ ] generates @xmath16 . for the closed surface without a marked point @xmath3 we have @xmath191 as @xmath15-modules
so @xmath21 $ ] generates @xmath25 as a @xmath15-module . on the other hand , the @xmath15-module structure of @xmath25 factors through @xmath26 so @xmath21 $ ] generates @xmath25 as a @xmath26-module . as a consequence of theorem [ thm : single_gen ]
there is a left ideal @xmath192 such that the @xmath26-module sequence @xmath193 is exact . in theory for a fixed genus @xmath2 the presentation in proposition [ prop : stein_res ] gives enough information to calculate a finite generating set for @xmath194 .
we will perform this calculation for the surfaces @xmath195 and @xmath157 in [ sec : g1andg2 ] below . a more `` geometric '' understanding of @xmath194
would certainly advance the understanding of @xmath25 . at minimum
one would like to know the stabilizer in @xmath26 of the class @xmath21 \in \st(\surface_g)$ ] . as illustrated above it
is useful to view the steinberg module as the reduced homology of the arc complex at infinity . on the other hand ,
the curve complexes @xmath196 and @xmath197 are more widely studied so we will now explain harer s homotopy equivalence @xmath198 between the arc complex at infinity and the curve complex
. we will use this map to get an explicit nontrivial @xmath31-sphere in the curve complex of the surface of genus @xmath31 ( see proposition [ prop:2-gen ] below ) which will motivate conjecture [ conj : nontriv ] .
we point out that the recent results of penner - mcshane @xcite provide an alternative approach to converting a @xmath20-filling system into a sphere in the curve complex .
harer ( * ? ? ?
* theorem 3.4 ) gives a homotopy equivalence from the arc complex at infinity to the curve complex .
harer s map @xmath199 is simplicial where @xmath200 denotes the second barycentric subdivision of @xmath80 and @xmath201 denotes the first barycentric subdivision of @xmath62 .
the map @xmath198 is defined on the vertices of @xmath200 , as follows .
a vertex @xmath202 of @xmath200 is a nested sequence of non - filling arc systems @xmath203 . for @xmath204
if one removes a small open regular neighborhood of the union of the arcs in @xmath190 from @xmath4 , one is left with a surface @xmath205 with nonempty boundary . after omitting redundancies and trivial curves , the boundary components of @xmath206
give a curve system @xmath207 in @xmath4 .
we define @xmath208 again omitting redundancies .
in fact , @xmath209 is a curve system since if @xmath210 then we can arrange that @xmath211 and hence the curves of @xmath207 can be taken to be disjoint from the curves of @xmath212 .
of course @xmath209 is a vertex of @xmath201 whose vertices are curve systems .
we then extend @xmath198 simplicially ; that is , the simplex with vertices @xmath213 is mapped linearly to the simplex with vertices @xmath214 .
the @xmath20-filling system @xmath22 in figure [ fig : st_gen ] gives an arc system whose class in @xmath215 is nontrivial . from the long exact sequence for the homology of the pair of spaces
@xmath216 we have the isomorphism @xmath217 in @xmath85 the arc system for @xmath22 gives a @xmath218-simplex all of whose proper subfaces are contained in @xmath219 . the class of @xmath220 \in \widetilde{\hh}_{2g-2 } ( \ac_\infty ; \z ) $ ] is represented by the boundary of this @xmath218-simplex . for the surface
@xmath157 we may compute the image of @xmath221 under @xmath198 directly .
figure [ fig : psi_image ] gives the image of the vertices of the first barycentric subdivision of @xmath221 .
one may easily fill in the images for vertices in the second barycentric subdivision by taking unions of the curve systems at the vertices of the first subdivision .
the sphere in figure [ fig : psi_image ] is nontrivial , but it is slightly unsatisfying in that it is specified in @xmath222 and not @xmath223 . figure [ fig : sphere ] and proposition [ prop:2-gen ] below provide a homotopic sphere with a nicer description .
the reader may recognize the shape in figure [ fig : sphere ] to be the boundary of the dual of the @xmath176-dimensional associahedron ( see @xcite ) .
[ prop:2-gen ] the @xmath156-orbit of the homology class of the @xmath31-sphere in @xmath223 pictured in figure [ fig : sphere ] generates @xmath224 . forgetting the marked point gives a @xmath31-sphere in @xmath225 whose @xmath226-orbit generates @xmath227 . the arc system for the generator @xmath21 \in
\st(\surface_2 ^ 1)$ ] from theorem [ thm : single_gen ] is pictured in figure [ fig : chord_diagram ] .
one may use harer s map directly to show that the vertices in the first barycentric subdivision of @xmath221 are mapped under @xmath198 to the curve systems pictured in figure [ fig : psi_image ] .
we will give a simplicial sphere in the unbarycentricly subdivided curve complex which is homotopic to the sphere pictured in figure [ fig : psi_image ] .
one may construct a homotopy equivalence @xmath228 as follows .
let @xmath229 be the set of exactly those curves appearing in figure [ fig : psi_image ] .
fix some linear order on @xmath229 so that each of the curves that appear in figure [ fig : sphere ] is greater than all of those that do not
. let @xmath230 be the map satisfying the following three properties : 1 .
the map @xmath231 fixes the vertices of @xmath223 .
2 . for any simplex @xmath232 of @xmath223 whose vertices are disjoint from @xmath229 the map @xmath233 is the identity .
if at least one vertex of the simplex @xmath232 of @xmath223 is in the set @xmath229 then @xmath233 is the map which is linear on the barycentric subdivision of @xmath232 , fixes the vertices of @xmath232 and sends the barycenter of @xmath232 to the greatest vertex of @xmath232 in the order on @xmath229 .
the map @xmath234 is homotopic to the identity .
in fact , sending each point of a simplex to weighted averages of itself and its image under @xmath231 gives the homotopy between @xmath231 and the identity .
it follows that the sphere @xmath235 is homotopic to @xmath236 .
some of the vertices in figure [ fig : psi_image ] coalesce under @xmath231 and we get the simpler homotopic sphere pictured in figure [ fig : sphere ]
. we may then conclude that the @xmath156-orbits of sphere in figure [ fig : sphere ] generate @xmath224 . forgetting the marked point
gives a @xmath31-sphere in @xmath225 whose @xmath226-orbit generates @xmath227 ( * ? ? ?
* lemma 3.6 ) .
the number of vertices in the boundary of the first barycentric subdivision of the @xmath218-simplex is @xmath237 .
hence , for large @xmath2 the direct approach to producing homologically nontrivial spheres in @xmath62 given in the proof of proposition [ prop:2-gen ] is impractical .
moreover , the asymmetry of the arc system for @xmath22 is likely to yield a sphere in @xmath62 which is difficult to even describe in any concise form . using the curves in figure [ fig : assoc ]
there is a simple construction of a map of a @xmath7-sphere into @xmath62 .
let @xmath238 be the boundary of the dual of the @xmath218-dimensional associahedron which we now define .
fix a regular @xmath239-gon @xmath240 . by definition @xmath238
is the simplicial complex whose vertices are diagonals of @xmath240 and whose simplices are given by sets of disjoint diagonals in @xmath240 .
it is well - known that @xmath238 is a @xmath218-dimensional sphere @xcite .
associate the vertices of @xmath240 with the @xmath241 curves in figure [ fig : assoc ] so that adjacent vertices of @xmath240 correspond to intersecting curves .
each pair of nonintersecting curves in figure [ fig : assoc ] corresponds to a diagonal of @xmath240 .
let @xmath242 be the simplicial map sending each pair of nonintersecting curves in figure [ fig : assoc ] to the curve system ( with at most @xmath155 nontrivial curves ) consisting of the boundary curves of the surface got by removing an open regular neighborhood of the union of all the other curves . for each vertex @xmath202 of @xmath238 choose a curve @xmath243 and let @xmath244 be the simplicial map sending @xmath202 to @xmath245 .
one may construct a homotopy like the one in the proof of proposition [ prop:2-gen ] above to show that @xmath246 and @xmath247 are homotopic and hence the homotopy class of @xmath246 is independent of the choices of @xmath243 .
[ conj : nontriv ] for @xmath18 the class @xmath248 \in \widetilde{\hh}_{2g-2}(\cc(\surface_g^1);\z)$ ] is nontrivial .
( when @xmath130 the proper picture for figure [ fig : assoc ] consists of @xmath31 pairs of parallel curves . ) by proposition [ prop:2-gen ] above the conjecture holds for @xmath249 .
it also holds and @xmath31 are somewhat exceptional .
( for instance they have nontrivial centers ( * ? ? ?
* theorem 7.5.d ) . ) ] for @xmath130 .
theorem [ thm : single_gen ] shows that the steinberg module is generated by a single element ; however , the presentation of the steinberg module given in proposition [ prop : st_pres ] has multiple generators . for the surfaces of genus one and two ( with or without marked points )
we will give presentations for the steinberg module based on the single generator from theorem [ thm : single_gen ] .
a natural first example to consider is @xmath250 . for the surface of genus @xmath30 we have @xmath251 .
identify @xmath252 with the column vector @xmath253 and @xmath254 with @xmath255 .
when @xmath130 we do not have the birman exact sequence ( [ eq : bes ] ) .
in fact we have @xmath256 which we may identify with @xmath257 .
while the dualizing modules for torsion - free , finite index subgroups of @xmath257 are already well - known , the reader may find the proof of proposition [ prop : g1_pres ] to be illustrative of the content of proposition [ prop : st_pres ] .
for this case we have the good fortune to be able to draw the entire arc complex @xmath258 which is a @xmath31-complex with @xmath30-skeleton the farey graph ( see figure [ fig : farey ] ) .
the arc complex at infinity @xmath259 is the @xmath20-skeleton of this complex . with @xmath30 marked point[fig :
farey],title="fig : " ] ( -116,139)@xmath260 ( -119,110 ) [ prop : g1_pres ] identify the groups @xmath261 the steinberg module @xmath262 has a @xmath257-module presentation with free generator @xmath263 and relations @xmath264 \mathbf{f}_0 \\ \big [ 1 + \left ( \begin{array}{cc}\scriptstyle 0 & \scriptstyle 1 \\ \scriptstyle -1 & \scriptstyle 0\end{array } \right ) \big ] \mathbf{f}_0.\end{gathered}\ ] ] by equations ( [ eq : a_infinity ] ) and ( [ eq : st_working_def ] ) we have @xmath265 the resolution of @xmath250 given in proposition [ prop : stein_res ] is of the form @xmath266 where @xmath110 is generated by the @xmath20-filling system @xmath22 in figure [ fig : gen_1_pres ] and @xmath158 is generated by the @xmath30-filling system @xmath260 . recall from [ sec : chord_diagrams ] above that we specify an orientation on a @xmath75-filling system by marking a starting point on the edge of a chord diagram with a `` @xmath149 '' and then ordering the chords as they are first encountered traveling clockwise around the edge of the diagram .
labelled with @xmath267.[fig : gen_1_pres],title="fig : " ] ( -11,46)@xmath149 ( -17,35)@xmath253 ( -27,11)@xmath255 ( -32,-10)@xmath22 labelled with @xmath267.[fig : gen_1_pres],title="fig : " ] ( -7,40)@xmath149 ( -17,35)@xmath253 ( -18.5,15)@xmath255 ( -39.5,6)@xmath268 ( -32,-10)@xmath260 proposition [ prop : st_pres ] gives a presentation for @xmath250 with the single generator @xmath22 and two relations .
we get the first relation by taking the boundary of @xmath260 . c ( 10,0 ) ( 0,15)@xmath269 ( -6.8,32)@xmath149 ( -16.5,29)@xmath270 ( -16.5,12)@xmath271 ( -33,6)@xmath272 ( 20,0 ) ( 3,15)@xmath273 ( -10.3,37)@xmath149 ( -18,29)@xmath271 ( -24.5,11)@xmath272 ( 20,0 ) ( 3,15)@xmath274 ( -10.3,37)@xmath149 ( -18,29)@xmath270 ( -24.5,11)@xmath272 ( 20,0 ) ( 3,15)@xmath275 ( -10.3,37)@xmath149 ( -18,29)@xmath270 ( -23,9)@xmath271 + ' '' '' @xmath276 - \left ( \begin{array}{cc}\scriptstyle 1 & \scriptstyle -1 \\ \scriptstyle 0 & \scriptstyle 1\end{array } \right ) [ \phi_0 ] + [ \phi_0 ] $ ] the other relation comes from the fact that @xmath110 is not a free @xmath277-module .
in fact , the arc system for the @xmath20-filling system @xmath22 has a cyclic stabilizer of order @xmath155 so @xmath22 satisfies the relation c ( -10.3,37)@xmath149 ( -18,29)@xmath270 ( -23,9)@xmath271 ( 20,0 ) ( 3,15)@xmath273 ( 20,0 ) ( 3,15)@xmath274 ( -10.3,37)@xmath149 ( -23,29)@xmath278 ( -23.5,9)@xmath270 + ' '' '' @xmath279 .
the proposition follows .
now we will use the same techniques as in the previous section to convert the @xmath155-generator presentations for @xmath280 and @xmath281 given in proposition [ prop : st_pres ] to the relatively efficient one - generator presentations given in proposition [ prop : g2_pres ] . before we can proceed to write down an explicit @xmath156-module presentation for @xmath280 we must give a set of mapping classes @xmath282 which generate @xmath156 along with their action on @xmath283 where @xmath284 are as in figure [ fig : chord_diagram ] .
let @xmath285 be the mapping class of order @xmath286 @xmath287 the mapping class @xmath288 generates the stabilizer of the arc system for a certain @xmath30-filling system in the @xmath156-orbit depicted in the top row and third column of figure [ fig : g2_relations ] .
let @xmath289 be the dehn twist indicated in figure [ fig : twists ] .
its action on @xmath290 is given by @xmath291 where @xmath289 fixes generators not listed .
let @xmath292 be the dehn twist indicated in figure [ fig : twists ] .
its action on @xmath290 is @xmath293 for convenience , for @xmath294 set @xmath295 [ prop : g2_pres ] for mapping class group of the surface of genus @xmath31 with one marked point the steinberg module @xmath280 has a @xmath156-module presentation with generator @xmath263 and relations @xmath296 where the mapping classes @xmath297 are defined in equations ( [ eq : mc_start]-[eq : mc_end ] ) .
taking the image in @xmath298 of the coefficients of the relations above , one gets a @xmath226-module presentation for @xmath281 .
firstly , by proposition [ prop : connected ] the connected @xmath20-filling systems in figure [ fig : con_gen_2 ] provide a sufficient generating set for @xmath280 , and since disconnected @xmath30-filling systems provide relations between disconnected ( and hence trivial ) @xmath20-filling systems , we need only consider relations coming from connected @xmath30-filling systems .
all orbits under the action of @xmath156 of all such @xmath30-filling systems are pictured in figure [ fig : g2_relations ] .
[ fig : con_gen_2],title="fig : " ] ( -5.8,15.5)@xmath149 ( -17,17)@xmath299 ( -31,24)@xmath300 ( -52,25)@xmath301 ( -34,44)@xmath302 ( -32,-10)@xmath22 [ fig : con_gen_2],title="fig : " ] ( -21,53)@xmath149 ( -21,40)@xmath300 ( -10,32)@xmath301 ( -28,6)@xmath303 ( -28,45)@xmath302 ( -32,-10)@xmath304 [ fig : con_gen_2],title="fig : " ] ( -6,38)@xmath149 ( -15,32)@xmath305 ( -12,16)@xmath301 ( -28,8)@xmath303 ( -43,10)@xmath302 ( -32,-10)@xmath306 let @xmath22 , @xmath304 , and @xmath306 be the @xmath20-filling systems pictured in figure [ fig : con_gen_2 ] . note that @xmath22 and @xmath304 are free generators since any mapping class fixing the corresponding arc systems must be the identity . on the other hand , @xmath306 is _ not _ a free generator .
setwise , its corresponding arc system has stabilizer a cyclic subgroup of @xmath156 of order 8 . ' '' '' ' '' '' * i * & + ' '' '' ' '' '' * ii * & & * iii * & & * iv * & [ cols="^ " , ] + ' '' '' ' '' '' * v * & & * vi * & + we could fix a proper labelling for each of the chord diagrams in figure [ fig : g2_relations ] and compute a presentation from the resulting relations .
however , before we introduce the added complexity of labels we will use the exactness of the sequence @xmath307 to eliminate much of the redundancy in the relations induced by these @xmath30-filling systems .
( actually we will only use the fact that @xmath308 . ) in fact , the smaller set of relations coming from representatives of the @xmath156-orbits of @xmath30-filling systems in figure [ fig : smaller_relations ] suffice which we now show .
firstly , using each @xmath31-filling system in the top row of figure [ fig : redun_rel](i ) we can exhibit the @xmath30-filling system below it as a linear combination of the @xmath30-filling systems in our smaller set in figure [ fig : smaller_relations ] . now using each @xmath31-filling system in the top row of figure [ fig :
redun_rel](ii ) we can exhibit the @xmath30-filling system below it as a linear combinations of @xmath30-filling systems in figure [ fig : smaller_relations ] and figure [ fig : redun_rel](i ) . continuing in this manner in for figure [ fig : redun_rel](iii - vi )
we can derive the remaining relations as consequences of the relations coming from preceding @xmath30-filling systems .
we note that the elimination of redundant relations that lead to the list in figure [ fig : smaller_relations ] was performed in a completely ad hoc manner .
it is not clear if this set of relations is minimal or if there are further redundancies .
ideally , one would like a systematic method for eliminating redundant relations .
such a method might yield a one - generator presentation for the steinberg module for every genus .
now we will assign arbitrary proper labels to our sufficient set of @xmath30-filling systems given in figure [ fig : smaller_relations ] and use the action of @xmath156 on @xmath309 to convert them into explicit @xmath156-module relations .
c ( 10,0 ) ( 0,15)@xmath269 ( -26.3,43.7)@xmath149 ( -23.5,38)@xmath310 ( -13.5,35)@xmath311 ( -10,15)@xmath312 ( -28,4)@xmath313 ( -45,27.5)@xmath314 ( 20,0 ) ( 4,15)@xmath273 ( -5.7,12.3)@xmath149 ( -10,13)@xmath311 ( -22,4)@xmath312 ( -38,20)@xmath313 ( -12,28)@xmath314 ( 20,0 ) ( 3,15)@xmath275 ( -5.7,12.3)@xmath149 ( -12,13)@xmath310 ( -22,5)@xmath311 ( -38,20)@xmath312 ( -15,21)@xmath313 + ' '' '' @xmath315 + [ \phi_0]$ ] ccl ( 10,0 ) ( 0,15)@xmath269 ( -5,14)@xmath149 ( -10,15)@xmath310 ( -16.5,6)@xmath316 ( -28,6)@xmath311 ( -41,19.5)@xmath312 ( -34.5,36.5)@xmath313 ( -37,-7)@xmath317 & ( 20,0 ) ( 4,15)@xmath273 & ( 20,0 ) ( 4,15)@xmath274 ( -5.7,12.3)@xmath149 ( -14,15)@xmath311 ( -22.5,6)@xmath312 ( -38,20)@xmath313 ( -11,28)@xmath316 ( -37,-7)@xmath317 ( 20,0 ) ( 3,15)@xmath274 ( -5.7,12.3)@xmath149 ( -13,13.5)@xmath310 ( -22.5,6.5)@xmath311 ( -38,20)@xmath312 ( -15,21)@xmath313 ( 20,0 ) ( 3,15)@xmath274 ( -5.7,12.3)@xmath149 ( -13,13.5)@xmath312 ( -22,4)@xmath313 ( -42,26)@xmath316 ( -15,21)@xmath310 ( -37,-7)@xmath317 + & & ' '' '' ( 20,0 ) ( 3,15)@xmath274 ( -5.7,12.3)@xmath149 ( -13,13.5)@xmath313 ( -29,4)@xmath316 ( -38,20)@xmath310 ( -15,21)@xmath311 ( -37,-7)@xmath317 ( 20,0 ) ( 3,15)@xmath274 ( -5.7,12.3)@xmath149 ( -17,7)@xmath316 ( -22,4)@xmath310 ( -38,20)@xmath311 ( -15,21)@xmath312 ( -37,-7)@xmath317 + c ( 10,0 ) ( 0,15)@xmath269 ( -39.5,2.5)@xmath149 ( -37,11)@xmath310 ( -44,19)@xmath311 ( -23,41)@xmath312 ( -30,12)@xmath313 ( -32,36)@xmath316 ( -44,-7)@xmath318 ( 20,0 ) ( 3,15)@xmath273 ( -5.7,12.3)@xmath149 ( -13,13.5)@xmath310 ( -22.5,6)@xmath311 ( -38,20)@xmath312 ( -15,21)@xmath313 ( 20,0 ) ( 4,15)@xmath274 ( -5.7,12.3)@xmath149 ( -13,13)@xmath316 ( -22.5,6)@xmath312 ( -41,28)@xmath311 ( -14,21)@xmath310 ( -44,-7)@xmath318 + ' '' '' @xmath319 - s_0^{-1}t_0s_0t_0s_0t_0t_2s_4^{-1}s_1s_4r[\phi_0 ] $ ] c ( 10,0 ) ( 0,15)@xmath269 ( -5,14)@xmath149 ( -9,15)@xmath311 ( -28,4)@xmath312 ( -17.5,7)@xmath320 ( -43,19)@xmath321 ( -14,20)@xmath313 ( 20,0 ) ( 4,15)@xmath273 ( -5.7,30)@xmath149 ( -12,27)@xmath320 ( -11,13)@xmath312 ( -22,6)@xmath316 ( -35,7)@xmath313 ( -35,-7)@xmath322 ( 20,0 ) ( 3,15)@xmath274 ( -17.7,41.7)@xmath149 ( -19,30)@xmath311 ( -9,26)@xmath312 ( -23.5,9)@xmath316 ( -23,36)@xmath313 ( -35,-7)@xmath322 ( 20,0 ) ( 3,15)@xmath275 ( -17.7,41.7)@xmath149 ( -19,30)@xmath311 ( -11,27)@xmath320 ( -23.5,9)@xmath316 ( -23,36)@xmath313 ( -35,-7)@xmath322 + ' '' '' @xmath323 - [ \phi_1 ] + t_1^{-1}[\phi_1]$ ] c ( 10,0 ) ( 0,15)@xmath269 ( -5,14)@xmath149 ( -10,15)@xmath310 ( -16.5,6)@xmath311 ( -37,10)@xmath312 ( -42,27.5)@xmath321 ( -14,20)@xmath313 ( 20,0 ) ( 4,15)@xmath273 ( -17.7,41.7)@xmath149 ( -19,30)@xmath311 ( -9,26)@xmath312 ( -23,8)@xmath316 ( -22,38)@xmath313 ( -35,-7)@xmath322 ( 20,0 ) ( 3,15)@xmath274 ( -5.7,12.3)@xmath149 ( -13,13.5)@xmath310 ( -22.5,6)@xmath311 ( -38,20)@xmath312 ( -15,21)@xmath313 ( 20,0 ) ( 4,15)@xmath275 ( -5.7,12.3)@xmath149 ( -13,13.5)@xmath310 ( -22.5,6)@xmath311 ( -38,19)@xmath312 ( -19,20)@xmath321 + ' '' '' @xmath324 - [ \phi_0 ] + s_1^{-1}[\phi_0]$ ] in order to get a proper presentation for @xmath280 as a quotient of a _
@xmath156-module we need one more relation to account for the fact that the @xmath20-filling system @xmath306 has a nontrivial ( but cyclic ) stabilizer : c ( -5.7,30)@xmath149 ( -15,26.5)@xmath325 ( -11,13)@xmath312 ( -22.5,6)@xmath316 ( -35,7)@xmath313 ( -35,-7)@xmath322 ( 20,0 ) ( 3,15)@xmath273 ( 20,0 ) ( 3,15)@xmath274 ( -5.7,30)@xmath149 ( -18,25)@xmath326 ( -13,14)@xmath327 ( -22,6)@xmath312 ( -35.5,7)@xmath316 ( -35,-7)@xmath322 ( -35,-17)@xmath328 + ' '' '' @xmath329 notice that this relation is satisfied by @xmath330 itself , not just the class @xmath331\in \st(\surface_2 ^ 1)$ ] . in summary , we now have the @xmath156-module presentation with free generators @xmath332 and relations @xmath333 combining the last three relations to eliminate the generators @xmath334 and @xmath335 we get the desired one - generator presentation .
the negative of the identity matrix @xmath336 stabilizes every subspace of @xmath50 so @xmath337 is in the kernel of the action of @xmath28 on its steinberg module @xmath55 .
consequently the action of @xmath28 on @xmath55 factors though an action of the simple group @xmath338 ( see ( * ? ? ?
* theorem 9.3 ) ) .
it follows that @xmath55 is a faithful @xmath339-module .
as we have seen in corollary [ cor : factors ] , for @xmath340 the kernel of the action of @xmath15 on the steinberg module contains the infinite point pushing subgroup @xmath127 . for the closed surface without a marked point
there is some hope that the action of the mapping class group ( modulo its center ) on the steinberg module is faithful .
when @xmath341 , the hyperelliptic involution generates the center @xmath342 of the mapping class group and acts trivially on the curve complex . for @xmath343 the center @xmath342 is trivial ( * ? ?
* theorem 7.5.d ) .
thus for @xmath18 the action of @xmath26 on @xmath25 factors through @xmath344 . the solomon - tits theorem ( _ cf .
_ @xcite , ( * ? ? ? * theorem 2 ) ) shows that the tits building @xmath49 has the homotopy type of a wedge of spheres , that its reduced homology is a cyclic module ( over the associated @xmath0-group ) , and futher gives a @xmath105-basis for the reduced homology .
harer has shown that the curve complex has the homotopy type of a wedge of spheres .
theorem [ thm : single_gen ] above shows that the reduced homology of the curve complex is a cyclic @xmath26-module . the full `` solomon - tits theorem '' for the mapping class group should also provide the following . * steinberg modules for automorphism groups of free groups .
* let @xmath346 be the free group on @xmath63 generators , and let @xmath347 and @xmath348 denote its automorphism group and outer automorphism group respectively ( see @xcite for a survey of these groups ) .
many of the results on the homological structure of the mapping class group have analogs for @xmath348 and @xmath347 .
for example , bestvina and feighn @xcite have shown that @xmath348 and @xmath347 are virtual duality groups . to date , the module structure of the dualizing modules for ( torsion - free , finite index subgroups of ) these groups has not been studied in depth . in the case of @xmath347 hatcher and vogtmann @xcite have proposed a likely candidate for this dualizing module .
they define a simplicial complex based on the poset of free factors of @xmath346 and show that it has the homotopy type of a wedge of spheres .
they define its reduced homology group to be the steinberg module of @xmath347 , and ask if this module provides the dualizing module for finite index , torsion - free subgroups of @xmath347 .
it is possible that the approach of the current paper might be used to address this question .
w. j. harvey , boundary structure of the modular group , in _ riemann surfaces and related topics :
proceedings of the 1978 stony brook conference ( state univ .
new york , stony brook , n.y . , 1978 ) _
, 245251 , ann . of math .
, 97 , princeton univ .
press , princeton , n.j .
l. mosher , a user s guide to the mapping class group : once punctured surfaces , in _ geometric and computational perspectives on infinite groups ( minneapolis , mn and new brunswick , nj , 1994 ) _ , 101174 ,
soc . , providence , ri . | harer has shown that the mapping class group is a virtual duality group mirroring the work of borel - serre on arithmetic groups in semisimple @xmath0-groups . just as the homology of the rational tits building provides the dualizing module for any torsion - free arithmetic group ,
the homology of the curve complex is the dualizing module for any torsion - free , finite index subgroup of the mapping class group .
the homology of the curve complex was previously known to be an infinitely generated free abelian group , but to date , its structure as a mapping class group module has gone unexplored . in this paper
we give a resolution for the homology of the curve complex as a mapping class group module . from the presentation coming from the last two terms of this resolution
we show that this module is cyclic and give an explicit single generator . as a corollary ,
this generator is a homologically nontrivial sphere in the curve complex . |
among the accretion powered x - ray pulsars , quasi periodic oscillations ( qpo ) were first discovered in the x - ray flux of the transient source exo 2030 + 375 ( angelini et al . 1989 ) . since then
, qpos have been detected in the x - ray power density spectra of almost a dozen accretion - powered pulsars ( psaltis 2004 ) .
qpo centre frequencies as seen in the pulsars vary widely , ranging from 1 mhz to 40 hz . in the models that deal with qpos in x - ray pulsars
, a correlation is expected between the qpo centre and the x - ray flux , which in turn is determined by the mass accretion rate .
hence , intensity dependent variable qpos can provide an interesting insight into the physics of accretion .
however , qpos are in general rare and transient events in x - ray pulsars . to date
, six out of the dozen x - ray binary pulsars ( xbp ) showing qpos are transient sources ( psaltis 2004 ) .
xte j1858 + 034 is one of the transient xbps in which qpos have been detected .
this x - ray binary was discovered with the rxte all sky monitor ( asm ) in 1998 february during a transient outburst ( remillard & levine 1998 ) .
observations made with the proportional counter array ( pca ) of the rxte during the first outburst of the source showed single peaked strong sinusoidal pulsations with a period of @xmath3221.0 s ( takeshima et al .
1998 , paul & rao 1998 ) .
qpos at 110 mhz with about 7% rms fluctuation were discovered from the same observation ( paul & rao 1998 ) .
the energy spectrum was measured to be hard , similar to the spectra of xbps . from the transient nature of this source , its hard x - ray spectrum , and presence of pulsations ,
it was suggested to be a be - x - ray binary ( takeshima et al .
1998 ) which is also supported from the optical observations of h@xmath4 emission line ( reig , kougentakis & papamastorakis 2004 ) .
the source had a second outburst that was first detected with integral in march 2004 and subsequently again in april , 2004 ( molkov et al .
the rxte - asm also detected the outburst and it was followed up with the rxte - pca .
we detected the qpo features again during this second outburst from an observation made on may 14 , 2004 .
subsequently , a series of target of opportunity observation of xte j1858 + 034 were made with rxte during may 1618 to study the evolution of the qpo properties .
we have investigated the evolution of the qpo parameters with time , x - ray flux and x - ray photon energy .
x - ray pulsation properties and the x - ray spectrum were also measured from these observations . in the following sections we present the x - ray observations , analysis and results and a discussion based on the results obtained .
the observations of xte j1858 + 034 reported here were made during may 14 - 18 , 2004 with the pointed mode instruments of the rxte satellite .
the rxte - asm lightcurve of xte j1858 + 034 is shown in figure 1 for the period of march - may 2004 with a bin - size of one and a half days .
the asm light curve clearly shows that during the rxte pointed observations , the source was in the decaying phase of this outburst .
the rxte satellite carries a large area proportional counter array ( pca ) , consisting of five xenon filled proportional counter units ( pcus ) sensitive in the energy range of 260 kev with a total effective area of 6250 cm@xmath1 ( jahoda et al .
it also has a set of high energy crystal scintillation experiment ( hexte ; 15200 kev ; 1600 cm@xmath1 area ) , and a continuously scanning all - sky monitor ( asm ; 210 kev ; 90 cm@xmath1 )
. a full description of the asm detector can be found in bradt et al .
( 1993 ) . here
we have used the rxte - asm long term light curve of xte j1858 + 034 available from the beginning of the rxte mission in 1995 to the present and data from six pointed observations with rxte - pca .
the details of the observations used in the present work have been given in a nutshell in table 1 .
in total there were data from eight orbits of the satellite .
two of the six observations , obs .
b & f had two orbits of data each ( b1 , b2 & f1 , f2 in table 1 ) .
the total good time interval ( gti ) with the pca from these six observations was 18.7 ks .
figure 1 shows the rxte - asm long term light curve ( bin size of one and a half days for clarity ) along with the background subtracted rxte - pca count rates superposed on it . though the beginning of the outburst is not clear , the gradual fall in the long term intensity can be clearly seen from the figure .
from all the eight segments in the six observations , we extracted the light curves , each with a bin size of 0.125 s and then applied barycentre correction on them .
all the individual light curves showed the pulses .
the combined background subtracted light curve of all the six observations binned at the pulse period of @xmath3220 s had an average count rate of @xmath3260 counts s@xmath6 and also had a gradual fall in intensity with time ( see figure 2 ) .
the pulse folding and the @xmath7 maximization method gave a pulse period of [email protected] s within 90@xmath9 confidence interval .
a slow transient pulsar is expected to have a large spin - up rate in a high intensity state like this . however
, a small duty cycle during the five days of observation does not allow us to measure the spin - up rate .
the 2 - 60 kev x - ray light curve , obtained from the pca detectors is shown in figure 2 .
this figure is generated with a binsize of 220 s , same as the spin period , which removes the pulsation related intensity variations . though the overall intensity changed with time , there was no significant intensity variation at a time scale of a few thousand seconds .
we have folded the energy resolved background subtracted light curves with the pulse period mentioned above and the resultant pulse profiles are shown in figure 3 for observation a. the pulse profiles shown for six out of the eight energy bands are single peaked sinusoidal in energy bands up to 25 - 35 kev range with a pulse fraction ( defined as maximum - minimum / maximum ) gradually increasing from 20% at 3 kev to 45% at 30 kev . from figure 3 , we see that pulsations exist upto 45 kev also with a pulse fraction of 35% in the energy band of 35 - 45 kev .
pulsations above 45 kev is not detectable with the pca due to low signal to noise ratio and low detection efficiency .
we have generated the power density spectra ( pds ) from the 2 - 60 kev 0.125 s time resolved light curves from each orbit of the satellite .
the light curves were broken into segments of length 128 s and the pds obtained from each of these segments were averaged to produce the final pds for data from each orbit of the satellite .
a broad qpo feature between 100 mhz and 200 mhz was very prominent in all the pds .
the continuum of the pds in the frequency range of 7 mhz to 2 hz fits well with a model consisting of a power law .
the prominent qpo feature is fitted well with a gaussian model component . in most of the pds ,
there is also an evidence for a second qpo peak , probably a harmonic of the fundamental .
however , we have refrained from quantifying the second peak as it has very low statistics . in figure 4
we have shown the power density spectrum from observation c which has a power - law index of -0.93 and the gaussian , qpo feature centered at 160 mhz .
evolution of the qpo frequency and the associated rms for the eight data segments are shown in figure 5 .
while the qpo frequency decreased with time from 185 mhz on may 14th to about 140 mhz on may 18th , the rms variability associated with the qpo did not have any significant change .
the rms variability in the qpo feature is of the order of 4.0 @xmath9 to 3.0 @xmath9 . in order to investigate the energy dependence of the qpo feature
, we generated pds in the five energy bands of 2 - 5 kev , 5 - 8 kev , 8 - 12 kev , 12 - 18 kev and 18 - 30 kev respectively for obs a. we did not find any evidence of a qpo beyond the 18 - 30 kev band though .
the pds in the different energy bands were more or less identical in shape , and were fitted with the same model as described above .
the rms variation in the qpo feature in different x - ray energy bands obtained from observation a is plotted in figure 6 .
we have obtained a clear energy dependence of the rms variation in the qpos , increasing from about 3% at 3 kev to 6% at 25 kev . the qpo centroid frequency is plotted against the 330 kev rxte - pca flux in figure 7 , which does not show a clear correlation between the two . for spectral analysis
, we used the standard-2 dataset from pca which is stored with 16 s time resolution .
we first extracted the source plus background spectra from all the eight segments of observation .
the background spectra were modeled using the * `` pcabackest '' * package provided by the xte guest observer facility ( gof ) . on the average three pcus
were on , and data from all the pcus were added together to produce the spectrum .
channels corresponding to energy less than 3 kev and greater than 30 kev were ignored because of low signal to noise ratio .
although the background simulation model that we have used takes care of the internal background , emission from an extended source like the galactic ridge or the cosmic x - ray background are not included in it . from a detailed observation of the galactic ridge obtained using the pcas ( valinia & marshall 1998 ) ,
we have estimated that about 1.0 @xmath9 of the observed flux can be accounted for by the galactic ridge emission and a very small fraction by the cosmic x - ray background .
we have , therefore , explicitly incorporated the galactic ridge emission as a sum of raymond - smith plasma and a power - law with appropriate normalizations . for fitting
the pulse - phase averaged x - ray spectrum of the x - ray we have used a high energy cut - off power law along with a gaussian emission line .
the pulsar spectrum in the 330 kev range is found to be very hard with a photon index @xmath31.5 , cut - off energy of @xmath317 kev and e - folding energy of @xmath316 kev along with a neutral absorption with a equivalent hydrogen column density of @xmath310@xmath10 atoms @xmath11 .
the spectrum did not show any significant variation over the five days of observations . during these observations , the x - ray flux in the 3 - 30 kev range varied between 2.9@xmath12 10@xmath13 to 5.5@xmath12 10@xmath13
erg @xmath11 s@xmath6 the x - ray spectrum from one of the observations is shown in figure 7 along with the best fit model and the residuals .
qpos in accretion powered x - ray pulsars are in general explained with either the beat frequency model or the keplerian frequency model . in the former ( alpar & shaham 1985 ) , blobs of matter orbit the neutron star ( ns ) at approximately the keplerian orbital frequency of the inner edge of the accretion disk .
the magnetic field rotates at the frequency of the ns and material inflow to the ns is modulated at the keplerian frequency
. if @xmath14 is the keplerian frequency of the inner edge of the disk and @xmath15 be the spin frequency of the ns , then according to the beat frequency model , the frequency of the qpo ( @xmath16 ) is given by the relation @xmath16 = @xmath14 - @xmath15 . according to the latter model ( van der klis et al .
1987 ) , the inner edge of the accretion disk contains structures that modulate the observed flux by obscuration and gives rise to a qpo at @xmath14 .
a third model by shirakawa & lai ( 2002 ) is known as the magnetic disk precession model .
it attributes the presence of qpos in xbps to the magnetically driven disk warping / precession , concentrated near the inner edge of the disk , at the magnetospheric boundary .
the warping and precession of the disk arises from the magnetic torques which are present due to the interactions between the stellar field and the induced electric currents in the disk . out of the six transient xbps showing qpos , luminosity dependent qpo properties
have been studied extensively for the two sources exo 2030 + 375 ( angelini et al .
1989 ) and 3a 0535 + 262 ( finger , wilson & harmon 1996 ) and both the beat and keplerian frequency models are applicable to a good extent . in several xbps with qpos like her x-1 , cen x-3 , 4u 1626 - 67 , lmc x-4 and smc x-1 , and two transients v 0332 + 52 and 4u 0115 + 63 , the qpo frequency is lower than the pulsation frequency ( psaltis 2004 ) . in these sources ,
the keplerian frequency model is not applicable , since a keplerian rotation frequency of the inner accretion disk that is smaller than the spin frequency of the neutron star will result into centrifugal inhibition of accretion .
again , for v 0332 + 52 , the beat frequency model may also not be viable since the value of the magnetospheric boundary calculated from the qpo properties and from the observed luminosity do not agree ( takeshima et al .
1994 ) . in the low mass xbp gro j1744 - 28 ,
both these models are eliminated since in this source also @xmath17 and a flux change by a factor of @xmath37.5 did not have any corresponding change in qpo frequency .
the magnetic disk precession model ( shirakawa & lai 2002 ) can explain the very low frequency qpos in the xbps , like in 4u 1626 - 67 ( 1 mhz ) and may also explain the lowest frequency qpos detected in some of the other xbps ( her x-1 , lmc x-4 , and 4u 0115 + 63 ) . while this model robustly accounts for qpos with frequencies of the order of 1 mhz , the qpo frequency of xte j1858 + 034 is two orders of magnitude higher and in this model , the production of qpos at higher frequency depends on the details of the physics at the inner edge of the disk .
applicabilty of the magnetic disk precession model is therefore uncertain in this source .
it has already been mentioned that both the beat frequency and the keplerian frequency models of qpos predict a systematic increase of qpo frequency with the mass - accretion rate , which can be investigated by looking at the change of qpo frequency with the x - ray flux .
this assumption is generally applicable to transient xbps , in which the relation between x - ray luminosity and spin - up torque is well understood .
the spin - up torque is in turn related to the mass accretion rate , inner radius of the accretion disk and rotation frequency of the inner disk . as the transient xbps experience a wide range of mass - accretion rates , these are ideal objects to understand the qpo phenomena . in the case of xte
j1858 + 034 during the present outburst , we have detected changes in the qpo frequency over a span of five days .
there is no discernible correlation between the qpo frequency and the instantaneous x - ray flux measured from the pca data , at least during the observation duration ( figure 7 ) .
however , as seen from the rxte - asm lightcurve , there is a gradual decrease in the overall one and a half day averaged x - ray intensity and a gradual decrease in the qpo frequency ( from 185 mhz to 140 mhz , figure 5 ) is also present during these observations . it may be possible that there is a correlation between the qpo frequency with the overall x - ray intensity or mass accretion rate which is not reflected in the short term . we note that in the rxte - pca observations of xte j1858 + 034 made during the 1998 outburst , the source had a count rate ( normalised for per detector unit ) which was about a factor of 2.53.0 times less compared to the observations reported here . a smaller qpo frequency of 110 mhz during the 1998 ourburst (
paul & rao 1998 ) compared to the 140 - 185 mhz frequency during the 2004 outburst supports an inner accretion disk origin of the qpos .
absorption of x - rays by structures in the accretion disk should not be effective in producing hard x - ray qpos and this process should definitely be more effective at lower x - ray energy .
the detection of qpos upto an energy of 30 kev and energy dependence of the rms variability of the qpos ( figure 6 ) shows that the keplerian frequency model is unlikely to be applicable in the case of xte j1858 + 034 .
considering the strong energy dependence of the qpo rms detected both during the 1998 and the 2004 outbursts , and the possibility of a correlation between the qpo frequency with the overall x - ray intensity as discussed above , the beat frequency model seems to be more suitable in the case of xte j1858 + 034 .
spectral parameters of xte j1858 + 034 obtained from the rxte - pca observations during the second outburst in 2004 are similar to those measured during the previous outburst ( paul & rao 1998 ) .
the iron emission line has an equivalent width in the range of 125 ev to 200 ev and the line centre energy is in the range of 6.36.7 kev . with the moderate resolution spectral data from the rxte - pca detectors we can not conclude about the emission line characteristics , but it is likely that the emission line is from neutral or lowly ionized iron in the circumstellar material . the column density measured during the present observations is in the range of ( 7.7 12.3)@xmath12 10@xmath18 atoms @xmath11 , which is higher than the same during the previous outburst . as the absorption column density depends on the mass loss history of the companion star prior to the outburst , a change in the column density is quite likely .
we note that the iron line equivalent width and the absorption column density is consistent with a scenario in which the iron line is produced in a spherical shell of material which also causes the low energy x - ray absorption .
this research has made use of data obtained from the high energy astrophysics science archive research center ( heasarc ) , provided by nasa s goddard space flight center . um and hr would like to acknowledge the kanwal rekhi scholarship of tifr endowment fund for partial financial support .
we also thank parag shah for his help in system administration .
alpar , m.,a . , and j. , shaham , 1985 , _ nature _ , * 316 * , 239 + angelini , l. , stella , l. , & parmar , a.,n . , 1989 , _
apj _ , * 346 * , 906 + bradt , h. , v. , rothschild , r. , e. , @xmath19 swank , j. , h. , 1993 , _ a@xmath19as _ , * 97 * , 355 + finger , m. , h. , wilson , r. , b. , & harmon , b. , a. , 1996 , _ apj _ , * 459 * , 288 + jahoda , k. , swank , j. , h. , giles , a. , b. , et.al . , 1996 , in : siegmund o.h.w .
, gummin m.a .
euv , x - ray and gamma - ray instrumentation for astronomy vii . * spie 2808 , p. 59
* + molkov , s.,v .
et al . , 2004 , _
atel _ , * 274 * , 1 m + paul , b. , and rao , a.,r . , 1998 , _
a&a _ , * 337 * , 815 + psaltis , d. , * astro - ph/0410536 * + remillard , r. , and levine , a. , 1998 , _ iauc 6826 _ + reig , p. , kougentakis , t. , & papamastorakis , g. , 2004 , _ atel _ , * 308 * + shirakawa , a. , & lai , d. , 2002 , _ apj _ , * 565 * , 1134 + takeshima , t. , dotani , t. , mitsuda , k. , & nagase , f. , 1994 , _ apj _ , * 436 * , 871 + takeshima , t. , corbet , r. , h. , d. , marshall , f. , e. , swank , j. , h. , & chakrabarty , d. , 1998 , _ iauc 6826 _ + valinia , a. , and marshall , f.,e . , 1998 , _
apj _ , * 505 * , 134 + van der klis et al . , 1987 , _
apj _ , * 313 * , * l19 * + cccccc + observations & date & start time & stop time & exposure & count rate + & & ( ut ) & ( ut ) & ( s ) & ( s@xmath6 ) + + a & 2004 - 05 - 14 & 11:30 & 12:26 & 3344 & 365 + + b1 & 2004 - 05 - 16 & 11:14 & 11:42 & 1665 & 422 + + b2 & 2004 - 05 - 16 & 13:54 & 14:48 & 3213 & 428 + + c & 2004 - 05 - 16 & 18:47 & 19:21 & 2017 & 399 + + d & 2004 - 05 - 16 & 20:24 & 20:45 & 1225 & 408 + + e & 2004 - 05 - 16 & 23:41 & 00:05 & 1403 & 331 + + f1 & 2004 - 05 - 18 & 13:11 & 14:06 & 3297 & 279 + + f2 & 2004 - 05 - 18 & 14:43 & 15:25 & 2517 & 277 + + | we have investigated the quasi periodic oscillation ( qpo ) properties of the transient accreting x - ray pulsar xte j1858 + 034 during the second outburst of this source in april - may 2004 .
we have used observations made with the proportional counter array ( pca ) of the rossi x - ray timing explorer ( rxte ) during may 14 - 18 2004 , in the declining phase of the outburst .
we detected the presence of low frequency qpos in the frequency range of 140185 mhz in all the rxte - pca observations .
we report evolution of the qpo parameters with the time , x - ray flux , and x - ray photon energy . though a correlation between the qpo centroid frequency and the instantaneous x - ray flux is not very clear from the data , we point out that the qpo frequency and the one day averaged x - ray flux decreased with time during these observations .
we have obtained a clear energy dependence of the rms variation in the qpos , increasing from about 3% at 3 kev to 6% at 25 kev .
the x - ray pulse profile is a single peaked sinusoidal , with pulse fraction increasing from 20% at 3 kev to 45% at 30 kev .
we found that , similar to the previous outburst , the energy spectrum is well fitted with a model consisting of a cut - off power law along with an iron emission line .
@xmath0tata institute of fundamental research , + homi bhabha road , colaba , mumbai 400005 + + @xmath1department of physics , college of science , + m. l. sukhadia university , rajasthan 313001 , india + + @xmath2joint astronomy program , indian institute of science , bangalore-560 012 + + * keywords * stars : pulsar : individual ( xte j1858 + 034 ) x - rays : pulsar |
the equality of krull dimension and transcendence degree for a finitely generated algebra over a field is one of the fundamental results in commutative algebra .
various extensions of this result have appeared in the literature . generalized the result to subalgebras of a finitely generated algebra over a field
. showed that for a finitely generated domain @xmath0 over a field and a prime ideal @xmath3 , the local ring @xmath4 has a subfield @xmath5 such that the transcendence degree of @xmath4 over @xmath5 equals @xmath6 .
some authors , among them and , proposed several notions of a transcendence degree of an algebra over a ring and studied their behavior . in the introduction , wrote that her paper might better be titled : `` why there is no notion of transcendence degree over arbitrary commutative rings . ''
this paper aims to take up the challenge posed by this comment .
we give a new definition of the transcendence degree of an algebra @xmath0 over a ring @xmath1 by calling elements @xmath7 algebraically dependent if they satisfy an equation @xmath8 , where @xmath9 a polynomial with coefficients in @xmath1 such that the trailing coefficient of @xmath9 , with respect to some monomial ordering , is @xmath2 ( see ) . if @xmath1 is a field , this definition coincides with the usual one .
but in other cases , the new transcendence degree behaves in unexpected ways .
for example , the transcendence degree of a ring over itself is `` usually '' not zero . as the main result
, we prove that if @xmath0 is finitely generated and @xmath1 is a noetherian jacobson ring , then the krull dimension of @xmath0 is equal to the transcendence degree of @xmath0 over @xmath1 . in fact , this result extends to the case that @xmath0 is contained in a finitely generated @xmath1-algebra , and if @xmath10 , the hypothesis that @xmath1 is a jacobson ring can be dropped ( see ) . in the case
@xmath10 the result was already proved for the lexicographic monomial ordering by ( see ) .
the paper is organized as follows .
the first section contains the new definition of the transcendence degree , some examples , and the statement of the main result ( ) .
this is proved in the second section .
we also show that the validity of characterizes jacobson rings ( see ) .
the last section is devoted to some applications and to the question whether the transcendence degree depends of the choice of a monomial ordering .
we conjecture that is does not ( conjecture [ conjecture ] ) , and prove some special cases ( see ) .
this work was inspired by reading the above - mentioned paper of , who characterized the krull dimension by certain types of identities . interpreting this in terms of the lexicographic monomial ordering
led to the new definition of the transcendence degree and prompted the questions to what extent this depends on the choice of the monomial ordering , and whether one can also prove a `` relative '' version involving the transcendence degree over a subring . in his bachelor thesis [ ] ,
christoph studied ( among other things ) the first question and found no example where the transcendence degree depends on the monomial ordering .
i wish to thank peter heinig for bringing s article to my attention .
all rings in this paper are assumed to be commutative with an identity element @xmath2 . if @xmath1 is a ring , an @xmath1-algebra is a ring @xmath0 together with a ring homomorphism @xmath11 .
we call an @xmath1-algebra if it is a subalgebra of a finitely generated @xmath1-algebra .
by @xmath12 we will always mean the krull dimension of @xmath1 .
we follow the convention that the zero ring @xmath13 has krull dimension @xmath14
. it will be convenient to work with the polynomial ring @xmath15 $ ] with infinitely many indeterminates over a ring @xmath1 , and to understand a as a total ordering `` @xmath16 '' on the set of monomials of @xmath15 $ ] such that the conditions @xmath17 and @xmath18 hold for all monomials @xmath19 , and @xmath20 with @xmath21 .
clearly any monomial ordering on a polynomial ring @xmath22 $ ] with finitely many indeterminates can be extended to a monomial ordering in the above sense .
the following notions of algebraic dependence and transcendence degree over a ring generalize the corresponding notions over a field .
[ dtrdeg ] let @xmath1 be a ring . 1 .
let `` @xmath16 '' be a monomial ordering .
a nonzero polynomial @xmath23 $ ] is called with respect to `` @xmath16 '' if its trailing coefficient ( i.e. , the coefficient of the least monomial having nonzero coefficient ) is @xmath2 . + a polynomial
is called if there exists a monomial ordering with respect to which it is submonic .
2 . let @xmath0 be an @xmath1-algebra .
elements @xmath7 are called over @xmath1 if there exists a submonic polynomial @xmath24 $ ] such that @xmath25 .
( of course the homomorphism @xmath11 is applied to the coefficients of @xmath9 before evaluating at @xmath26 . ) otherwise , @xmath26 are called over @xmath1 .
+ we can also restrict @xmath9 to be submonic with respect to a specified monomial ordering
@xmath16 , in which case we speak of algebraic dependence or independence with respect to
@xmath16 .
3 . for an @xmath1-algebra @xmath0 ,
the of @xmath0 over @xmath1 is defined as @xmath27 n \in \nn \mid \text{there exist } \
a_1 \upto a_n \in a \
\text{that are algebraically independent over r}\right\}. \end{gathered}\ ] ] if every @xmath28 is algebraically dependent over @xmath1 , we set @xmath29 in the case @xmath30 and @xmath31 in the case @xmath32 .
we write @xmath33 for the transcendence degree of @xmath1 over itself . + if
@xmath16 is a monomial ordering , we define @xmath34 by requiring algebraic independence with respect to @xmath16 .
[ extrdeg ] 1 .
[ extrdeg1 ] if @xmath1 is an integral domain , then the elements of @xmath1 that are algebraically dependent over @xmath1 are @xmath35 and the units of @xmath1 .
[ extrdeg2 ] if @xmath1 is a nonzero finite ring , then @xmath36 since for each @xmath37 there exist nonnegative integers @xmath38 such that @xmath39 .
[ extrdeg3 ] the following example shows that the notion of algebraic dependence with respect to a monomial ordering depends on the chosen monomial ordering .
let @xmath40 $ ] be a polynomial ring in two indeterminates and let @xmath41 and @xmath42 .
the relation @xmath43 tells us that @xmath44 are algebraically dependent over @xmath1 with respect to the lexicographic monomial ordering with @xmath45 . on the other hand , algebraic dependence over @xmath1 with respect to the lexicographic ordering with @xmath46
would mean that there exist @xmath47 such that @xmath48 lies in the @xmath1-ideal @xmath49 which is not the case .
[ extrdeg4 ] we consider @xmath50 and claim that @xmath51 . since @xmath52 has nonzero elements which are not units , we have @xmath53 . we need to show that all pairs of integers @xmath54 are algebraically dependent over @xmath52 .
we may assume @xmath55 and @xmath56 to be nonzero and write @xmath57 where the @xmath58 are pairwise distinct prime numbers and @xmath59 .
choose @xmath60 such that @xmath61 for all @xmath62 with @xmath63 .
then @xmath64 so there exist @xmath65 such that @xmath66 . since @xmath67 is submonic ( with respect to the lexicographic ordering with @xmath45 ) , this shows that @xmath44 are algebraically dependent .
+ clearly this argument carries over to any principal ideal domain that is not a field .
it is remarkable that although the transcendence degree is an algebraic invariant , the above calculation has a distinctly arithmetic flavor .
it becomes clear from that sums of algebraic elements need not be algebraic , and from that the transcendence degree does not behave additively for towers of ring extensions .
proved that for a ring @xmath1 and an integer @xmath68 , the inequality @xmath69 holds if and only if for all @xmath70 there exist @xmath71 such that @xmath72 ( also see ) . using and writing @xmath73 for the lexicographic ordering with @xmath74 for all @xmath62
, we can reformulate this result as follows .
[ tcl ] if @xmath1 is a ring , then @xmath75 the following is the main result of this paper .
[ tmain ] 1 .
[ tmaina ] if @xmath1 is a noetherian ring , then @xmath76 2 .
[ tmainb ] if @xmath1 is a noetherian jacobson ring and @xmath0 is a subfinite @xmath1-algebra , then @xmath77 3 .
[ tmainc ] if @xmath1 and @xmath0 are as in and @xmath0 is noetherian , then @xmath78 parts and generalize the classical result that the krull dimension of a finitely generated algebra over a field is equal to its transcendence degree .
the proof of is subdivided into various steps , which contain results that are themselves of some interest . recall that in a noetherian ring @xmath1 with @xmath79 there exist elements @xmath26 and a prime ideal @xmath80 of height @xmath81 that lies minimally over @xmath82 ( see [ , theorem 7.8 ] ) .
therefore the following theorem implies the inequality @xmath83 [ tlower ] let @xmath1 be a noetherian ring
. if @xmath70 are elements such that @xmath1 has a prime ideal of height @xmath81 lying minimally over @xmath82 , then the @xmath84 are algebraically independent over @xmath1 .
let @xmath85 be a prime ideal of height @xmath81 lying minimally over @xmath82 .
clearly it suffices to show that the images of the @xmath84 in the localization @xmath86 are algebraically independent . substituting @xmath1 by @xmath86
, we may therefore assume that @xmath1 is a local ring and @xmath26 form a system of parameters .
with @xmath87 , this implies that for all @xmath88 the module @xmath89 has finite length , and for sufficiently large @xmath90 this length is given by a polynomial of degree @xmath81 in @xmath90 ( see ) . by way of contradiction , assume that @xmath26 are algebraically dependent .
this means that there exists a monomial @xmath91 and further monomials @xmath92 that are greater than @xmath93 with respect to some monomial ordering such that @xmath94 where the @xmath95 are elements of @xmath1 . by
[ , exercise 9.2(b ) ] , there exist positive integers @xmath96 such that @xmath97 holds for @xmath98 .
( the exercise uses the so - called convex cone of the monomial ordering , and a solution is provided in the book . ) writing @xmath99 and @xmath100 , we can express the inequalities as @xmath101 for @xmath88 we define the ideal @xmath102 with @xmath103 we have @xmath104 for all @xmath90 , so for @xmath90 sufficiently large , the length of @xmath105 is bounded below by a polynomial of degree @xmath81 in @xmath90 .
clearly @xmath106 .
we write @xmath107 and claim that @xmath108 is generated as an @xmath1-module by @xmath109 elements .
in fact , @xmath108 is clearly generated by all @xmath110 with @xmath111 .
but if @xmath112 has the form @xmath113 with @xmath114 , then @xmath115 since @xmath116 so we can exclude such a product @xmath110 from our set of generators and are left with @xmath117 generators , as claimed .
the definition of @xmath118 implies that @xmath119 , so @xmath108 is an @xmath120-module . with @xmath121
, we obtain @xmath122 in the ring @xmath123 $ ] of formal power series over @xmath52 , we have @xmath124 where @xmath125 and @xmath126 $ ] .
it follows that @xmath127 is bounded above by a polynomial of degree at most @xmath128 in @xmath90 , contradicting the fact that @xmath129 is bounded below by a polynomial of degree @xmath81 for large @xmath90 .
this contradiction finishes the proof .
the converse statement of may fail : for example , the only prime ideal lying minimally over the class of @xmath130 in @xmath131/(x \cdot y)$ ] ( with @xmath132 a field and @xmath130 and @xmath133 indeterminates ) has height @xmath35 , but the class of @xmath130 is nevertheless algebraically independent over @xmath1 . for any monomial ordering `` @xmath16 '' , it follows directly from that @xmath134 . applying this to the lexicographic ordering and using and yields .
let @xmath0 be an algebra over a ring @xmath1 and let
@xmath16 be a monomial ordering .
then the inequalities @xmath135 follow directly from .
the next goal is to prove @xmath136 in the case that @xmath1 is a noetherian jacobson ring and @xmath0 is finitely generated over @xmath1 . to achieve this goal , we need four lemmas .
the proof of part of the following lemma was shown to me by viet - trung ngo .
[ lngo ] let @xmath1 be a noetherian ring and @xmath85 . 1 .
[ lngoa ] if @xmath137 such that @xmath138 and @xmath139 has height at least 2 , then there exist infinitely many prime ideals @xmath140 such that @xmath141 and @xmath142 .
[ lngob ] if @xmath143 is an infinite set of prime ideals such that every @xmath144 satisfies @xmath145 and @xmath142 , then @xmath146 . 1 . by factoring out @xmath80 and localizing at @xmath147
we may assume that @xmath1 is a local domain with maximal ideal @xmath147 , and @xmath148 .
for @xmath149 there exists @xmath150 which is minimal over @xmath151 . by the principal ideal theorem , @xmath152 has height one .
so @xmath153 if there existed only finitely many @xmath140 of height one , it would follow by the prime avoidance lemma ( see [ , lemma 7.7 ] ) that @xmath147 is contained in one of them , contradicting the hypothesis @xmath154 2
. clearly @xmath155 .
if @xmath156 , then every @xmath157 would be a minimal prime ideal over @xmath158 and so @xmath159 would be finite ( see [ , corollary 3.14(d ) ] ) .
[ ljacobson ] a noetherian ring @xmath1 is a jacobson ring if and only if for every @xmath160 with @xmath161 there exist infinitely many maximal ideals @xmath162 with @xmath163 .
we prove that the negations of both statements are equivalent .
first assume that @xmath1 is not jacobson .
choose @xmath85 to be maximal among the prime ideals that are not intersections of maximal ideals .
then @xmath80 itself is not maximal , so @xmath164 . on the other hand , it is impossible that @xmath165 , since by that would imply that @xmath80 is the intersection of strictly larger prime ideals and therefore ( by its maximality ) of maximal ideals .
so @xmath161 .
therefore every maximal ideal @xmath166 containing @xmath80 satisfies @xmath167 , so by only finitely many such @xmath166 exist .
conversely , assume that @xmath1 has a prime ideal @xmath80 with @xmath168 such that only finitely many maximal ideals , say @xmath169 , contain @xmath80 .
if @xmath170 , then @xmath80 would contain at least one of the @xmath171 , so @xmath80 would be maximal . since this is not the case , @xmath1 is not jacobson .
the following lemma may be surprising since it does not require the ring @xmath1 to be jacobson .
[ lboundary ] let @xmath55 be an element of a noetherian ring @xmath1 and set @xmath172 then the localization @xmath173 is a jacobson ring .
we will use the criterion from and the inclusion - preserving bijection between the prime ideals in @xmath174 and the prime ideals @xmath85 satisfying @xmath175 ( see [ , theorem 6.5 ] ) .
let @xmath85 with @xmath176 such that @xmath177 .
then there exists @xmath178 with @xmath179 and @xmath180 .
the latter condition implies @xmath181 and @xmath182 , so there exists @xmath137 such that @xmath183 .
it follows that @xmath184 , so by , the set @xmath185 is infinite .
assume that the subset @xmath186 is also infinite .
then would imply @xmath187 , so @xmath188 , contradicting @xmath189 .
we conclude that @xmath190 is infinite .
let @xmath191 .
then @xmath192 and @xmath193 imply that @xmath194 for every @xmath195 , so @xmath196 .
therefore @xmath197 , and we also have @xmath198 . since @xmath177 , this implies that @xmath199 is a maximal ideal .
so by , the infinity of @xmath190 implies that @xmath200 is a jacobson ring .
the localization @xmath173 from was also used by .
they called it the _ boundary _ of @xmath55 in @xmath1 .
[ lfield ] let @xmath1 be a jacobson ring and let @xmath0 be a finitely generated @xmath1-algebra that is a field .
then the kernel of the map @xmath11 is a maximal ideal .
we may assume that the map @xmath11 is injective , so we may view @xmath1 as a subring of @xmath0 .
we need to show that @xmath1 is a field .
since @xmath0 is finitely generated as an algebra over @xmath201 ( the field of fractions ) , it is algebraic over @xmath132 ( see [ , lemma 1.1(b ) ] ) .
so @xmath0 is a finite field extension of @xmath132 , and choosing a basis yields a map @xmath202 sending each @xmath28 to the representation matrix of the linear map given by multiplication by @xmath55 .
let @xmath26 be generators of @xmath0 as an @xmath1-algebra and choose @xmath203 to be a common denominator of all matrix entries of all @xmath204
. then @xmath205^{n \times n}$ ] , and since @xmath206 is a homomorphism of @xmath1-algebras , its image is contained in @xmath207^{n \times n}$ ] .
if @xmath208 , then @xmath209 , so @xmath210^{n \times n}.\ ] ] this implies @xmath211 $ ] , so @xmath207 $ ] is a field . from this
it follows by that @xmath1 is a field .
as announced , we can now prove the upper bound for the transcendence degree .
as above , @xmath73 stands for the lexicographic ordering with @xmath74 for all @xmath62 .
[ pupper ] 1 .
[ puppera ] if @xmath1 is a ring , then @xmath212 2 .
[ pupperb ] if @xmath0 is a finitely generated algebra over a noetherian jacobson ring @xmath1 , then @xmath213 part is contained in s result ( ) .
we include a proof of this part for the reader s convenience .
we prove both parts simultaneously , setting @xmath10 in the case of part .
it is clear that without loss of generality we may assume the map @xmath11 to be injective , so we may view @xmath1 as a subring of @xmath0 .
we may also assume @xmath30 and @xmath214 .
we use induction on @xmath215 .
let @xmath7 .
consider the multiplicative set @xmath216 \
\text{is submonic}\right\ } \subseteq a\ ] ] and set @xmath217 . by way of contradiction , assume @xmath218 .
using the correspondence between prime ideals in @xmath219 and prime ideals in @xmath0 that do not intersect with @xmath220 , we obtain a chain @xmath221 with @xmath222 and @xmath223 .
since @xmath224 , @xmath225 must be a field . in the case of part
, it follows by that @xmath226 is a maximal ideal . in the case of part , this is also true since @xmath227 .
so @xmath225 is an algebraic field extension of @xmath228 ( see [ , lemma 1.1(b ) ] ) .
from @xmath229 we conclude that @xmath230 is invertible .
so there exists @xmath231 $ ] such that @xmath232 . but
@xmath233 is submonic , so @xmath234 , contradicting @xmath229 .
we conclude that @xmath235 . if @xmath236 ( which must happen if @xmath237 ) , then @xmath238 , so @xmath239 satisfies a submonic equation and therefore @xmath26 are algebraically dependent with respect to @xmath73 .
having dealt with this case , we may assume @xmath240 .
clearly @xmath241 \subseteq a'$ ] . in the case of part
we have @xmath242 . in the case of part , @xmath219 is finitely generated as an @xmath243-algebra , and by , @xmath243 is a jacobson ring .
so in both cases the induction hypothesis tells us that @xmath244 satisfy a polynomial @xmath245 $ ] that is submonic with respect to @xmath73 . multiplying the coefficients of @xmath246 by a suitable element from @xmath220
, we obtain @xmath247[x_1 \upto x_{n-1}]$ ] whose trailing coefficient @xmath248 $ ] lies @xmath220 such that @xmath249 . by replacing every coefficient
@xmath250 $ ] of @xmath251 by a @xmath252 $ ] with @xmath253 , we obtain @xmath24 $ ] with @xmath254 .
since @xmath255 , we may choose the coefficient @xmath256 $ ] of @xmath9 to be submonic .
the trailing coefficient of @xmath9 ( with respect to @xmath73 ) is equal to the trailing coefficient of @xmath257 , which is @xmath2 .
therefore @xmath9 is submonic with respect to @xmath73 .
we conclude that @xmath26 are algebraically dependent with respect to @xmath73 .
since they were chosen as arbitrary elements of @xmath0 , this shows that @xmath258 .
we can now finish the proof of . and imply @xmath259 if @xmath0 is noetherian , and yield the finer inequality @xmath260
so for the proof of and it suffices to show that @xmath136 .
we may assume that @xmath261 and write @xmath262 . by hypothesis
, @xmath0 is a subalgebra of a finitely generated @xmath1-algebra @xmath263 .
let @xmath264 be the minimal prime ideals of @xmath263 , and assume that we can show @xmath265 for all @xmath62 .
then for @xmath266 there exist polynomials @xmath267 $ ] that are submonic with respect to @xmath73 such that @xmath268 .
so @xmath269 lies in the nilradical of @xmath263 , hence there exists @xmath270 such that @xmath271 satisfy the polynomial @xmath272 , which is also submonic with respect to @xmath73 .
this shows that we may assume @xmath263 to be an integral domain . by ( or
[ , exercise 10.3 ] ) , there exists a nonzero @xmath28 such that @xmath273 $ ] is finitely generated as an @xmath1-algebra .
so we may apply and get @xmath274:r\right ) \le \dim\left(a[a^{-1}]\right)$ ] .
we obtain @xmath275:r\right ) \le \dim\left(a[a^{-1}]\right ) \le \dim(a),\ ] ] where the first inequality follows directly from , and the last since @xmath273 $ ] is a localization of @xmath0 .
this completes the proof .
[ rconverse ] 1 .
[ rconversea ] the hypothesis that @xmath1 be a jacobson ring can not be dropped from and .
in fact , the validity of and characterizes jacobson rings in the following sense : if @xmath1 is a non - jacobson ring , then by , @xmath1 has a nonmaximal prime ideal @xmath80 such @xmath276 contains a nonzero element @xmath56 for which @xmath277 $ ] is a field .
so @xmath278 , but @xmath56 is not a unit in @xmath200 , so yields @xmath279 since @xmath0 is a finitely generated @xmath1-algebra , the assertions of and fail for @xmath1 .
[ rconverseb ] neither can the hypothesis that @xmath0 is subfinite be dropped . in fact , if @xmath1 is any nonzero ring , we can choose a maximal ideal @xmath166 of @xmath1 and form the polynomial ring @xmath280 $ ] and @xmath281 .
it seems to be rare that helps to compute the dimension of rings .
in fact , the transcendence degree seems to be the less accessible quantity in most cases , so knowledge of the dimension provides information about the structure of the ring that is encoded in the transcendence degree . contains such an instance .
here is a further example .
[ exnumber ] let @xmath55 and @xmath56 be two nonzero algebraic numbers ( i.e. , elements of an algebraic closure of @xmath283 )
. there exists @xmath284 such that @xmath55 and @xmath56 are integral over @xmath285 $ ] , so @xmath286 $ ] has krull dimension @xmath2 . by , @xmath287 ,
so @xmath44 satisfy a polynomial @xmath288 $ ] that is submonic with respect to @xmath73 .
if @xmath289 is the trailing monomial of @xmath9 , then all monomials of @xmath9 are divisible by @xmath290 , so we may assume @xmath291 .
we obtain @xmath292 with @xmath293 $ ] polynomials .
it is not so clear how the existence of such a relation follows directly from the properties of algebraic numbers .
[ exinvariant ] let @xmath1 be a ring that is finitely generated as a @xmath52-algebra , @xmath296 a group of automorphisms of @xmath1 and @xmath297 a subgroup .
then [ cab ] tells us that @xmath298 even though the invariant rings need not be finitely generated ( see ) . by part of the following ,
we need to show that if @xmath302 for an integer @xmath81 , then @xmath303 .
( we will invoke numerous times during this proof without always mentioning it . ) by , there exists an open affine subset @xmath220 of @xmath299 such that @xmath304 . by
, @xmath305 is the closure of @xmath306 in @xmath220 .
it is also clear that @xmath307 is locally closed in @xmath220 . moreover , if @xmath308 , then @xmath0 is finitely generated as an @xmath1-algebra ( see ) .
so by substituting @xmath299 by @xmath220 and @xmath300 by @xmath307 , we may assume that @xmath309 with @xmath0 a finitely generated @xmath1-algebra .
since @xmath310 with @xmath263 a quotient ring of @xmath0 , we may substitute @xmath0 by @xmath263 and assume that @xmath300 is dense in @xmath299 .
@xmath299 has an irreducible component @xmath311 with @xmath312 . since @xmath313 is nonempty , @xmath313 is dense in @xmath311 .
so by substituting @xmath299 by @xmath311 and factoring out the nilradical of @xmath0 , we may assume @xmath0 to be an integral domain .
since @xmath314 is nonempty and open , there exists a nonzero ideal @xmath315 such that @xmath316 choose @xmath317 .
then @xmath318 so it suffices to show that @xmath319 .
but @xmath320 is homeomorphic to @xmath321\right)$ ] ( see [ , theorem 6.5 and exercise 6.5 ] ) . since @xmath0 is a jacobson ring ( see ) , [ cab ] yields @xmath322\right ) \ge \dim(a),\ ] ] so @xmath323\right ) \ge n$ ] . 1 .
[ ltopa ] @xmath300 is irreducible if and only if its closure @xmath324 is irreducible .
[ ltopb ] the inequality @xmath325 holds for the dimensions of @xmath299 and @xmath300 as topological spaces .
[ ltopc ] if @xmath326 is an open subset , then @xmath327 is the closure of @xmath307 in @xmath220 .
[ ltopd ] if @xmath328 is a set of open subsets of @xmath299 such that @xmath329 and if @xmath330 holds for an integer @xmath81 , then there exists @xmath331 such that @xmath332 . under the hypothesis of
, we may assume @xmath337 since @xmath338 is an open covering of @xmath300 .
there exists a chain @xmath339 of closed irreducible subsets of @xmath299 .
we can choose @xmath331 with @xmath340 .
then @xmath341 , and the @xmath342 form an ascending chain of closed subsets in @xmath220 .
since @xmath342 is nonempty and open in @xmath311 , its closure equals @xmath311 .
this implies that the inclusions between the @xmath342 are strict , and , by , that the @xmath342 are irreducible .
therefore @xmath343 .
jacobson rings are characterized by the validity of .
indeed , every non - jacobson ring @xmath1 has a nonmaximal prime ideal @xmath80 such @xmath276 contains a nonzero element @xmath56 for which @xmath344 $ ] is a field ( see ) .
then with @xmath345 und @xmath346 we have @xmath347 , but @xmath348 and @xmath349 .
it seems odd that the lexicographic ordering plays such a special role in theorems [ tcl ] and [ tmain ] .
can it be substituted by other monomial orderings ? below gives answers in some special cases .
we need some preparations for its proof .
let `` @xmath16 '' be a weight - graded monomial ordering .
consider the set @xmath356 of all trailing monomials of polynomials @xmath357 $ ] with @xmath254 .
it is easy to see that @xmath358 is an ideal and @xmath359 .
we need to show that @xmath360 .
suppose that we can show that for all maximal ideals @xmath162 with @xmath361 , the elements @xmath362 are algebraically dependent with respect to `` @xmath16 '' .
then @xmath363 is nonempty , so @xmath360 .
this shows that we can assume that @xmath1 is a local ring and @xmath158 is contained in its maximal ideal . by hypothesis ,
@xmath364 is artinian , so by , the length of @xmath365 is given by a polynomial of degree @xmath69 for @xmath90 large enough . since multiplying all weights @xmath366 by the same positive constant
does not change the monomial ordering , we may assume that @xmath367 for @xmath368 .
for @xmath369 write @xmath370 , and for @xmath88 set @xmath371 then @xmath372 , so the length of @xmath373 is bounded above by a polynomial of degree @xmath374 .
take @xmath88 and consider the set @xmath375 by way of contradiction , assume that the @xmath84 are algebraically independent with respect to `` @xmath16 '' . for @xmath376 , this assumption and the fact that `` @xmath16 '' is weight - graded imply @xmath377 so ordering @xmath378 according to `` @xmath16 '' yields a chain of length @xmath379 of ideals between @xmath380 and @xmath118 .
this implies @xmath381 , so @xmath382 . with @xmath383
, we obtain @xmath384 contradicting the fact that the length of @xmath373 is bounded above by a polynomial of degree @xmath374 .
this contradiction finishes the proof . 1
. [ torderinga ] if @xmath0 is an algebra over a ring @xmath1 that contains a field @xmath132 such that @xmath0 is a subfinite @xmath132-algebra , then @xmath385 2 .
[ torderingb ] if @xmath386 with @xmath263 a finitely generated algebra over a field and @xmath387 , then @xmath388 3 .
[ torderingc ] if @xmath389 , then @xmath388 4 .
[ torderingd ] if @xmath389 and @xmath0 is an subfinite algebra over a noetherian jacobson ring @xmath1 , then @xmath385 by we may assume @xmath30 . by
and we have @xmath390 where we set @xmath227 in the case of and .
so we may assume @xmath391 and need to show that all @xmath7 are algebraically dependent over @xmath1 with respect to `` @xmath16 '' . 1 . by ,
the @xmath84 are algebraically dependent over @xmath132 .
since @xmath132 is a field , the algebraic dependence is with respect to `` @xmath16 '' , and since @xmath392 , it is over @xmath1 .
2 . let @xmath393 be the minimal prime ideals in @xmath0 .
for each @xmath62 , @xmath394 is the localization of a finitely generated domain over a field at a prime ideal , so by , @xmath394 contains a field @xmath395 such that @xmath396 .
therefore @xmath397 so we obtain polynomials @xmath398 $ ] that are submonic with respect to `` @xmath16 '' such that @xmath399 .
a power of the product of the @xmath400 yields a submonic equation for @xmath26 .
3 . if @xmath278 , the algebraic dependence of the @xmath84 follows from since every monomial ordering restricts to the lexicographic ordering on @xmath401 $ ] .
so we may assume @xmath402 .
it follows by that the restriction of `` @xmath16 '' to @xmath403 $ ] is either a lexicographic ordering or weight - graded . by
we may assume the latter .
+ let @xmath404 be the minimal prime ideals of @xmath0 satisfying @xmath405 and set @xmath406 ( with @xmath407 in the case @xmath408 ) .
every @xmath3 with @xmath409 contains at least one of the @xmath410 , and if @xmath411 , then @xmath412 , so @xmath413 .
this shows that @xmath414 . applying to @xmath415 yields @xmath416
$ ] that is submonic with respect to `` @xmath16 '' such that @xmath417 . by multiplying @xmath9 by @xmath418
, we may assume @xmath419 , so by the definition of @xmath158 , @xmath420 lies in every minimal prime ideal of @xmath0
. therefore a suitable power of @xmath9 yields a submonic equation for @xmath421 .
4 . by [ cab ]
, the subalgebra @xmath422 generated by the @xmath84 has dimension at most @xmath2 . by ,
the @xmath84 satisfy an equation @xmath423 $ ] that is submonic with respect to `` @xmath16 '' .
obtain @xmath424 $ ] from @xmath9 by replacing every coefficient @xmath425 of @xmath9 be a @xmath426 $ ] with @xmath427 , where the trailing coefficient of @xmath9 is replaced by @xmath428 .
it follows that @xmath251 is submonic with respect to `` @xmath16 '' and @xmath429 . in view of , a candidate that comes to mind for a ring @xmath1 such that @xmath430 for some monomial ordering `` @xmath16 '' is the polynomial ring @xmath431 $ ] . using a short program written in magma [ ] ,
i tested millions of randomly selected triples of polynomials from @xmath431 $ ] and verified that they were all algebraically dependent with respect to the graded reverse lexicographic ordering , even over the subring @xmath52 .
this prompts the following conjecture : | for a finitely generated algebra over a field , the transcendence degree is known to be equal to the krull dimension .
the aim of this paper is to generalize this result to algebras over rings .
a new definition of the transcendence degree of an algebra @xmath0 over a ring @xmath1 is given by calling elements of @xmath0 algebraically dependent if they satisfy an algebraic equation over @xmath1 whose trailing coefficient , with respect to some monomial ordering , is @xmath2 .
the main result is that for a finitely generated algebra over a noetherian jacobson ring , the transcendence degree is equal to the krull dimension . |
in its basic features , our implementation @xcite of the coherent potential approximation ( cpa ) follows the formulation @xcite based on iterating the `` coherent interactor '' matrix @xmath82 for sites with chemical and/or spin disorder , which is defined as @xmath83 where @xmath11 is the complex energy , @xmath84 is the coherent potential and @xmath85 the on - site auxiliary green s function .
the latter is computed as the on - site block of the @xmath86-integrated crystal green s function @xmath87^{-1}$ ] , where @xmath88 and @xmath89 denote basis sites within the unit cell , and @xmath90 is the structure constant matrix of the tight - binding linear muffin - tin orbital ( lmto ) method .
the coherent potential is in turn calculated from the lmto potential parameters @xmath91 and the @xmath82 matrix : @xmath92^{-1 } + \omega \label{p}\ ] ] where @xmath93 is the probability to find an atom of the component @xmath94 on the given site . in the vector disordered local moment method ( vdlm ) ,
different directions of the local magnetic moments are treated as different alloy components . the component index @xmath94 in eq .
( [ p ] ) then combines the index of the chemical component @xmath15 and the orientation of the local moment @xmath95 , and the probabilities become @xmath96 , where @xmath97 is the concentration of the component @xmath15 on the given site , and @xmath98 is the distribution function for the spin orientations . the lmto potential parameter matrix for each spin direction is initially computed in the reference frame in which the polar axis is aligned with the direction @xmath99 of the total effective magnetic field for this direction ( more on this below ) .
let us denoting this matrix @xmath100 . to obtain @xmath91 in eq .
( [ p ] ) , we need to rotate @xmath100 to the global reference frame : @xmath101 where @xmath102 is the rotation operator that rotates the @xmath103 vector to @xmath99 . in this work spin - orbit coupling
is neglected .
therefore , @xmath100 is spin - diagonal and the rotation @xmath102 only affects the spin indices .
if spin - orbit coupling is included @xcite , @xmath102 acts on both spin and orbital indices and is generated by the total angular momentum operator @xmath104 .
the summation over @xmath94 in eq .
( [ p ] ) involves a sum over chemical components and an integration over the sphere for each component treated within the vdlm scheme . for our present purposes the vdlm scheme
is implemented for partially ordered magnetic states with a global axial symmetry and without spin - orbit coupling .
this symmetry implies that the lmto potential matrices @xmath100 and the spin distribution functions @xmath98 depend only on the chemical identity @xmath15 and the polar angle @xmath105 with respect to the magnetic symmetry axis . in this special case
the integral over the azimuthal angle in eq .
[ p ] has a very simple effect : it eliminates all spin off - diagonal elements and leaves spin - diagonal elements ( which do not depend on @xmath106 ) unchanged . the remaining integration over the polar angle
@xmath105 is discretized using the 16-point gauss - legendre quadrature .
thus , formally we are dealing with 32 cpa components on the fe / mn sites corresponding to 16 values of the polar angle for fe and mn .
the spin distribution functions for partially ordered vdlm states are assumed to have the curie - weiss form @xmath107 .
the factors @xmath108 are treated as variational parameters in the free energy minimization .
in addition to vdlm states , in the construction of the energy functional we also use cpa calculations for the @xmath12 phases ( f / g , g / c , and f / c ) with definite ( non - random ) spin directions .
these calculations are essential to capture the higher - order interactions between the order parameters corresponding to different @xmath31 vectors ; indeed , axially - symmetric vdlm calculations represent partially ordered states with only one ordering @xmath31 vector . in these calculations , the sum over @xmath94 in eq .
( [ p ] ) has only two terms for fe and mn with their spins pointing in certain directions within the @xmath109 plane ( as is sufficient for the @xmath12 states ) .
of course , the spin off - diagonal terms are retained in eq .
( [ p ] ) , whereby @xmath110 , @xmath82 , and @xmath85 also become spin off - diagonal .
the magnetic configuration in a @xmath12 phase can be written as @xmath111 with orthogonal order parameters ( @xmath112 ) , where @xmath113 is the coordinate of site @xmath14 .
because of the orthogonality , there are four real order parameters in a @xmath12 phase , which can be simply denoted as @xmath114 and @xmath115 .
@xmath116 represents the averaged local moment of an atom of component @xmath15 appearing at lattice site @xmath14 .
the site magnetizations are defined as @xmath21 , i. e. the dependence of the local moment on its orientation is neglected _ in this definition_. this choice does impair the accuracy of the calculations , because the fitting of the calculated total energies and the calculation of the magnetic entropy are consistent with this specific definition of the order parameters .
the relevant ordering wave vectors are @xmath6 , @xmath7 , and @xmath8 in units of @xmath9 ( or @xmath10 for the @xmath11 component ) .
all these wave vectors are commensurate with a tetragonal @xmath117 unit cell , which is therefore used for the calculation of the total energies in all magnetic phases .
( the use of the same unit cell improves the accuracy of the calculations . )
the magnetic structures of the considered @xmath12 states are shown in fig .
[ structures ] along with the @xmath118-type and @xmath119-type orderings ( the ferromagnetic state is not shown ) . using the symmetries
@xmath120 and @xmath121 ( applied to all @xmath15 at the same time ) , the accessible space of ( @xmath34 is reduced to the range @xmath35 , @xmath36 with additional @xmath37 symmetry at @xmath38 and @xmath39 symmetry at @xmath40 .
this irreducible domain is covered by a uniform square mesh of 38 inequivalent points .
the lmto charges are computed in the usual way from the conditionally - averaged physical green s function @xcite @xmath122 in the reference frame in which the polar axis is aligned with the prescribed spin direction @xmath123 . here
@xmath124 and @xmath125 are the spin - diagonal potential parameters related to @xmath100 , and @xmath126^{-1}$ ] .
we emphasize that the lmto charges and potentials are different for different orientations of the local moment , and self - consistency is achieved independently for all of them . since the coherent interactor @xmath82 , which includes the magnetic coupling to the rest of the crystal , and @xmath91 , describing the local exchange - correlation field , are diagonal in different reference frames , the magnetic moment obtained from @xmath127 deviates from the direction @xmath99 of the total effective magnetic field .
if the transverse component of this output magnetic moment is discarded , the obtained magnetic state is not self - consistent in the sense of the density functional theory ( dft ) . in order to remedy this problem ,
constraining transverse magnetic fields need to be introduced @xcite .
we implement these fields in the form suggested in ref . .
specifically , the constraining field for each atom is assumed to have the same radial dependence as the exchange - correlation field on the same atom .
the total effective magnetic field is then collinear within the atomic sphere , oriented at an angle @xmath128 to the exchange - correlation field , and rescaled by a factor @xmath129 @xcite . at each iteration toward self - consistency , the misalignment angle @xmath128
is adjusted to eliminate the deviation of the magnetic moment from the prescribed direction @xmath123 .
thus , at self - consistency , the local moment is collinear with the exchange - correlation field direction @xmath123 , while the total effective magnetic field is rotated by an angle @xmath128 with respect to @xmath123 . for the @xmath12 states there is an additional complication that the orientations of the induced local moments on the pt atoms are not known in advance .
these orientations are also updated at each iteration using the output directions of the induced magnetic moments , guaranteeing full dft self - consistency in the final state . as explained in the main text , the vdlm and noncollinear cpa calculations were performed for a number of concentrations with experimental lattice parameters .
the calculated ( about 200 ) total energies for each candidate @xmath12 phase at the given concentration are then fitted to a polynomial in the four order parameters @xmath130 , @xmath131 ( where 1 and 2 label the ordering types , e.g. f and g for phase f / g ) : @xmath132 . the polynomials @xmath133 and @xmath134 include all even terms up to sixth order , and @xmath135 includes all products of second - order monomials in @xmath130 and @xmath131 ( lower - order terms in @xmath135 are forbidden by symmetry ) .
this fit also covered the two @xmath136 phases , which are obtained at @xmath137 or @xmath138 .
3 of the main text illustrates the quality of the fit and its energy profile for fully ordered f / g phase at one selected concentration @xmath45 . the actual expression
@xcite is the following ( @xmath139 , @xmath140 ) : the polynomial @xmath135 describes interaction between the order parameters at two different @xmath31 vectors .
information needed to fit this polynomial is only contained in the noncollinear cpa results for the given @xmath12 phase .
( @xmath135 vanishes in the @xmath136 phases described by the vdlm method . )
since these calculations ( contrary to vdlm ones ) are done only for magnetically - ordered phases , certain higher - order terms in @xmath135 would be linearly dependent within the available dataset and could not be distinguished .
indeed , in a fully - ordered @xmath12 phase we have @xmath142 , and terms like @xmath143 , @xmath144 , and @xmath145 are linearly dependent .
therefore , for most concentrations we restricted @xmath135 to fourth - order terms , which are expected to be the most important .
however , at the mn - rich end the interaction of the fe spins with the mn host has a particularly strong non - heisenberg character in the f / c phase , making the fourth - order fit for @xmath135 insufficient .
therefore , at @xmath146 we added sixth - order terms to @xmath147 that are second - order in f and fourth - order in c order parameters ( such as @xmath148 , etc . ) .
in addition , to provide a consistent fit for both g / c and f / c phases , in this concentration range we combined all input data to obtain a single fit for the @xmath149 structure involving all three order parameters .
( possible interaction between the f and g order parameters , which is irrelevant in this region , was left undetermined . ) | a first - principles approach to the construction of concentration - temperature magnetic phase diagrams of metallic alloys is presented .
the method employs self - consistent total energy calculations based on the coherent potential approximation for partially ordered and noncollinear magnetic states and is able to account for competing interactions and multiple magnetic phases .
application to the fe@xmath0mn@xmath1pt `` magnetic chameleon '' system yields the sequence of magnetic phases at @xmath2 and the @xmath3-@xmath4 magnetic phase diagram in good agreement with experiment , and a new low - temperature phase is predicted at the mn - rich end . the importance of non - heisenberg interactions for the description of the magnetic phase diagram is demonstrated .
magnetic substitutional alloys are often found to excel in applications @xcite , because alloying broadens the parameter space for tuning the desired properties .
however , wide tunability , combined with the need to target certain operating temperature ranges , presents a challenge for empirical materials design
. competing magnetic interactions in alloys can produce complicated magnetic phase diagrams ( mpd ) with multiple magnetic phases @xcite .
understanding of the @xmath3-@xmath4 mpd s is therefore essential for the development of advanced magnetic materials .
some mpd s can be computed using the heisenberg model with empirical or calculated exchange parameters combined with the mean - field approximation ( mfa ) @xcite monte - carlo simulations @xcite , or spin - fluctuation theory @xcite .
however , many systems are not adequately described by the heisenberg model . in metallic alloys
the interaction parameters are sensitive to the electronic structure and population and thereby to the content of the alloy @xcite and to the degree of spin disorder @xcite .
to avoid the limitations of the heisenberg model , one can use first - principles spin - dynamics simulations @xcite or construct a generalized spin hamiltonian to map the adiabatic energy surface @xcite for use in thermodynamic calculations .
the energies of disordered spin configurations can also be obtained using the disordered local moment ( dlm ) model @xcite , where the spin - rotational averaging is done in the coherent potential approximation ( cpa ) .
while all these approaches fail in strongly itinerant magnets , they are applicable when the spin moments do not vary by more than 10 - 20% in different spin configurations .
we restrict ourselves to such systems here .
first - principles spin - dynamics and the construction of a microscopic generalized spin hamiltonian are computationally very demanding and unfeasible for most systems of practical interest .
we have developed @xcite an alternative approach , in which self - consistent dlm and noncollinear cpa calculations are used to construct a ginzburg - landau - type total - energy functional expressed through a small number of relevant magnetic order parameters .
combined with the mfa expression for the magnetic entropy , this method provides the variational free energy .
a similar scheme was used to describe the phase transitions in ferh @xcite .
here we show , using the fe@xmath0mn@xmath1pt disordered alloy system as a test case , that this efficient approach is sufficiently powerful to explain and refine a complicated mpd , not only reproducing the five known magnetic phases but also predicting another , hidden low - temperature phase in this system .
fe@xmath0mn@xmath1pt alloys are of interest for ultrahigh - density magnetic recording and medical applications @xcite .
their structural ordering is of the l1@xmath5 type in the fcc sublattice , with ( 001 ) layers of pt alternating with disordered fe / mn layers .
neutron diffraction measurements revealed three collinear and two noncollinear phases @xcite .
the collinear phases are the ferromagnetic ( f ) phase at the fe - rich end , the c - type antiferromagnetic phase at the mn - rich end , and the g - type antiferromagnetic phase in the middle of the diagram .
the corresponding ordering wave vectors are @xmath6 , @xmath7 , and @xmath8 in units of @xmath9 ( or @xmath10 for the @xmath11 component ) .
the transitions between the collinear phases occur through intermediate @xmath12 phases combining the corresponding orderings for two orthogonal spin components ( see supplemental material @xcite for an illustration ) .
the samples studied in ref . show a high degree of l1@xmath5 order at all concentrations , which is consistent with the fact that magnetic ordering occurs well below the structural ordering temperatures in these alloys . therefore , as a practical simplification we assumed perfect l1@xmath5 ordering and complete disorder of fe and mn atoms within their own sublattice .
the method can also be applied to alloys with partial chemical ordering and in principle allows one to study the coupling between magnetic and chemical order parameters @xcite . to construct the energy functional ,
we have extended our cpa implementation @xcite based on the tight - binding linear muffin - tin orbital formalism @xcite by special features designed to describe complicated magnetic states .
first , we implemented the vector dlm ( vdlm ) model , in which partially ordered magnetic states are specified by the curie - weiss distribution functions @xmath13 , where @xmath14 is the lattice site and @xmath15 the component index ; @xmath16 are regarded as variational parameters .
this formulation is suitable for systems with axial spin symmetry , i. e. those with collinear magnetic order .
different spin moment orientations are treated as different alloy components in the cpa formalism .
the integral over the azimuthal angle in the cpa equations is taken analytically , while the @xmath17 dependence is discretized using the 16-point gauss - legendre quadrature .
the potentials for all atoms are determined by embedding the cpa self - consistency loop into the density - functional theory ( dft ) charge iteration . to enforce magnetic self - consistency , constraining transverse magnetic fields
@xcite are introduced for each orientation of the spin moment and determined self - consistently .
local density approximation is used for exchange and correlation .
the second feature extends cpa to the noncollinear case , in which the orientations of the spin moments of different components on the same lattice site can be different .
this method is suitable , in particular , for the studies of @xmath12 structures appearing in the fe@xmath0mn@xmath1pt system .
self - consistent constraining fields are also used in these calculations .
both vdlm and noncollinear cpa calculations yield the dft total energy .
let us first examine the magnetic interactions in the vicinity of the paramagnetic state .
we set up a unit cell for each of the three magnetic orderings and calculate the total energy for about 70 partially ordered vdlm states with @xmath18 and @xmath19 ranging from 0 to 3 .
experimental lattice constants are used at the concentrations reported in ref .
@xcite . at each concentration the calculated total magnetic energy @xmath20 ( per fe / mn atom , referenced from the paramagnetic state )
is fitted to even fourth - order polynomials in the reduced magnetizations @xmath21 , which are expressed through the fields @xmath22 by the langevin function .
the quadratic part of these polynomials , @xmath23 , defines the component - resolved effective exchange interactions @xmath24 for the three orderings .
since the total energies are calculated with constraining fields , these exchange parameters are free from the errors associated with the long - wave approximation @xcite .
the results are shown in fig .
[ j0 ] .
the concentration dependence of the parameters @xmath24 shows that , in agreement with experiment , the f , g , and c - type orderings are favored at the fe - rich end , in the middle , and at the mn - rich end , respectively .
further insight can be obtained from the reduced exchange parameters that are normalized by the concentrations , @xmath25 , shown in fig .
[ j0c ] .
these reduced quantities would be concentration - independent in a heisenberg system with pair exchange parameters @xmath26 depending only on the distance and the identity of the atoms in a pair .
we see that the parameters @xmath27 for like components ( i. e. fe - fe and mn - mn ) are almost constant for all ordering vectors , as well as the reduced fe - mn coupling for the f ordering .
however , the reduced fe - mn couplings at the g - type and c - type ordering vectors depend strongly on the concentration , which reflects the effect of band filling on the exchange interaction in metallic systems .
the local spin moments of fe and mn increase by about 10% as @xmath28 goes from 0 to 1 ( for example , from 2.87 to 3.11 @xmath29 for fe and from 3.35 to 3.74 @xmath29 for mn in the paramagnetic phase ) , and similar variations are observed for different phases and spin directions .
the adiabatic approach @xcite is thus well suited for this system .
we also repeated some calculations without the constraining magnetic fields , which is equivalent to making the long - wave approximation @xcite , and found that the resulting errors in @xmath30 for all phases do not exceed 5 - 7% . to further extend the mapping of the magnetic configuration space , we performed noncollinear cpa calculations for the f / g , g / c , and f / c noncollinear @xmath12 phases in the relevant concentration ranges .
these calculations are needed to reveal the possible interaction between orderings at different @xmath31 in the @xmath12 structures , which can appear in quartic and higher - order interaction terms .
a @xmath12 structure is parameterized by two angles , @xmath32 and @xmath33 that the spin moments of fe and mn atoms make with the @xmath11 axis .
( the @xmath11 and @xmath28 components of the magnetization order with one or the other of the @xmath31 vectors . )
using the symmetries , the accessible space of ( @xmath34 is reduced to the range @xmath35 , @xmath36 with additional @xmath37 symmetry at @xmath38 and @xmath39 symmetry at @xmath40 .
this irreducible domain is covered by a uniform mesh of 38 points . for a particular @xmath12 phase ( say , f / g )
we then combine the data from the separate vdlm calculations for the f and g phases with those from the noncollinear cpa results for the @xmath12 phase and fit the magnetic energy at the given concentration to a polynomial in the order parameters @xmath41 , @xmath42 , @xmath43 , and @xmath44 .
we allowed all symmetry - respecting terms in the polynomials of up to sixth order ( see supplemental material @xcite for details ) .
the high accuracy of the fits is illustrated in fig .
[ fit]a for the f / g phase at @xmath45 ; all other fits are of similar accuracy .
fig .
[ fit]b shows the magnetic energy as a function of @xmath32 , @xmath33 in the same phase at @xmath45 . at this concentration
there are two minima : the global one at @xmath46 corresponding to the collinear g - type phase , and a local one near @xmath47 corresponding to the @xmath12 phase . at a lower concentration
there is a first - order transition where the @xmath12 minimum goes below the g - type minimum . .
( b ) the magnetic energy @xmath20 ( mev per fe / mn atom ) in the f / g phase at @xmath45 .
the global minimum at @xmath46 corresponds to the collinear g - type phase.,title="fig:",scaledwidth=23.0% ] .
( b ) the magnetic energy @xmath20 ( mev per fe / mn atom ) in the f / g phase at @xmath45 .
the global minimum at @xmath46 corresponds to the collinear g - type phase.,title="fig:",scaledwidth=23.0% ] using the combined fits for the magnetic energy , we now determine the ground states for all concentrations by choosing the lowest energy of all the competing phases .
the results are shown in fig .
[ ground ] .
we find that the f phase is stable at @xmath48 and the g phase at @xmath49 .
there is a @xmath12 f / g phase at @xmath50 separated from f by a second - order and from g by a first - order transition .
the g / c phase is stable at @xmath51 .
surprisingly , at @xmath52 we find a first - order transition from the g / c to the f / c phase .
the existence of this first - order transition is in excellent agreement with the observed abrupt drop in the mean magnetic moment at this concentration @xcite .
however , since the f component in the low - temperature mn - rich phase has not been previously identified , the existence of the f / c phase is a prediction that needs to be verified experimentally .
the f / c and g / c phases at the mn - rich end differ essentially in the fe ordering alone , as the ordering of the mn spins is almost purely c - type .
the energy difference between the f / c and g / c phases reaches about 20 mev per fe atom near @xmath53 ; it is barely visible in fig .
[ ground ] because of the small fe concentration .
the first - order transition from f / g to g at @xmath54 is also in excellent agreement with experiment , while the transition from g to g / c occurs at a larger @xmath28 compared to experiment , where it is close to @xmath55 .
note , however , that our calculations are for systems with perfect @xmath56/@xmath57 ordering , while the order parameter in the experimental samples for @xmath55 and @xmath58 was 0.79 @xcite . ) .,scaledwidth=45.0% ] fig .
[ angles ] shows the angles @xmath32 , @xmath33 in the ground states as a function of @xmath28 .
( by convention , f and c amplitudes lie along the @xmath11 and g along the @xmath28 axis ; except for the f / c phase , where f is along @xmath28 . ) as could be inferred from fig .
[ j0 ] , the spin moments of fe and mn are antiparallel in the f phase and parallel in the g phase .
first - order phase transitions appear as discontinuous jumps of the angles .
note that in a wide concentration range at the mn - rich end the fe spin moments are almost perpendicular to the mnpt host in the ground state , while mn ordering is almost pure c - type .
( a full set of first - principles calculations with interpolated lattice constants was performed with a small 0.01 step in @xmath59 range to confirm this . )
this feature highlights the importance of non - heisenberg terms in the magnetic energy . made by the spin moments of fe and mn with the @xmath11 axis at zero temperature . by convention
, the g order parameter , as well as the f order parameter in the f / c phase , are assumed to be orthogonal to the @xmath11 axis ; all others are parallel to @xmath11.,scaledwidth=45.0% ] we now turn to the full concentration - temperature phase diagram .
we have the fits for the magnetic energy @xmath60 , where the @xmath28 and @xmath11 components of @xmath61 and @xmath62 correspond to the two @xmath31 vectors ; in the collinear phase one of these components vanishes .
@xmath20 is available for discrete concentrations , and we use linear interpolation between them .
the entropy is approximated as @xmath63 , where @xmath64 is the entropy of a classical spin in a weiss field of such a magnitude that the magnetization is @xmath65 .
this corresponds to the mean - field - like distribution function @xmath66 , where @xmath67 is the angle with respect to the direction of @xmath68 .
given that @xmath69 in the f / g phase , etc .
, we minimize the free energy for each phase with respect to four parameters @xmath32 , @xmath33 , @xmath70 and @xmath71 . for a system with purely heisenberg interaction this scheme is identical to mfa applied to @xmath12 phases @xcite .
therefore , both second - order transitions to the paramagnetic phase ( where non - heisenberg terms have no effect in mfa ) and ground - state properties ( where only energy is important ) are correctly described by this approach . in the intermediate temperature
range our scheme can therefore be treated as a thermodynamic interpolation . at such intermediate temperatures
we are essentially assuming that the presence of non - heisenberg terms does not strongly change the distribution functions and that our magnetic energy fits remain valid for partially disordered @xmath12 phases .
in most cases , these assumptions are likely consistent with the accuracy of the mfa itself
. significant improvement of the thermodynamic description could be reached by mapping the total energies to a microscopic spin hamiltonian , but this expensive procedure is beyond the scope of this paper .
fig .
[ pd ] shows the mpd obtained both from the full magnetic energy fits and from the same fits truncated at the quadratic ( heisenberg ) terms .
first - order transitions are shown by dashed lines .
the overall structure of the phase diagram agrees well with experiment .
the first - order transition from the f / g to the g phase may help explain the peculiarities of the observed temperature - dependent magnetization peaks @xcite .
indeed , there is a concentration range from @xmath72 to @xmath73 where the ground - state g - type ordering turns into f / g and then to f on heating .
the physics is complicated by configurational disorder , which may lead to the formation of small fe - rich clusters at elevated temperatures @xcite .
if only heisenberg terms are kept in the magnetic energy , the transitions into the paramagnetic phase remain unchanged .
however , all first - order transitions turn into second - order ; the f / g to g transition is shifted to much larger concentrations , and the f / c phase disappears completely .
thus , in order to describe the observed first - order transitions at @xmath72 and @xmath52 it is important to take the non - heisenberg interaction terms into account .
the f / c phase disorders at rather low temperatures in our description , although its stability may be underestimated in our thermodynamic scheme .
the non - zero magnetization in this phase should facilitate an easy experimental verification of its existence .
mn@xmath1pt .
temperatures are rescaled by the ratio @xmath74 , where @xmath75 ; @xmath76 and @xmath77 are the ordering temperatures of fept and mnpt from ref .
( @xmath78 ) or theory ( @xmath79 ) . ( mfa gives @xmath80 k and @xmath81 k. ) thick ( blue ) lines : full fit for @xmath20 .
thin ( red ) lines : same fit but with non - heisenberg terms set to zero .
solid ( dashed ) lines : second ( first ) order phase transitions . symbols : experimental data @xcite.,scaledwidth=45.0% ] to conclude , we developed a computational tool based on a combination of first - principles calculations that is capable of describing a complicated @xmath3-@xmath4 magnetic phase diagram of a metallic alloy with competing interactions .
its application to the fe@xmath0mn@xmath1pt system produced a detailed interpretation of the experimental phase diagram and also predicted the existence of a previously unknown low - temperature magnetic phase on the mn - rich end .
the correct first - order transitions and the mn - rich phase are only captured if non - heisenberg terms are included in the magnetic energy , showing the limitations of the conventional approach based on the heisenberg - model hamiltonian .
the wide applicability and predictive power of this approach makes it useful for the design of magnetic materials with desired properties .
the work at unl was supported by the national science foundation through grant no .
dmr-1308751 and the nebraska mrsec ( dmr-0820521 ) and was performed utilizing the holland computing center of the university of nebraska .
work at ames lab is supported in part by the critical materials institute , an energy innovation hub funded by the us doe and by the office of basic energy science , division of materials science and engineering .
ames laboratory is operated for the us doe by iowa state university under contract no .
de - ac02 - 07ch11358 . 99 _ handbook of magnetism and advanced magnetic materials _ , edited by h. kronmller and s. parkin ( wiley , chichester , 2007 ) , vol
. 4 and 5 . _ handbook of advanced magnetic materials _ , edited by y. liu , d. j. sellmyer , and d. shindo ( springer , berlin , 2006 ) .
d. fruchart and e. f. bertaut , j. phys .
soc .
jpn * 44 * , 781 ( 1978 ) .
m. shimizu , j. magn .
magn . mater . *
50 * , 319 ( 1985 ) .
a. z. menshikov , physica b * 149 * , 249 ( 1988 ) .
y. tokura and y. tomioka , j. magn .
magn . mater . *
200 * , 1 ( 1999 ) .
f. a. kassan - ogly and m. v. medvedev , phys .
status solidi b * 76 * , 45 ( 1976 ) .
g. c. kaphle , s. ganguly , r. banerjee , r. banerjee , r. khanal , c. m. adhikari , n. p. adhikari , and a. mookerjee , j. phys . : condens .
matter * 24 * , 295501 ( 2012 ) .
j. rusz , l. bergqvist , j. kudrnovsk , and i. turek , phys .
rev .
b * 73 * , 214412 ( 2006 ) .
a. v. ruban , s. khmelevskyi , p. mohn , and b. johansson , phys .
rev .
b * 75 * , 054402 ( 2007 )
. p. pal , r. banerjee , r. banerjee , a. mookerjee , g. c. kaphle , b. sanyal , j. hellsvik , o. eriksson , p. mitra , a. k. majumdar , and a. k. nigam , phys . rev .
b * 85 * , 174405 ( 2012 ) .
v. d. buchelnikov , v. v. sokolovskiy , h. c. herper , h. ebert , m. e. gruner , s. v. taskaev , v. v. khovaylo , a. hucht , a. dannenberg , m. ogura , h. akai , m. acet , and p. entel , phys .
rev .
b * 81 * , 094411 ( 2010 ) .
l. m. sandratskii , m. uhl , and j. kbler , in _ itinerant electron magnetism ; fluctuation effects _ , edited by d. wagner _
et al . _
( kluwer , amsterdam , 1998 ) , pp .
161 - 192 .
d. bttcher , a. ernst , and j. henk , j. magn .
magn . mater . *
324 * , 610 ( 2012 ) .
v. p. antropov , m. i. katsnelson , b. n. harmon , m. van schilfgaarde , and d. kusnezov .
phys .
rev .
b * 54 * , 1019 ( 1996 ) . a. l. wysocki , k. d. belashchenko , and v. p. antropov , nature phys .
* 7 * , 485 ( 2011 ) .
j. k. glasbrenner , j. p. velev , and i. i. mazin , phys .
rev .
b * 89 * , 064509 ( 2014 ) .
t. oguchi , k. terakura , and n. hamada , j. phys .
f : met .
phys .
* 13 * , 145 ( 1983 ) .
b. l. gyorffy , a. j. pindor , j. staunton , g. m. stocks , and h. winter , j. phys .
f : met .
phys . * 15 * , 1337 ( 1985 ) .
k. belashchenko , b. pujari , p. larson , v. antropov , and m. van schilfgaarde , bull . am .
. soc . * 58 * , a24.13 ( 2013 ) , http://meetings.aps.org/link/baps.2013.mar.a24.13 .
j. b. staunton , r. banerjee , m. dos santos dias , a. deak , and l. szunyogh , phys .
rev .
b * 89 * , 054427 ( 2014 ) .
m. e. gruner and p. entel , beilstein j. nanotechnol . * 2 * , 162 ( 2011 ) .
a. z. menshikov , v. p. antropov , g. p. gasnikova , y. a. dorofeyev , and v. a. kazantsev , j. magn .
magn . mater . *
65 * , 159 ( 1987 ) .
see supplemental material for computational details and pictures of the magnetic phases . a. bieber and f. gautier , j. magn .
magn . mater . *
99 * , 293 ( 1991 ) .
l. ke , k. d. belashchenko , m. van schilfgaarde , t. kotani , and v. p. antropov , phys .
rev .
b * 88 * , 024404 ( 2013 ) .
k. d. belashchenko , l. ke , m. dne , l. x. benedict , t. n. lamichhane , v. taufour , a. jesche , s. l. budko , p. c. canfield , and v. p. antropov , appl .
. lett . * 106 * , 062408 ( 2015 ) .
i. turek , v. drchal , j. kudrnovsk , m. sob , and p. weinberger , _ electronic structure of disordered alloys , surfaces , and interfaces _ ( kluwer , boston , 1997 ) .
p. h. dederichs , s. blgel , r. zeller , and h. akai , phys .
rev .
lett .
* 53 * , 2512 ( 1984 ) .
g. m. stocks , b. ujfalussy , xindong wang , d. m. c. nicholson , w. a. shelton , yang wang , a. canning , and b. l. gyrffy , phil . mag .
b * 78 * , 665 ( 1998 ) .
v. p. antropov , j. magn .
magn . mater . *
262 * , l192 ( 2003 ) . |
we consider the microscopic model which describes the interaction of a single particle of mass @xmath5 with a weakly interacting gas of bosons of mass @xmath6 .
we assume that the interactions between the particles are described by contact interactions .
this approximation is valid for the description of ultracold gases close to a broad inter - species feshbach resonance @xcite .
the hamiltonian reads @xmath89 where @xmath90 and @xmath91 are the creation operators that describe the bosonic host particles and an impurity with the corresponding dispersion relations @xmath92 and @xmath93 .
the parameters of the contact interaction ( bare interaction ) are given by the coefficient @xmath94 for the interaction between the host particles and the impurity and @xmath95 for the interaction among the host particles .
we allow for the general case of a mass imbalanced system where the mass of the bosonic field @xmath6 and the mass of the impurity @xmath5 are not fixed to be equal .
we account for the macroscopic occupation of the @xmath96 mode within the bogoliubov approximation @xcite . following the corresponding truncation of the hamiltonian on the quadratic level in the bosonic creation and annihilation operators we diagonalize the purely bosonic part of the hamiltonian by a unitary bogoliubov rotation@xmath97 here
the coefficients are given by @xmath98 , @xmath99 .
this transformation introduces the bogoliubov quasiparticles with dispersion relation @xmath100 which are created by the operators @xmath101 . reexpressing the resulting hamiltonian in explicit impurity coordinates yields the hamiltonian in the main text . in order to take advantage of momentum conservation
we perform a canonical transformation to the so - called polaron frame using @xmath28 where @xmath29 is the total momentum operator of the bosons .
the operator @xmath102 transforms the momentum of the impurity and bosonic operators in the following way @xmath103 applying this transformation to the hamiltonian , @xmath104 , we obtain @xmath105 since @xmath19 now commutes with the hamiltonian , the hilbert space of the hamiltonian @xmath65 becomes a direct product of the sub - spaces characterized by the total momentum of the system .
in this section we derive the saddle point solution of the equations of motions of the parameters in eq . .
we set the left hand side of the first equation to zero and obtain the following equation which defines the saddle point @xmath106 here @xmath107 , @xmath108 , and @xmath109 denotes the total system momentum .
the real and imaginary parts of the coherent amplitudes @xmath110 can be expressed as @xmath111 the first equation allows for an analytical solution which is found by a multiplication with @xmath112 , a summation over all modes and expressing the problem in terms of the averaged quantity @xmath113 .
after some algebra one obtains @xmath114 the microscopic interaction strength @xmath12 is related to the impurity - boson scattering length @xmath20 by the zero - momentum limit of the lippmann - schwinger equation which yields @xmath115 where @xmath59 denotes a high - momentum cutoff which can be taken to infinity . upon insertion into eq .
one obtains @xmath116 where we defined @xmath117 as @xmath118 the second equation has the trivial solution @xmath119 .
after the coherent amplitudes are expressed in terms of the inverse scattering length , a self - consistent procedure for the determination of @xmath117 and the total momentum of the phonons can be set up . to this end
we substitute the eq . into the total momentum of the phonons , @xmath120 , and
obtain the expression @xmath121 the self - consistent solution of eqs . and allows to find the saddle point for given initial conditions , @xmath20 and @xmath122 .
finally , the energy is obtained as expectation value of the hamiltonian in this state .
after some algebra one obtains the expression for energy @xmath123
in this section we provide the derivation of the bound state energy .
we propose the wave - function which accounts for a single bogoliubov excitation above the polaron state @xmath69 @xmath124 the polaron state in the lee - low - pines frame , i.e. after the transformation with @xmath102 , can be represented as a displacement operator @xmath125 = \exp\left ( \sum_{\v k } \beta_{\v k } \hat b_{\v k}^\dag - \text{h.c.}\right)$ ] acting on the vacuum state , @xmath126 \ket{0}$ ] . here
the @xmath127 are defined by eq . in the previous section of the sm . after some algebra we can rewrite the state @xmath128 in the form @xmath129 \left ( \gamma_0(t ) + \sum_{\mathbf k}\gamma_{\mathbf k}(t ) { \hat b^\dag } _ { \mathbf k}\right ) \ket{0}\ ] ] where @xmath130 .
we calculate the lagrangian of the system with respect to this state , @xmath131 , and obtain the equations of motion for the parameters of the wave - function @xmath132 the first equation can be solved trivially assuming @xmath133 .
this solution implies that the bound state solution is orthogonal to the polaron state .
we are searching for a solution of these equations of motion such that the parameters @xmath134 of the wave - function are time evolving with frequency @xmath72 , @xmath135 . to this end
we transform the second line of eq . to the frequency domain and find @xmath136 multiplying by @xmath112 or @xmath137 and summing over over the momentum we obtain the following matrix equation , @xmath138 this equation has a nontrivial solution if and only if the determinant of the matrix is equal to zero .
neglecting terms that are vanishing in the limit @xmath139 , @xmath140 , we obtain the following equation for the eigenenergy @xmath141 of the bound state @xmath142 using eq .
we recast eq . in the form of eq . in the main text .
this simple analysis agrees with the features below the polaron branch in the absorption spectrum as shown in fig .
[ fig : spectrum ] .
in this section we compare the results for the spectral function @xmath143 of the bose polaron in the vicinity of a feshbach resonance , as described by the model , using different theoretical approaches . while in fig .
[ fig : chevy](a ) we show @xmath143 as obtained from a single - particle expansion using the time - dependent variational state eq . , fig .
[ fig : chevy](b ) shows the result using a self - consistent field - theoretical t - matrix approach @xcite .
finally the results are compared with fig . [
fig : chevy](c ) which shows the result from the time - dependent coherent state approach as described by eq . .
all results are obtained for @xmath144 and a cutoff scale @xmath145 .
we note that the result shown in fig .
[ fig : chevy](a ) has been obtained previously in ref .
@xcite using a non - self - consistent t - matrix ( nsct ) approach which is equivalent to the time - dependent single particle excitation ansatz eq . when @xmath144 .
the main features in the nsct / single particle excitation approach are the coherent polaron excitations at negative and positive energies . in this approach
the attractive polaron peak exists at arbitrary interaction strength , and the excitation follows the mean - field prediction @xmath146 ( green line ) in the limit of small dimensionless interaction parameter @xmath147 . as the feshbach resonance is crossed to positive @xmath0 the attractive polaron peak looses weight .
however , in contrast to the coherent state approach [ cf .
[ fig : chevy](c ) ] , neither the destruction of the polaron quasiparticle at resonance is captured , nor the formation of the series of bound states for increasing @xmath0 . at positive scattering length @xmath20 the repulsive polaron excitation at positive energies is recovered .
it follows the saddle point prediction only for very weak effective impurity bath interactions .
away from the coherent excitations , the nsct / single - particle excitation approach predicts a spectral gap between the attractive polaron and zero energy , cf .
[ fig : chevy](a ) .
this gap is , however , an artifact of the approximation .
for instance , in a self - consistent t - matrix ( sct ) approach , shown in fig .
[ fig : chevy](b ) , such a gap is absent . in this approach
an infinite number of particle excitations is taken into account at the basis of a self - consistent calculation of the in - medium t - matrix and impurity self - energy ( the computational details are described in @xcite ) .
the , compared to the nsct approach , increased number of bosonic excitations leads to an additional renormalization of the spectrum .
as one consequence the spectral gap , visible in the nsct approach , is absent and furthermore the repulsive polaron looses additional spectral weight as the feshbach resonance is approached from the side of negative scattering length .
this loss of spectral weight is even more pronounced in the coherent state approach , cf .
[ fig : chevy](c ) , where multiple boson fluctuations lead to the destruction of the attractive polaron peak .
in contrast to the coherent state approach , the single - particle excitation and sct approaches do not describe the formation of the series of bound states . in the case of the single - particle excitation approach
this fact is simply explained since only one boson excitation is allowed in the wave function . the sct approach , which allows in principle more than one bosonic excitation , yet does not reveal higher order bound states .
we attribute this fact to the specific momentum structure generated by the sct approach which only allows for fluctuations in the pairing channel of the interaction vertex .
the particular correlations captured by this momentum structure seems to be insufficient for a description of the multiple bound state formation .
finally we note that both the sct and the coherent state approach show a more pronounced repulsive polaron peak as compared to the single - particle excitation approach which can be tested by radio - frequency spectroscopy of impurities immersed in a bec of ultracold atoms . | we analyze the dynamics of bose polarons in the vicinity of a feshbach resonance between the impurity and host atoms .
we compute the radio - frequency absorption spectra for the case when the initial state of the impurity is non - interacting and the final state is strongly interacting .
we compare results of different theoretical approaches including a single excitation expansion , a self - consistent t - matrix method , and a time - dependent coherent state approach .
our analysis reveals sharp spectral features arising from metastable states with several bogoliubov excitations bound to the impurity atom .
this surprising result of the interplay of many - body and few - body efimov type bound state physics can only be obtained by going beyond the commonly used frhlich model and including quasiparticle scattering processes .
close to the resonance we find that strong fluctuations lead to a broad , incoherent absorption spectrum where no quasi - particle peak can be assigned . of a single impurity immersed in a weakly interacting bec as a function of the inverse interaction strength @xmath0 .
the polaron energy is shown as green line and the energies , cf .
eq . , of bound states with one , two , and three quasiparticles attached to the polaron are shown as red dashed lines . in the regime of weak repulsion
the energy of the first bound state is approaching the dimer binding energy ( gray dash - dotted line ) .
the spectrum is shown for a finite microscopic interaction range corresponding to a momentum cutoff @xmath1 and boson - boson scattering length @xmath2 . ]
[ [ introduction . ] ] * introduction . *
+ + + + + + + + + + + + + + + + + understanding the role of few - body correlations in many - body systems is a challenging problem that arises in many areas of physics .
few particle systems in isolation can be studied using powerful techniques of scattering theory such as faddeev equations , the hyperspherical formalism , or effective field theory @xcite .
these approaches have been successfully applied to investigate collisions of neutrons and protons @xcite and efimov resonances in ultracold atoms @xcite . on the other hand the common approach to interacting many - body systems
is to use the mean - field approximation , which reduces a many - body problem to an effective single particle hamiltonian with self - consistently determined fields . while this approach provides a good description of many fundamental states , including magnetic , superconducting , and superfluid phases @xcite , in many cases it is important to go beyond the mean - field paradigm and include correlations at a few particle level .
recent notable examples include 4e pairing in high tc superconductors @xcite , spin nematic states @xcite , chains and clusters of molecules in ultracold atoms @xcite , and the qcd phase diagram in high - energy physics @xcite . a particularly important class of problems where few - body correlations play a crucial role is the formation of quasiparticles and polarons .
the key feature of both is the dramatic change in the particle dynamics due to the interaction with collective excitations of the many - body system .
famous examples include lattice polarons in semiconductors @xcite , magnetic polarons in strongly correlated electron systems @xcite , and @xmath3he atoms in superfluid @xmath4he @xcite .
recent experiments with ultracold atoms opened a new chapter in the study of polaronic physics @xcite .
these systems have tunable interactions between impurity and host atoms @xcite and powerful experimental techniques for characterizing many - body states including spectroscopy @xcite , ramsey interferometry @xcite , time of flight experiments @xcite , and in - situ measurements with single atom resolution @xcite . in this paper
we explore the dynamics of bose polarons in the specific setting of radio - frequency ( rf ) spectroscopy of impurity atoms immersed in a bose - einstein condensate ( bec ) .
the most surprising finding of our study is the appearance of sharp spectral lines arising from states of several bogoliubov quasiparticles bound to the impurity atom ( see fig .
[ fig : spectrum ] ) . this can be contrasted to earlier theoretical studies that predicted only a single molecular state in addition to attractive and repulsive polarons @xcite .
our result demonstrates the importance of the interplay of many - body physics of the bec of host atoms and few particle efimov - type physics in the vicinity of a feshbach resonance @xcite .
this interplay can not be studied in the commonly used frhlich model of bec polarons because the latter does not include two particle scattering processes that results in the feshbach resonance .
our work also demonstrates that a broad spectral feature at unitarity can be understood as a superpolaronic state discussed previously in rydberg systems @xcite .
[ [ model . ] ] * model*. + + + + + + + + + + we consider an impurity of mass @xmath5 interacting with a weakly interacting bec of atoms of mass @xmath6 in the vicinity of an inter - species feshbach resonance . within the bogoliubov approximation
the system is described by the hamiltonian ( a derivation is provided in the supplementary material ( sm ) ) @xcite , @xmath7 here the operators @xmath8 ( @xmath9 ) create ( annihilate ) bogoliubov quasiparticles ( ` phonons ' ) with momentum @xmath10 and dispersion @xmath11 .
the bare inter - species interaction is given by @xmath12 .
furthermore @xmath13 , and @xmath14 define the interaction vertices where @xmath15 is the dispersion relation of bare host atoms ; @xmath16 is the condensate density , and @xmath17 the system s volume .
the last two lines in eq .
describe the interaction of the impurity at position @xmath18 and momentum @xmath19 with the host bosons . in the frhlich model
only the interaction term linear in the bosonic operators is present and it describes the creation of excitations directly from the bec .
however , a microscopic derivation reveals that in cold atomic systems also the additional quadratic terms , included in eq . , are present which lead to rich physics beyond the frhlich paradigm @xcite .
the extended hamiltonian allows for a proper regularization of the contact interaction between the impurity and bosons which is beyond the frhlich model . from the solution of the two - body scattering problem of eq
.
follows the relation of the microscopic interaction strength @xmath12 to the impurity - boson scattering length @xmath20 by the lippmann - schwinger equation @xmath21 here @xmath22 is the reduced mass and @xmath23 denotes an ultraviolet ( uv ) cutoff scale related to a finite range @xmath24 of the interaction potential . in the limit
@xmath25 contact interactions are recovered .
we note that all our numerical results are obtained for the mass - balanced case @xmath26 .
we describe the impurity - bath system in the frame co - moving with the polaronic quasiparticle @xcite .
this is achieved using a canonical transformation @xmath27 with @xmath28 where @xmath29 is the total momentum operator of the bosons .
after the transformation sectors with different total system momentum @xmath30 are decoupled in the hamiltonian @xmath31 .
the bosons now interact with each other since the impurity kinetic energy transforms according to @xmath32 @xcite , for details see sm .
[ [ quantum - quench - dynamics . ] ] * quantum quench dynamics*.
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + in this work we predict the excitation spectrum of eq . .
in experiments the spectrum can be explored using rf spectroscopy where the impurity is initially in a spin - state @xmath33 in which it does not interact with the bath .
then the response is measured with respect to a weak monochromatic perturbation @xmath34 flipping the spin to a state @xmath35 which does interact with the bosons with a strength determined by the dimensionless interacting parameter @xmath36 .
using fermi s golden rule we calculate the absorption spectrum
i.e. the impurity spectral function in linear response @xmath37 , where @xmath38 is a time - dependent overlap . here
@xmath39 denotes the initial state of the system and the overlap @xmath40 describes the dynamics of the system after a quench of the interactions between impurity and the bath . in real - time
the overlap @xmath40 can be measured using ramsey interferometry @xcite .
of a single impurity immersed in a weakly interacting bose gas for the fixed scattering length @xmath0 : ( a ) @xmath41 , ( b ) @xmath42 , ( c ) @xmath43 , ( d ) @xmath44 .
for the negative scattering length , panel ( a ) , the absorption spectrum reveals the attractive polaron which follows closely the polaron energy . at unitarity , panel ( b )
, the spectrum shows a broad spectral feature and no quasi particle peak can be assigned .
as the feshbach resonance is crossed towards positive scattering lengths , panel ( c ) , a series of bound states emerges . for small positive scattering length , panel ( d ) , the binding energy of the negative frequency peak increases while at positive frequencies a long - lived repulsive polaron becomes the dominant excitation .
as for fig .
[ fig : spectrum ] we artificially broadened the spectrum to make sharp features visible . ] in order to predict the real - time evolution as well as the excitation spectrum of the system , we invoke the time - dependent variational principle @xcite .
the approach relies on a projection of the many - body wave function onto a submanifold of the full hilbert space spanned by a set of trial wave functions .
specifically , we employ a variational state in the form of a product of coherent states @xcite @xmath45 where @xmath46 are the coherent amplitudes , @xmath47 is a global phase which ensures energy conservation , and @xmath48 denotes the vacuum of bogoliubov quasiparticles .
the ansatz provides an exact solution when describing the sudden immersion of an impurity of infinite mass into a gas of non - interacting bosons .
as such the choice of the wave function is based on an exact limit of the model which is valid for arbitrary interaction strength between the impurity and bose gas . before turning to the dynamics , we consider the parameters in eq . as time - independent variables and study the variational energy @xmath49 of the system .
its variation yields the polaron energy which for zero total momentum , @xmath50 , reads @xcite @xmath51 where @xmath52 defines the shift of the scattering resonance due to the many - body environment .
we note that the expression correctly accounts for the regularization of the short - range interaction as given by eq .
and , accordingly , contains only experimentally accessible parameters .
the attractive and repulsive polaron branches described by eq . are shown in fig .
[ fig : spectrum ] and play an important role in the dynamics as we describe below . for the time - dependent problem
we promote the variational parameters @xmath53 and @xmath47 to time dependent quantities .
the equation of motions @xmath54 for the parameters @xmath55 are obtained from the lagrangian @xmath56 of the system .
they can be cast in the form @xmath57\right)}{m_i}\right ) \beta_{\v k } \nonumber\\ & + & \frac{g_{\lambda}}{2 } \left[w_{\v k}\sum_{\v k ' } w_{\v k'}\left ( \beta_{\v k'}+ \beta_{\v k'}^*\right ) + w_{\v k}^{-1}\sum_{\v k ' } w_{\v k'}^{-1}\left ( \beta_{\v k'}- \beta_{\v k'}^*\right ) \right ] \nonumber\\ \dot \phi(t ) & = & g_{\lambda } n + \frac{1}{2 } g_{\lambda}\sqrt{n } \sum_{\v k } w_{\v k } \left ( \beta_{\v k } + \beta_{\v k}^*\right ) + \frac{{\v p}^2 -\v p_b^2 \left [ \beta_{\v k}\right]}{2m_i}\end{aligned}\ ] ] where @xmath58=\sum_{\v k } \v k \abs{\beta_{\v k}}^2 $ ] is the total momentum of the phonons . in eq .
@xmath12 is defined by the lippman - schwinger equation with given scattering length @xmath20 and a finite uv cutoff @xmath59 .
this ensures a time - evolution which is fully regularized and free of any divergencies .
[ [ absorption - spectra . ] ] * absorption spectra*. + + + + + + + + + + + + + + + + + + + + + + in the class of coherent states , eq .
, the dynamical overlap becomes @xmath60 $ ] . from its fourier
transform we obtain the rf absorption spectrum , shown in fig .
[ fig : spectrum ] , as function of the inverse interaction strength @xmath61 .
the spectrum exhibits two main excitation branches : the attractive polaron for @xmath62 and the repulsive polaron for @xmath63 .
both follow the energy eq .
( indicated as green lines in fig .
[ fig : spectrum ] ) , and in the weak coupling regime both are well described by a frhlich polaron model with renormalized interaction parameters @xcite .
the attractive polaron is formed when the impurity is dressed by bosonic excitations due to the weak attractive interactions with the bath for @xmath62 .
the corresponding spectral signature , shown in fig .
[ fig : spectrumcuts](a ) for @xmath64 , is a sharp quasiparticle peak at negative frequencies given by eq . .
as the feshbach resonance at @xmath42 is approached the attractive polaron peak looses spectral weight to the scattering continuum at higher frequencies .
close to unitarity , no particular eigenstate of @xmath65 yields a distinct contribution to the dynamical overlap @xmath40 , and many overlaps between the eigenstates of the many - body hamiltonian and the non - interacting state of the system are of the same order . hence the spectrum becomes broad and no coherent quasiparticle excitation is possible any longer .
in consequence , perturbative approaches based on expansions around the non - interacting state become particularly unreliable in this strong coupling regime . for positive scattering length ,
i.e. @xmath66 , the effective interaction between the mobile impurity and the bosons is repulsive and leads to the formation of a repulsive polaron .
this state manifests itself as a quasiparticle peak at positive frequency . as can be seen in fig .
[ fig : spectrum ] , the energy of the repulsive polaron increases as the shifted resonance at @xmath67 is approached , and it follows the saddle point prediction eq . .
similar to the attractive polaron , close to resonance , the repulsive polaron quickly looses quasiparticle weight .
however , the spectral weight is transferred not only to incoherent excitations but also to coherent spectral features which appear below the repulsive polaron branch . in fig .
[ fig : spectrum ] and fig .
[ fig : spectrumcuts](c ) those features are visible as a series of equidistant peaks .
such excitations corresponding to multiple bosons bound to the impurity are absent in the frhlich model since they are a consequence of strong pairing correlations which originate from the quadratic interaction terms in eq . .
as unitarity is approached , the spacing between these bound state peaks decreases until they eventually can not be resolved .
we note that such a crossover in the spectral profile from discrete bound state levels to a broad distribution is reminiscent of the formation of superpolarons which has been studied in rydberg molecular systems @xcite .
[ [ many - body - bound - states . ] ] * many - body bound states . * + + + + + + + + + + + + + + + + + + + + + + + + + + + the emergence of the series of bound states on the repulsive side of the resonance ( @xmath68 ) is a novel feature of impurities immersed in atomic becs .
each peak corresponds to a single , two , and more bogoliubov quasiparticles bound to the repulsive polaron .
the structure of the bound state spectrum can be understood analytically .
we consider a wave - function which accounts for a single bogoliubov excitation above the polaron state @xmath69 @xcite @xmath70 the ansatz eq . can be regarded as a molecular wave function which fully accounts for two - body bound state physics on top of the repulsive polaron state @xmath69 .
a calculation shows that the equations of motion of this state have an eigenmode @xmath71 with eigenfrequency @xmath72 determined by the solution of the equation ( a derivation is given in the sm @xcite ) @xmath73=0\ ] ] where @xmath74 . in fig .
[ fig : spectrum ] we show the energy @xmath72 as a dashed red lines .
the lines occur in integer multiples of @xmath72 as the bound state can be occupied by several phonons at the same times , an effect taken into account by the exponentiated creation operators in eq . .
for a further understanding of these states we first consider the limit of the bec density going to zero , ie .
@xmath75 , @xmath76 , @xmath77 and @xmath78 .
this limit defines the two - body problem where , for zero - range interactions , the bound state energy becomes @xmath79 @xcite , which is fully recovered by eq . .
for any finite density of non - interacting bosons , and assuming an infinitely heavy impurity , this bound state can be occupied by arbitrarily many bosons , and each bound atom contributes an energy @xmath80 . in an interacting bose gas , the infinitely massive impurity scatters with bogoliubov quasiparticles instead of bare bosons . as a consequence
the binding energy @xmath80 is modified . since in the bogoliubov model quasiparticles
do not interact with each other they can still occupy the bound state multiple times which leads to the series of spectral lines visible in fig . [
fig : spectrumcuts](c ) .
the emergence of the sequence of many - body bound states revealed in our approach is related to the efimov effect @xcite .
indeed , an exact solution of the corresponding three - body problem reveals that , as the infinite mass condition is relaxed , recoil leads to the splitting of three- , four- , etc .
body bound states into an infinite series of bound states situated in a regime exponentially close to the feshbach resonance @xcite . within our approach
this splitting is absent since the effective interactions between phonons vanishes for the class of coherent states eq . .
in consequence , in the limit of vanishing density , we effectively recover the efimov physics of an infinitely mass - imbalanced system , as discussed by efimov in his seminal work @xcite . [ [ comparison - to - other - approaches . ] ] * comparison to other approaches*. + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + as can be seen from the previous analysis , a simple ansatz such as eq .
can already account for aspects of complex many - body bound state physics .
indeed , the ansatz is related to a variational wave function which is based to an expansion in terms of single particle excitations . in the context of cold atoms ,
it was first introduced in the work by chevy for fermionic systems @xcite , and later generalized to bosonic systems @xcite . to connect our results to previous work we present here an extension of the equilibrium approach to real - time dynamics by studying the time - dependent variational wave function @xmath81 this wave - function accounts for a single phonon excitation on top of the unperturbed bec state .
we calculate the excitation spectrum of the system from the dynamical overlap @xmath82 as obtained from the equation of motions of the variational parameters .
this ansatz is naturally connected to the coherent state approach as an expansion in low occupation numbers , which justifies the validity of eq . in the limit of low densities .
while in the weak coupling regime the simple ansatz reproduces the predictions using the coherent state approach , it fails to describe the intricate many - body physics in regimes where multiple boson excitations become relevant ( for a comparison see the sm @xcite ) . , for attractive and repulsive interactions , ( a ) @xmath83 , ( b ) @xmath84 , ( c ) @xmath85 , obtained by the coherent state approach and the single - excitation expansion ( yellow solid and red dashed lines , respectively ) . ]
the difference between the coherent states approach and the single - excitation expansion can be highlighted when the time evolution of the number of phonon excitations @xmath86 is compared , where @xmath87 denotes the expectation value with respect to the chosen variational state . in fig .
[ fig : phnum ] we show @xmath88 for attractive and repulsive interactions . on the attractive side
both approaches predict a saturation of the phonon number in the long time limit , see panels ( a ) and ( b ) in fig .
[ fig : phnum ] . in the single - excitation expansion ( dashed lines ) ,
the number of excitations is restricted to one .
this limitation of eq . becomes apparent when comparing it to the coherent state approach ( solid lines ) .
we find that already for moderate attractive coupling strength the number of phonons exceeds one .
hence , due to the restriction in the number of excitations for the ansatz , both approaches agree only in the initial short time evolution . on the repulsive side
the coherent state approach predicts oscillations of the number of phonons in the long time limit , see panel ( c ) in fig .
[ fig : phnum ] .
these oscillations appear due to the competition between the many - body polaron branch and the few - body bound states in real time .
in contrast , in the single - excitation expansion these oscillations decay gradually and in the long time limit , the number of excitations saturates at one , reflecting the formation of a single molecular state .
we note that similar to the time - dependent coherent state approach presented in this letter , a self - consistent t - matrix approach @xcite also accounts for an infinite number of bosonic excitations . using this diagrammatic approach
it had been found that the inclusion of multiple boson excitations has a profound influence on the spectrum .
however multiphonon bound states have not been observed in this earlier work .
yet we emphasize that when considering only moderate interaction strengths or short - time evolution , expansions in terms of a few particle excitations such as eq . remain a viable approach .
they correctly describe the weakly attractive and repulsive polaron branches as well as the one - boson bound state present in the spectrum sufficiently far away from the feshbach resonance .
[ [ summary - and - outlook . ] ] * summary and outlook*. + + + + + + + + + + + + + + + + + + + + + + + in summary , we analyzed the dynamics and absorption spectra of an impurity immersed in a bec .
we demonstrated both the disappearance of the sharp quasiparticle spectral feature at strong coupling and the presence of a novel type of excitations in which several bogoliubov quasiparticles are bound to the impurity particle .
our analysis highlights the importance of quasiparticle scattering processes that are not present in the commonly used frhlich model .
they result in strong short distance correlations that give rise to efimov type physics of multi - particle bound states and play a crucial role in suppressing the quasiparticle spectral weight close to the feshbach resonance .
our work opens new directions for studying non - perturbative phenomena in bose polarons at strong coupling .
we expect that new insight into such systems can be gained by extending our analysis to more sophisticated approaches such as gaussian variational wave functions @xcite and renormalization group analysis @xcite .
our method can also be extended to problems beyond linear response such as rabi oscillations of strongly driven bose impurities @xcite .
[ [ acknowledgment . ] ] * acknowledgment . *
+ + + + + + + + + + + + + + + + + + + + we thank dima abanin , gregory astrakharchik , immanuel bloch , ignacio cirac , thierry giamarchi , christoph gohle , deborah jin , marton kanasz - nagy , dries sels , tao shi , lars wacker , and martin zwierlein for inspiring discussions .
the authors acknowledge support from the nsf grant no .
dmr-1308435 , harvard - mit cua , afosr new quantum phases of matter muri , the aro - muri on atomtronics , aro muri quism program .
e. d. acknowledges support from the simons foundation , the humboldt foundation , dr .
max rssler , the walter haefner foundation , and the eth foundation .
r. s. is supported by the nsf through a grant for the institute for theoretical atomic , molecular , and optical physics at harvard university and the smithsonian astrophysical observatory .
f. g. is grateful for financial support from the gordon and betty moore foundation .
75ifxundefined [ 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 \doibase http://dx.doi.org/10.1063/1.452944 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.71.4103 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/0370-1573(94)00094-j [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/s0370-1573(00)00107-1 [ * * , ( ) ] @noop _ _ ( , ) link:\doibase 10.1103/revmodphys.81.1773 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/0370-1573(95)00085-2 [ * * , ( ) ] http://dx.doi.org/10.1016/0370-2693(70)90349-7 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/j.physrep.2006.03.001 [ * * , ( ) ] @noop _ _ ( , ) http://dx.doi.org/10.1038/nphys1389 [ * * , ( ) ] http://link.aps.org/doi/10.1103/physrevb.44.4693 [ * * , ( ) ] http://dx.doi.org/10.1103/physrevlett.97.180413 [ * * , ( ) ] http://dx.doi.org/10.1103/physrevlett.107.163202 [ * * , ( ) ] link:\doibase 10.1103/physrevb.86.064501 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.113.243002 [ * * , ( ) ] http://stacks.iop.org/1126-6708/2004/i=04/a=050 [ * * , ( ) ] http://dx.doi.org/10.1038/nature06080 [ * * , ( ) ] @noop _ _ ( , ) @noop _ _ ( , ) http://dx.doi.org/10.1070/pu1975v018n11abeh005234 [ * * , ( ) ] link:\doibase 10.1103/physrevb.37.1597 [ * * , ( ) ] @noop _ _ ( , ) link:\doibase 10.1103/physrev.156.207 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.85.483 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.103.150601 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.102.230402 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.103.170402 [ * * , ( ) ] link:\doibase 10.1103/physreva.77.011603 [ * * , ( ) ] http://dx.doi.org/10.1038/nature11151 http://www.nature.com/nature/journal/v485/n7400/abs/nature11151.html#supplementary-information [ * * , ( ) ] http://dx.doi.org/10.1038/nature11065 http://www.nature.com/nature/journal/v485/n7400/abs/nature11065.html#supplementary-information [ * * , ( ) ] link:\doibase 10.1103/physrevlett.108.235302 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.109.235301 [ * * , ( ) ] http://dx.doi.org/10.1038/nature12592 [ * * , ( ) ] http://dx.doi.org/10.1038/nphys2561 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.115.135302 [ * * , ( ) ] link:\doibase 10.1103/physreva.93.043607 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.96.210401 [ * * , ( ) ] link:\doibase 10.1088/1367 - 2630/9/11/411 [ * * , ( ) ] link:\doibase 10.1103/physrevb.80.184504 [ * * , ( ) ] link:\doibase 10.1103/physreva.82.063614 [ * * , ( ) ] http://stacks.iop.org/0953-4075/43/i=10/a=105301 [ * * , ( ) ] link:\doibase 10.1103/physreva.84.063612 [ * * , ( ) ] link:\doibase 10.1134/s1054660x11150035 [ * * , ( ) ] link:\doibase 10.1103/physreva.86.043614 [ * * , ( ) ] link:\doibase 10.1103/physreva.88.013613 [ * * , ( ) ] link:\doibase 10.1103/physreva.89.023612 [ * * , ( ) ] link:\doibase 10.1038/srep12124 [ * * , ( ) ] link:\doibase 10.1103/physreva.92.063635 [ * * , ( ) ] http://stacks.iop.org/1367-2630/17/i=3/a=033023 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.114.203001 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.115.160401 [ * * , ( ) ] link:\doibase 10.1103/physreva.93.043606 [ * * , ( ) ] link:\doibase 10.1103/physrevx.6.011012 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.82.1225 [ * * , ( ) ] http://dx.doi.org/10.1038/nature08482 [ * * , ( ) ] link:\doibase 10.1103/physreva.88.053632 [ * * , ( ) ] link:\doibase 10.1103/physreva.90.013618 [ * * , ( ) ] http://dx.doi.org/10.1103/physrevlett.105.195301 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.115.125302 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.116.105302 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.116.053001 [ * * , ( ) ] http://prola.aps.org/abstract/pr/v90/i2/p297_1 [ * * , ( ) ] link:\doibase 10.1103/physreva.89.053617 [ * * , ( ) ] @noop _ _ link:\doibase 10.1103/physrevx.2.041020 [ * * , ( ) ] @noop http://www.sciencedirect.com/science/article/pii/0375960179901518 [ * * , ( ) ] link:\doibase 10.1140/epjb / e2012 - 30841 - 3 [ * * , ( ) ] @noop * * , ( ) http://dx.doi.org/10.1038/nphys1253 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/0375-9474(73)90510-1 [ * * , ( ) ] link:\doibase 10.1103/physreva.74.063628 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.111.265302 [ * * , ( ) ] @noop _ _ ( , ) |
elemental abundances contain significantly more information than just ` metallicity ' , since different elements are synthesised in stars of different masses , and on different timescales .
in particular , oxygen and the alpha - elements are produced predominantly by massive stars , ejected into the interstellar medium by core - collapse , type ii , supernovae .
iron has important contributions by both type ii supernovae and type ia supernovae , the latter believed to result from the explosive nucleosynthesis of degenerate material , associated with a white dwarf in a binary system .
the vast majority of local stars belonging to the galactic halo population have enhanced values , compared to that of the sun , of the ratio of the abundances of oxygen and of the alpha - elements to that of iron ( e.g. wheeler , sneden & truran 1989 ; nissen et al .
1994 ; gratton et al .
further , these halo stars have very similar values of this ` over - abundance ' of alpha - elements , [ @xmath1/fe]@xmath3 .
this ` halo ' value is , within nucleosynthetic yield uncertainties , just that predicted for chemical enrichment by a mean , imf - averaged , type ii ( core collapse ) supernova , assuming a normal ( i.e. solar - neighborhood ) massive - star mass function ( wyse & gilmore 1992 ; nissen et al .
1994 ) .
thus a typical halo star shows no evidence for nucleosynthetic products from type ia supernovae .
this lack restricts the duration of star formation that created these stars to be shorter than the time it takes for there to have occurred significant type ia explosions .
this timescale is rather model - dependent . in the commonly - discussed double - degenerate models ( iben & tutukov 1984 ;
see iben et al .
1997 for a recent discussion , including the expected statistics of close wd binaries ) it is plausibly of order 1 gyr after the formation of the progenitor stars ( e.g. smecker - hane & wyse 1992 ; matteucci & franois 1992 ) .
a similar timescale for significant supernovae follows from models in which the white dwarf is accreting due to roche - lobe overflow from a low - mass , evolved companion ( yungelson & livio 1998 ) .
this restriction on the duration of star formation is not necessarily the same as a restriction on the global age range of the halo , since different regions of the halo could initiate ( short - lived ) star formation at different times , resulting in a significantly larger total age range in the halo ( see also gilroy et al .
1988 for an analogous argument for rapid local enrichment , based on the appearance of @xmath4process elements in metal - poor halo stars ) .
it is very implausible that the duration of star formation across the entire halo was as short as 1 gyr , no matter one s preconceptions of how the halo formed
if ` monolithic ' , this requires remarkable synchronisation over @xmath5 kpc , while if ` chaotic ' , the regions that formed the halo had remarkably uniform star formation properties .
star formation is likely to occur in transient overdense regions , perhaps of the mass of present - day giant molecular cloud complexes in the galactic disk , even in models that envisage a well - defined ` monolithic ' galactic potential well during the formation of the stellar halo ( e.g. fall & rees 1985 ) .
was there pre - enrichment of one region by another ? did some regions self - enrich sufficiently long to incorporate the products of type ia supernovae ?
isochrone - based determinations of the ages of field halo stars are of limited accuracy , and indeed the only technically - feasible means of detecting an extended duration of star formation that may be as short as a gyr is to identify halo stars with evidence for pre - enrichment by type ia supernovae .
the manifestation of this is lower values of [ @xmath1/fe ] than can be understood from type ii supernovae alone .
given the expectation of some increase in metallicity with time as star formation proceeds , the most natural place to look for such an effect is among the most chemically - enriched halo stars .
we note that a variation in the massive - star imf can also provide for lower values of element ratios , still with enrichment by only type ii supernovae
. however , there is firm observational evidence for universality of the imf , surprising as this may be ( see papers in gilmore & howell 1998 ) , and indeed any invoked imf variation to explain lower values of [ @xmath1/fe ] would be extremely _ ad hoc_. thus we favor interpretations within the framework of a fixed imf .
several recent analyses , most notably that by nissen & schuster ( 1997 ) , have discovered exactly the element ratio behavior expected for some enrichment by type ia supernovae .
the authors of these analyses have favored , as the explanation for ` low - alpha ' halo stars , accretion of stars from distinct objects such as satellite dwarf galaxies . here
we argue rather that the observations are more naturally explained by the variation in the duration of star formation in regions of the proto - halo .
fragmentation into cool , dense self - gravitating regions is fairly generic to models of the collapse of galaxy - scale perturbations ; assimilation and mixing of these transient regions is not to be thought of as accretion .
of course , accretion , in the sense of a distinct other proto - galaxy merging into our proto - galaxy , remains a very attractive explanation for the creation of the thick disk ( e.g. gilmore & wyse 1985 ; see velazquez & white 1998 , for a recent simulation ) , some @xmath6 gyr ago ( e.g. gilmore , wyse & jones 1995 ) .
halo stars with low values of the [ @xmath1/fe ] element ratios were identified in a study of metal - rich halo stars , those with [ fe / h ] @xmath7 dex , by nissen & schuster ( 1997 ) , and serendipitously in samples of local , low - metallicity halo stars by king ( 1997 ) and by carney et al .
nissen & schuster studied approximately equal numbers of ` disk ' and ` halo ' dwarfs ( 16 disk , 13 halo ; so designated based on kinematics ) in the metallicity range @xmath8 \simlt -0.4 $ ] .
two giant stars in each of the ` younger ' ( e.g. richer et al .
1996 ) globular clusters ruprecht 106 and pal 12 were analysed by brown , wallerstein & zucker ( 1997 ) and found to have approximately solar values of alpha - element - to - iron ratios .
figure 1 shows the [ mg / fe ] data , as a function of [ fe / h ] , for the nissen & schuster ( 1997 ) sample of disk and halo stars , plus the anomalous metal - poor halo stars from king ( 1997 ) and carney et al .
( 1997 ) ; the shaded region indicates the locus of the majority of metal - poor halo stars ( e.g. nissen et al .
1994 ; gratton et al . 1997a
; note that there are sometimes zero - point offsets in the value of [ mg / fe ] between different observational data sets , which can lead to apparent gradients in [ @xmath1/fe ] plotted against [ fe / h ] if all published data are naively combined .
each data set shows similar behavior in their regions of overlap in [ fe / h ] .
since it is probable that these offsets reflect differences in analysis methods rather than real systematic variations in [ @xmath1/fe ] as a function of [ fe / h ] , we do not consider them further here ) .
as may be seen in the figure , the majority of the metal - rich halo stars observed shows significantly lower values of [ @xmath1/fe ] than the enhanced , super - solar values of the alpha - element ratios derived for the more metal - poor halo stars .
thus the halo stars with enhanced element ratios are ` anomalous ' in this metallicity range .
this contrasts with the typical ( metal - poor ) halo where the stars with low values of [ @xmath1/fe ] are the anomalous ones .
the halo stars of all metallicity with low values of [ @xmath1/fe ] share one property
they tend to be on orbits with large values of apogalactic distances , @xmath9 kpc , whereas the ` normal ' halo stars are not to so biased to the outer halo .
this trend suggests a real physical meaning to the present orbital parameters .
further , all the halo stars of the nissen & schuster ( 1997 ) sample are on derived orbits with very low perigalactic distances , @xmath10 kpc ; this is consistent with the apparent correlation between @xmath11 and metallicity established for such kinematically - selected samples by ryan & norris ( 1991 ) , such that the more metal - rich halo stars tend to be on orbits of smaller perigalactic distances .
thus all the metal - rich halo stars observed are on orbits with small values of @xmath11 , while of these , the stars with low values of [ @xmath1/fe ] are on orbits with large values of @xmath12 .
note that it is _ not _ the case that all metal - poor halo stars with large @xmath13 necessarily have low values of [ @xmath1/fe ] , although such a trend exists for the metal - rich stars with low values of [ @xmath1/fe ] . as demonstrated in table 10 of carney et al .
( 1997 ) there are many metal - poor stars that have orbits associating them with the outer halo , but that have normal , enhanced , values of the alpha - element ratios .
the metal - poor halo stars with low values of [ @xmath1/fe ] are on retrograde orbits ( king 1997 ; carney et al .
1997 ) , but there is no such signature in the kinematics of the metal - rich stars with low values of [ @xmath1/fe ] .
we wish to understand why a small fraction of stars in the stellar halo , biased to metal - rich stars on high - energy , high - eccentricity orbits , have low values of [ @xmath1/fe ] .
consider the pattern of element ratios in self - enriching systems ( of fixed imf ) with given star formation rates .
the solid lines in figure 1 illustrate schematically the trends produced in simple ( homogeneous ) models of chemical evolution with differing star formation rates .
the important feature is the location of the turndown due to iron from type ia supernovae since the _ shortest _ pre - explosion lifetime of type ia supernovae is presumed to be a fixed value , the location of this turndown occurs at a higher value of [ fe / h ] for a higher star formation rate , reflecting the more rapid metallicity increase with time ( note that the effect of gas flows on the location of the turndown from the ` type ii plateau ' at early times / lower metallicities can be incorporated by a suitably - modified star - formation rate ; inflow of unenriched material will reduce [ fe / h ] keeping [ @xmath1/fe ] fixed .
outflow driven by type ii supernovae will behave similarly ) .
such an effect has been proposed to explain the element ratios seen in local disk stars , born at different locations in the disk ( edvardsson et al .
1993 ) .
figure 1 shows that the observed differences in element ratio patterns are inevitable consequences of a variation of the star formation rate .
the amplitude of the invoked variation in star formation rates is required only to be such as to allow up to @xmath14 gyr longer duration of star formation to produce lower values of [ @xmath1/fe ] .
this difference in age ranges is undetectable with current isochrone - based techniques .
note that there is no requirement from the element ratio data that the progenitor stars of the type ia supernovae be part of the same star forming region as the gas they will eventually enrich .
however , the observed trend for low values of [ @xmath1/fe ] to be observed for higher - metallicity halo stars favors rather self - enrichment scenarios , in ( sub)structure of relatively deep potential wells . clearly , the data are consistent with the lower [ @xmath1/fe ] stars forming over longer timescales , allowing incorporation of iron from type ia supernovae . but where did these halo stars form ? in a distinct separate galaxy , or essentially where we observe them now ?
nissen & schuster ( 1997 ) considered two possibilities , either that the outer regions of the stellar halo of the galaxy had a longer duration of star formation than did the inner halo , or that the halo stars with low values of [ @xmath1/fe ] were accreted from a dwarf galaxy within which star formation is presumed to have been sustained over a sufficiently long period , longer than in the field halo .
the authors favor the latter explanation . indeed , ` distinct accretion events ' are the preferred explanation of all the authors for the data in figure 1 , and also for the ` anomalous ' globular clusters ( it has been suggested that these two clusters have been captured from the large magellanic cloud ; lin & richer 1992 ) .
motivation for the interpretation in terms of ` accretion ' came partially from the prediction of low values of the alpha - element to iron ratios in the stars of dwarf galaxies in which there is extended star formation , perhaps even successive bursts of star formation , as inferred for many of the extant satellite galaxies associated with the milky way ( gilmore & wyse 1991 ; unavane , wyse & gilmore 1996 ) .
it is clear from observations that only the more massive dwarf systems can self - enrich to a metallicity of 1/10 the solar value ( see , for example the metallicity luminosity ( mass ) relation of lee et al .
1993 ) . of the local dwarf galaxy companions to the milky way , only the fornax dsph , the sagittarius dsph and the magellanic clouds have a significant population of stars as metal - rich as the observed low - alpha halo stars .
however , these systems also contain stars of a wide range of ages ( e.g. smecker - hane 1996 ) , and late accretion of stars from systems like these satellite galaxies would produce stars in the halo which are _ many _ gyr younger than the bulk of the halo , reflecting the star - formation history of the satellite galaxy parent .
the present sample of ` low - alpha ' halo stars shows no obvious signature of having a relative youth of this amplitude , which would be manifest in their colors ( e.g. unavane , wyse & gilmore 1996 ) .
thus late accretion from systems like the extant satellite galaxies is not favored .
what other kind of accretion is possible ?
models of the formation of the stellar halo generically invoke early star formation in transient @xmath15 m@xmath16 local potential wells , perhaps reflecting the initial primordial fluctuation power spectrum , or formed by thermal and/or gravitational instability within a larger system . the turn - around , collapse and relaxation to dynamical equilibrium of the present - day galactic potential takes at least a couple of global dynamical times , or a few gyr .
thus disruption and assimilation of transient systems during the relaxation of the large - scale galactic potential should not be considered as ` accretion ' , since these systems never really had a separate identity .
systems that do retain coherence through this period can be said to be ` accreted ' if assimilated subsequently , and it is these we consider next .
since we are discussing essentially old stars , we shall focus on the stellar components of this substructure ; in any case a satellite galaxy would be most likely stripped of its gas after a few passages through the gaseous disk of the milky way with perigalactic distances like those observed for the halo stars . `
substructure ' orbiting within the ( dark halo ) potential of a larger ` parent ' galaxy is subject to tidal forces , and to dynamical friction .
the mass and the ( mean ) density of the substructure will to a large extent determine its fate .
dynamical friction will cause the orbit of the substructure to decay in radius , on a timescale inversely proportional to the mass of the substructure ( for given orbit and large galaxy ) ; relatively massive substructure will sink to the center of the larger galaxy on a few orbital times .
tidal stripping will cause the substructure to lose material when its internal gravity is overcome by the gravitational field of the larger galaxy ; the simple jacobi criterion implies that a system will lose material unless ( or until ) its internal density is of order three times the mean density of the larger galaxy interior to its orbit ; simulations have shown that this simple criterion works remarkably well ( e.g. johnston , hernquist & bolte 1996 ) .
diffuse substructure will lose material while orbiting even in the far outer regions of the larger galaxy .
the orbits of stars that have been accreted into the galaxy by tidal effects on a satellite galaxy are only slightly different from that of the satellite at the time the stars were captured ; this produces a kinematic signature of moving groups of stripped stars , that can persist for many orbital times ( cf .
tremaine 1993 ; johnston 1998 ) .
thus one may consider four broad classes of dissipationless ( stars plus dark matter ) substructure : massive , dense systems will survive fairly intact while sinking to the center of the larger galaxy ; massive , diffuse systems will be largely disrupted by tides during their orbital decay ; low mass , dense systems will be largely unaffected , while low mass , diffuse systems will be disrupted with no orbital decay .
the stars under consideration here are on rather eccentric orbits , but with even @xmath12 much less than that of typical extant satellite galaxies . as discussed by velazquez & white ( 1998 ) , for example , in their study of the interaction between satellite galaxies and large disk galaxies , _ local _ application of the standard ( chandrasekhar ) dynamical friction formula
provides reliable results in the case of non - circular orbits .
thus due to the higher densities in the central regions of the bigger galaxy , the frictional effects are higher there .
further , most tidal stripping occurs at pericenter passage ( e.g. velazquez & white 1998 and references therein ) .
thus if the observed halo stars were removed from a satellite galaxy by tidal stripping , their very low perigalactic distances would require that the satellite galaxy be able to penetrate deep inside the milky way .
their low values of @xmath12 also would , in this scenario , require a rather massive satellite , so that dynamical fraction can operate on less than a hubble time .
the perigalactic distances provide the more stringent constraint . as discussed above , the low [ @xmath1/fe ]
stars are on radial orbits , with perigalactic distances @xmath17 kpc , and indeed the majority of the stars having @xmath10 kpc . a lower limit to the mean density of the milky way galaxy interior to 1 kpc
is given by adopting just the mass of the central bulge , @xmath18 ( e.g. kent 1992 ) .
the predominant stellar populations in the galactic bulge and inner disk are probably at least @xmath19 gyr old ( ortolani et al .
1995 ; holtzman et al . 1993 ) , consistent with theoretical expectations for the timescale of assembly of the ` core ' of the milky way in hierarchical - clustering models ( e.g. lacey & cole 1993 ) .
thus one may use inferences from the mass distribution of the inner galaxy to constrain the fate of substructure back to that time . the mean density interior to the perigalactic approach ( @xmath20 kpc ) of the ` low - alpha ' stars is then @xmath21 m@xmath16 pc@xmath22 . to be robust against tides ,
a satellite galaxy on an orbit with this small a perigalactic distance would need an mean internal density of around three times this , or @xmath23
m@xmath16 pc@xmath22 .
this value is significantly larger ( by orders of magnitude ) than the typical mean densities derived for dwarf galaxies , both gas - poor and gas - rich ( e.g. puche & carignan 1991 ; ibata et al .
thus essentially all dwarf galaxies as we see them now ( presumably the survivors ) would be disrupted after only a couple of such pericenter passages ( see johnston , hernquist & bolte 1996 ) . indeed , they would likely be disrupted significantly earlier in their orbital decay , when the pericenter distances are of order the present derived apocenter distances of the anomalous halo stars ( e.g. ibata et al .
1997 ) .
thus , if the halo stars under discussion represent stars captured from a satellite galaxy , that galaxy had to be sufficiently massive to spiral into the center of the milky way in less than a hubble time this implies a mass perhaps 10% that of the older stars in the central milky way , or equal to the entire mass of the stellar halo ( e.g. carney , latham & laird 1990 ) .
this satellite had also to be significantly more robust than any of the extant satellite galaxies ( the survivors ) .
if this satellite contributed to the formation of the thick disk , it did so without donating any of its own stars to the disk .
each of these is rather surprising .
the requirements for the progenitor structures for the metal - rich , ` low - alpha ' halo stars may be summarised as being dense enough to survive to a perigalactic distance of @xmath20 kpc , and have a deep enough local potential well to self - enrich to @xmath24 dex , while sustaining star formation for @xmath20 gyr , being robust not only to internal feedback from star formation , but also external effects .
plausibly the first two of these requirements are satisfied by a density criterion for the substructure , given that the escape energy per unit mass scales as @xmath25 .
sustaining star formation requires gas retention , and the further requirement that the effects of collisions with other star - forming clouds , or passages through the ( forming ) gas disk of the galaxy , be minimised . as discussed by many authors ( see e.g. gunn 1980 ; jones , palmer & wyse 1981 ) gravitational instability in the cooling layers behind shocked gas in the haloes of protogalaxies produces systems of density @xmath26m@xmath16pc@xmath22 and of characteristic ( jeans ) mass of a few times @xmath27m@xmath16 .
a more sophisticated analysis by fall & rees ( 1985 ) included thermal instability in the hot , gaseous halo of a proto - galaxy , like the milky way , to produce cool , dense regions , that are seeds for gravitational instability .
fall & rees ( 1985 ) argue that a fairly massive proto - galaxy , with circular velocity greater than @xmath28 km / s , is required for this scenario ( essentially to provide the hot phase of the gaseous halo ) . in flat cold - dark - matter dominated cosmologies ( e.g. white & frenk 1991 ) , dark haloes of this circular velocity have peak abundance at redshifts @xmath29 , or lookback times of @xmath30 gyr .
thus this picture of fragmentation inside dark haloes is reasonable even in hierarchical clustering scenarios .
the fragments undergo star formation and are subsequently dispersed , either by the internal effects of star formation , or the external effects of cloud - cloud collisions or tidal disruption by the galactic potential .
it is plausible that the fragments probing smaller galactocentric distance have larger internal densities ( indeed , as predicted by the model of fall & rees 1985 ) and larger escape velocities , allowing greater self - enrichment and thus providing a relation between r@xmath31 and [ fe / h ] as observed in the remnant stars , now the field halo ( higher metallicity for lower r@xmath31 ; ryan & norris 1991 ) .
the most dense regions could plausibly form globular clusters , as suggested by fall & rees ( 1985 ) , even some of which are subsequently destroyed ( e.g. gnedin & ostriker 1997 ) . the bias to large apogalactic distances for the orbits of the ` low - alpha ' stars follows naturally from the expected longer survival times of large , self - enriching clouds in the outer regions of proto - galaxy , due to the longer collision times there , and the larger orbital times .
additionally , a large - scale density gradient could provide slower local star formation rates in the outer regions , provided the properties of density perturbations correlate with their local background .
thus fragmentation would appear to provide the kind of substructure required by the observations .
a striking aspect of the data in figure 1 is that the value of the ` type ii plateau ' of element ratios is the same for the halo stars as for the disk stars of the same metallicity .
following the same logic as wyse & gilmore ( 1992 ) , this implies that the imf of the ( massive ) stars that enriched these stars was the same in the early disk and halo .
inversion of the value of the element ratios to determine the actual slope of the massive star imf requires adoption of theoretical supernova yields and is hence uncertain , but the robust statement that the imf was invariant in the enrichment of disk and halo may be made immediately .
it is also clear from figure 1 that most of the metal - rich halo stars do not have element ratios equal to this ` type ii plateau ' , and that the _ typical _ halo and disk stars of equal metallicity ( [ fe / h ] @xmath32 dex ) do _ not _ have the same values of the element ratios .
there is not the smooth continuity in chemical enrichment from halo to disk often derived in models of chemical evolution .
of course , these halo stars are on orbits of very different angular momentum than those of the disk stars , further emphasizing the distinction between local samples of disk and halo , and the low likelihood that the halo , as sampled , pre - enriched the disk ( see wyse & gilmore 1992 ; gilmore 1995 ) .
there is also a spread in [ fe / h ] at fixed values of [ mg / fe ] ; as mentioned earlier in section 3 , this may be achieved by different gas flows / mixing and star formation histories .
the onset of star formation in the local disk , as measured by the white dwarf luminosity function , has most recently been estimated to be 1012 gyr ago ( an increase over earlier estimates , in part due to the inclusion of phase separation and crystallisation energy in the white dwarf cooling curves ; hernanz et al .
1994 ; chabrier 1998 ) .
this older age is in agreement with the ages of the oldest open clusters ( e.g. phelps 1997 ) and the age of the thick disk ( e.g. gilmore , wyse & jones 1995 ) and in better agreement with the ages of individual stars from edvardsson et al .
( 1993 ; see also ng & bertelli 1998 for post - hipparcos age estimates for stars from this sample ) .
the recent downward revision in the ages of the halo globular clusters to @xmath6 gyr , based on hipparcos calibrations of the luminosities of field subdwarfs ( e.g. gratton et al . 1997b ; reid 1998 ) , leaves little room for a significant hiatus between the epoch of halo star formation and the onset of star formation in the disk ( however , this calibration depends on many corrections and is not unique see e.g. pont et al .
it would follow that disk and halo were independent entities from the very early stages of galaxy formation .
the recent observations of halo stars with elemental abundances reflecting pre - enrichment from intermediate - mass stars and type ia supernovae can be understood in many models of halo formation .
the establishment of trends with kinematics , suggested at in the present data , will be important in distinguishing models , as will be age determinations .
the fact that the ` low - alpha ' halo stars tend to be metal - rich favors self - enrichment scenarios , and the survival of individual star - forming complexes for approximately a gyr . that the observed low - alpha stars are on orbits that plunge close to the galactic center also implies that a reasonably robust parent system is required .
we argue for an interpretation that simply appeals to a small variation in the duration of star formation within regions of the stellar halo , rather than ` distinct accretion events ' .
the present data implicates a significant fraction of the metal - rich , outer halo .
one may normalise to the stellar halo as a whole , using the metallicity distribution for local halo stars of carney et al .
( 1996 ) in which @xmath33% of the stellar halo has [ fe / h]@xmath34 dex , combined with the halo structural parameters of unavane , wyse & gilmore ( 1996 ) in which @xmath35% of the halo is located beyond @xmath36 kpc , to conclude that at most a few percent ( 10% times 15% ) of the stellar halo is being discussed .
more data not restricted to detailed elemental abundances for selected halo stars will clearly help to substantiate these estimates .
for example , the metallicity distribution of the halo is derived from local , kinematically - selected samples , and the inner halo is poorly studied .
further , there is currently little information about the _ wings _ of the metallicity distribution of the stellar halo . the present data for disk and halo stars of ` overlapping ' metallicities [ fe / h]@xmath37 dex show a disk halo dichotomy , in that the elemental abundances are different for typical disk stars and for typical halo stars in this metallicity range , with the halo stars having lower values of [ @xmath1/fe ] than the disk stars .
this complicates the requirements for chemical evolution models for the disk that appeal to gas inflow from the halo ; indeed if the halo had ejected gas that pre - enriched the disk , one might have expected the opposite situation , with the disk stars having lower values of element ratios than the halo stars .
a simplification of chemical evolution models is that the elemental abundance data favor a fixed massive - star imf for both disk and halo .
carney , b. , latham , d. & laird , j. 1990 , aj , 99 , 572 .
carney , b. , wright , e. , sneden , c. , laird , j. , aguilar , l. & latham , d. 1997 , aj , 114 , 363 carney , b. , laird , j. , latham , d. & aguilar , l. 1996 , aj , 112 , 668 carney , b. , latham , d. , laird , j. & aguilar , l. 1994 , aj , 107 , 2240 chabrier , g. 1998 , in iau symposium 189 , ` fundamental stellar properties : the interaction between theory and observation ' , eds t.r .
benning , a.j .
booth & j. davis ( kluwer : dordrecht ) in press edvardsson , b. , andersen , j. , gustafsson , b. , lambert , d. , nissen , p. & tomkin , j. 1993 , 275 , 101 fall , s.m . &
rees , m.j .
1985 , apj , 298 , 18 gilmore , g. 1995 , in iau symposium 164 , ` stellar populations ' , eds g. gilmore & p. van der kruit ( kluwer , dordrecht ) p99 gilmore , g. & wyse , r.f.g .
1985 , aj , 90 , 2015 gilmore , g. & wyse , r.f.g .
1991 , apjl , 367 , l55 ibata , r. , wyse , r.f.g . ,
gilmore , g. , irwin , m. & suntzeff , n. 1997 , aj , 113 , 634 .
iben , i. & tutukov , a.v .
1984 , apjs , 54 , 335 iben , i. , tutukov , a.v . & yungelson , l. 1997 , apj , 475 , 291 johnston , k. , hernquist , l. & bolte , m. 1996 , apj , 465 , 278 puche , d. & carignan , c. 1991 , apj , 378 , 487 reid , i.n .
1998 , aj , 115 , 204 richer , h. , harris , w.e . , fahlman , g. , bell , r. , bond , h.e .
, hesser , j.e . ,
holland , s. , pryor , c. , stetson , p. , vandenberg , d.a . & van den bergh , s. 1996 , apj , 463 , 602 ryan , s. & norris , j. 1991 , aj , 101 , 1835 smecker - hane , t. a. 1996 , in ` star formation near and far ' aip conf proc vol 393 , eds .
s. holt & l. g. mundy ( aip press : new york ) , p571 | it is well - established that the vast majority of metal - poor galactic halo stars shows evidence for enrichment by solely massive stars .
recent observations have identified more varied behavior in the pattern of elemental abundances measured for metal - rich , [ fe / h]@xmath0 dex , halo stars , with both high and low values of the ratio of [ @xmath1/fe ] .
the low values are most naturally due to the incorporation of iron from type ia supernovae .
these ` low - alpha ' halo stars have been interpreted as being accreted from dwarf galaxies . however , these stars are on very high - energy radial orbits , that plunge into the galactic center , from the outer regions of the stellar halo .
we demonstrate here that known dwarf galaxies could not reproduce the observations , which require the proposed parent dwarf to have a very high mean density to provide ` low - alpha ' stars on orbits of such low values of perigalactic distance .
rather , the observations are consistent with fragmentation of a gaseous proto - halo into transient star - forming regions , some of which are sufficiently dense to survive close to the galactic center , and self - enrich to relatively high metallicities .
those that probe the outer halo sustain star formation long enough to incorporate the ejecta from type ia supernovae .
further , we point out that the values of the ` type ii plateau ' in the element ratios for disk stars and for halo stars are equal .
this implies that the ( massive ) stars that enriched the early disk and halo had the same imf .
however , the bulk of the halo stars with [ fe / h]@xmath2 dex that have been observed have significantly lower values of [ @xmath1/fe ] than does a typical disk star of that metallicity .
thus there is a discontinuity in the chemical enrichment history between the halo and disk , as probed by these samples , consistent with previous inferences based on angular momentum considerations . |
in the context of classical algebraic geometry , consider a given embedded complex projective manifold @xmath0 .
one of the typically used techniques was to replace @xmath1 by some of its smooth hyperplane sections , @xmath2 .
thus , the dimension of @xmath1 is decreased and classification results may be obtained inductively .
the efficiency of the method depends on the possibility of transferring some known special properties from @xmath3 to @xmath1 . in general ,
given @xmath4 , there is no _ smooth _
@xmath0 such that @xmath3 is one of its hyperplane sections .
one can say that _
@xmath1 is more special then @xmath3_. the present paper is a survey of contemporary aspects of the hyperplane section technique . a first important `` modern '' incarnation of the above principle is given by lefschetz s theorem , showing that the topology of @xmath3 strongly reflects that of @xmath1 ( see @xcite ) . from a geometrical point of view
, we are usually given some regular map , say @xmath5 , making @xmath3 _ special _ ; e.g. , a fano fibration .
we would like to extend this map to @xmath1 .
it was discovered by sommese , in his innovative early paper @xcite , that the extension is always possible , if one only assumes that the general fiber of @xmath6 has dimension at least two .
his proof is based on lefschetz s theorem and on ( very much related ) vanishing results of kodaira type . in the same paper ,
sommese showed that when @xmath6 is smooth and extends , the dimension of @xmath7 can not be too large .
it soon became clear that the extension problem is much harder when @xmath8 .
fujita @xcite further refined some of the techniques and considered new applications e.g. , when @xmath6 is a @xmath9-bundle or a blowing - up . in the case of three folds , fine results were found by bdescu @xcite , when @xmath3 is a @xmath10-bundle over a curve and by sommese @xcite , when @xmath3 is not relatively minimal .
it is worth pointing out that the classical context of hyperplane sections was gradually replaced by the more general situation when @xmath2 is merely an ample divisor , and no projective embedding of @xmath1 is given .
this is a substantial generalization , since in the new setting the normal bundle of @xmath3 in @xmath1 is _ not _ specified . the appearance of mori theory made possible a change of both the point of view and the techniques ( see @xcite ) .
the isomorphism between the picard groups of @xmath1 and @xmath3 given by the lefschetz theorem leads to an inclusion between the kleiman
mori cones @xmath11 and @xmath12 .
as is well known , faces of these cones describe non - trivial morphisms defined on @xmath3 and @xmath1 , respectively .
so , the original question of extending maps from @xmath3 to @xmath1 translates into a comparison problem between these cones .
ideally , when the two cones are equal , all morphisms from @xmath3 extend to @xmath1 ( see e.g. , @xcite and section 8 for results in this direction , usually when @xmath1 or @xmath3 are fano manifolds ) . in the general case , what we can hope for is to extend the contraction of an extremal ray of @xmath3 ( cf .
this is not always possible , but very few counterexamples are known ( see section 4 for this intriguing aspect ) .
the techniques used in this setting are the cone theorem , due to mori , and the contraction theorem , due to kawamata reid shokurov , combined with the well behaved deformation theory of families of rational curves @xcite .
see also @xcite and @xcite for some useful facts about special families of rational curves , coming from extremal rays .
general results on extending morphisms are discussed in section 3 ; in section 5 we concentrate on the special situation when @xmath6 is a @xmath9-bundle or a blowing - up .
we pay special attention to the case of @xmath10-bundles , which is the most difficult . a ( still open ) main conjecture on the subject is stated and various related facts are proved in section 7 . the afore mentioned results by bdescu and sommese on three folds are recovered in section 6 , using the mori theory point of view ( cf .
@xcite ) . in the last section
we discuss the ascent of some good properties from @xmath3 to @xmath1 : e.g. , being uniruled , or rationally connected , or rational , etc .
we have tried to write a complete and coherent exposition , also accessible to the nonspecialist .
we included several new proofs and sometimes substantial simplifications of the original arguments .
several possible generalizations are mentioned at the end of the paper , together with appropriate references to the existing literature .
we work over the complex field @xmath13 . throughout the paper we deal with irreducible , reduced , projective varieties
we use the term manifold if @xmath1 is moreover assumed to be smooth .
we denote by @xmath14 the structure sheaf of @xmath1 . for any coherent sheaf @xmath15 on @xmath1
, @xmath16 denotes the complex dimension of @xmath17 .
if @xmath18 is a morphism , we write @xmath19 for its @xmath20-th direct image .
let @xmath21 be a line bundle on @xmath1 .
@xmath21 is said to be _ numerically effective _
( _ nef _ , for short ) if @xmath22 for all effective curves @xmath23 on @xmath1 .
we say that @xmath21 is _ strictly nef _ ( or _ numerically positive _ ) if @xmath24 for all effective curves @xmath23 on @xmath1 .
@xmath21 is said to be _ big _ if @xmath25 , where @xmath26 denotes the iitaka dimension of @xmath21 .
if @xmath21 is nef then this is equivalent to @xmath27 , where @xmath28 is the first chern class of @xmath21 and @xmath29 .
the pull - back @xmath30 of a line bundle @xmath21 on @xmath1 by an embedding @xmath31 is denoted by @xmath32 .
we denote by @xmath33 the normal bundle of @xmath3 in @xmath1 and by @xmath34 the canonical bundle of a smooth variety @xmath1 .
we use standard notation from algebraic geometry , among which we recall the following ones : 1 .
@xmath35 , the linear equivalence of line bundles ; @xmath36 , the numerical equivalence of line bundles ; 2 .
@xmath37 , the complete linear system associated to a line bundle @xmath21 ; 3 .
@xmath38 , the iitaka dimension of the line bundle associated to a @xmath39-cartier divisor @xmath40 on @xmath1 ; and @xmath41 , the kodaira dimension of @xmath1 , for @xmath1 smooth .
@xmath42 , the @xmath20-th homotopy group , omitting the base point when its choice is irrelevant .
@xmath43 denotes the projective @xmath44-space , @xmath45 denotes the @xmath44-dimensional hyperquadric . for a vector bundle @xmath46
, we write @xmath47 for the associated projective bundle and @xmath48 , or @xmath49 when @xmath50 , for the tautological line bundle , using the grothendieck convention .
line bundles and divisors are used with little ( or no ) distinction .
we almost always use the additive notation .
we say that a line bundle @xmath21 is _ spanned _ if it is spanned , i.e. , globally generated , at all points of @xmath1 by @xmath51 .
let @xmath1 be a projective manifold and let @xmath2 be a smooth ample divisor .
it is a natural classical question to try to understand how the structure of @xmath3 determines the one of @xmath1 .
more precisely , given a surjective morphism @xmath52 associated to a linear system @xmath53 , we look for a linear system @xmath54 on @xmath1 defining a regular map @xmath55 onto a projective variety @xmath56 , such that the following diagram @xmath57_p \ar @{^{(}->}[r ] & x \ar[d]^{\overline{p } } \\ z \ar[r]_{\alpha } & w } \ ] ] commutes .
if the morphism @xmath58 is finite we say that @xmath59 is a _ lifting _ of @xmath6 . if @xmath60 , that is if @xmath61 is an isomorphism onto its image , we say that @xmath59 is a _ strict lifting _ of @xmath6 , or that @xmath6 is _ extendable _ to @xmath59 .
note that this is always the case whenever the restriction map @xmath62 is surjective .
note also that this further condition will be a posteriori satisfied in our setting ( see the proof of theorem [ six ] ) .
assume that the morphism @xmath6 has a lifting @xmath59 .
up to taking the remmert
stein factorization , we can always assume that @xmath6 has connected fibers and @xmath7 is normal .
therefore , by using the ampleness of @xmath3 in @xmath1 , it is a standard fact that one of the following holds : 1 .
@xmath63 and @xmath64 ( in particular @xmath6 is extendable ) ; 2 .
@xmath6 , @xmath59 and @xmath65 are birational ; so , @xmath66 is the normalization morphism ; 3 .
@xmath6 is birational and @xmath67 ; in this case @xmath58 may be of degree @xmath68 .
a simple example is obtained as follows .
consider @xmath69 embedded in @xmath70 by @xmath71 , @xmath72 . by bertini s theorem , we can choose a hyperplane @xmath73 in @xmath70 such that the restriction @xmath3 of @xmath73 to @xmath1 is a smooth ample divisor .
then we get a commutative diagram @xmath74_p \ar @{^{(}->}[r ] \ar[dr ] & x={{\mathbb p}^{1}}\times { { \mathbb p}^{n-1 } } \ar[d]^{\overline{p } } \\ z \ar[r]_{\alpha}&{{\mathbb p}^{n-1 } } } \ ] ] where @xmath6 and @xmath66 are given by the remmert
stein factorization of the restriction @xmath75 and @xmath76 is the natural projection .
note that the morphism @xmath66 is finite of degree two .
moreover , @xmath6 is not an isomorphism .
indeed , assume otherwise .
then @xmath77 is a two - to - one finite covering , so that it induces an isomorphism @xmath78 ( see ( * ? ? ?
* ii , 7.1.20 ) for details and complete references ) . on the other hand , @xmath79 by the lefschetz theorem ; a contradiction .
if the morphism @xmath6 is extendable , our aim is to describe @xmath1 by using the structure morphism @xmath59 .
the occurrence that @xmath6 is not extendable forces @xmath1 to satisfy geometric constraints which , in turn , make @xmath1 special enough to be completely classified . as a typical example , consider the following natural question , formulated in the classical context : [ zeroq ] let @xmath1 be an @xmath44-dimensional manifold embedded in a projective space @xmath70 .
assume that a smooth hyperplane section , @xmath80 , of @xmath1 is a @xmath9-bundle over some manifold @xmath7 , such that the fibers are linearly embedded .
does it follow that the bundle projection @xmath81 extends to @xmath1 giving a @xmath82-bundle projection @xmath83 ?
as soon as @xmath72 , the ( positive ) answer to this question relies on some non - trivial results from the deformation theory of rational curves .
it turns out that the key - fact is the condition @xmath84 , where @xmath85 is a line in a fiber @xmath9 of @xmath81 , i.e. , @xmath85 is a linear @xmath10 with respect to the embedding of @xmath1 in @xmath70 given by @xmath73 .
moreover , the above can happen only if @xmath86 ( proposition [ qnot ] ) .
let @xmath1 be a projective manifold of dimension @xmath44 .
we say that @xmath1 is a _ fano manifold _ if @xmath87 is ample ; its _ index _ , @xmath20 , is the largest positive integer such that @xmath88 for some ample line bundle @xmath89 on @xmath1 .
let @xmath21 be a given ample line bundle on @xmath1 .
we say that @xmath90 is a _ del pezzo variety _
( respectively a _
mukai variety _ ) in the adjunction theoretic sense if @xmath91 ( respectively @xmath92 .
note that del pezzo manifolds are completely described by fujita ( * ? ? ? * i , section 8) .
we refer to mukai @xcite for results on mukai varieties .
we say that @xmath90 is a _ scroll _ over a normal variety @xmath7 of dimension @xmath93 if there exists a surjective morphism with connected fibers @xmath94 , such that @xmath95 for some _ ample _ line bundle @xmath89 on @xmath7 .
we refer to @xcite and ( * ? ? ?
* sections 14.1 , 14.2 ) for relations between the adjunction theoretic and the classical definition of scrolls .
let @xmath1 be a projective manifold and let @xmath96 be a surjective morphism onto another manifold , @xmath7 .
we say that @xmath1 , @xmath96 is a @xmath9-_bundle _ if each closed fiber of @xmath6 is isomorphic to the projective space @xmath9 .
we also say that @xmath1 , @xmath96 is a _
linear @xmath9-bundle _ if @xmath97 for some rank @xmath98 vector bundle @xmath46 on @xmath7 .
we say that @xmath1 , @xmath96 is a _ conic fibration _ over a normal projective variety @xmath7 if every fiber of the morphism @xmath6 is a conic , i.e. , it is isomorphic to the zero scheme of a non - trivial section of @xmath99 .
note that the above definition is equivalent to saying that there exists a rank @xmath100 vector bundle @xmath46 over @xmath7 such that its projectivization @xmath101 contains @xmath1 embedded over @xmath7 as a divisor whose restriction to any fiber of @xmath102 is an element of @xmath103 .
the push - forward @xmath104 can be taken as the above @xmath46 .
it is a standard fact to show that @xmath96 is a flat morphism ; since @xmath1 is smooth , it follows that the base @xmath7 is smooth , too .
a basic tool for dealing with the problems discussed above are lefschetz s theorems , which , in turn , are very much related ( in fact , almost equivalent ) to vanishing results of kodaira type ( see ( * ? ? ?
* i , chapters 3 , 4 ) for a nice general presentation and complete references ) .
see @xcite for the classical statement of lefschetz s theorem . _
( hamm lefschetz theorem)_[lefthm ] let @xmath21 be an ample line bundle on a projective manifold , @xmath1 , and let @xmath105 . then given any point @xmath106 it follows that the @xmath107-_th _ relative homotopy group , @xmath108 , vanishes for @xmath109 . in particular , the restriction mapping , @xmath110 is an isomorphism for @xmath111 , and is injective with torsion free cokernel for @xmath112 . _
( barth lefschetz theorem)_[barthlefthm ] let @xmath3 be a connected submanifold of a projective manifold , @xmath1 .
let @xmath29 , @xmath113 .
assume that @xmath33 is ample .
then for any @xmath114 , we have @xmath115 for @xmath116 . in particular , under the natural map we have @xmath117 if @xmath118 .
moreover : 1 . if @xmath119 , the restriction map @xmath120 is injective with torsion free cokernel ; and 2 . if @xmath121 , then @xmath122 via @xmath123 .
kawamata and viehweg showed that the kodaira vanishing theorem holds for any nef and big line bundle ( see e.g. , ( * ? ? ?
* sections 12 ) ) . _
( kawamata viehweg vanishing theorem)_[kvvan ] let @xmath1 be a projective manifold of dimension @xmath44 , and let @xmath40 be a nef and big divisor on @xmath1 .
then @xmath124 let us recall some definitions and a few facts from mori theory we need .
basic references for details are @xcite , and @xcite .
let @xmath1 be a connected normal projective variety of dimension @xmath125 .
= 20 pt = 20 pt = 1 @xmath126 ; @xmath127 ; @xmath128-cycles}@xmath129 ; @xmath130 , the convex cone in @xmath131 generated by the effective 1-cycles ; @xmath132 , the closure of @xmath130 in @xmath131 with respect to the euclidean topology ; @xmath133 , the _ picard number _ of @xmath1 ; @xmath134 for given @xmath135 ; @xmath136 for given @xmath135 ; @xmath137 for given @xmath135 ; @xmath138 , the dual cone of @xmath12 , namely , the cone in @xmath139 spanned by classes of nef divisors . if @xmath140 is a @xmath141-dimensional cycle in @xmath1 we denote by @xmath142 $ ] its class in @xmath12 .
note that the vector spaces @xmath139 and @xmath131 are dual to each other via the usual intersection of cycles @xmath143 " .
assume that @xmath1 is smooth .
we say that a half line @xmath144 $ ] in @xmath132 , where @xmath145 , is an _
extremal ray _
if @xmath146 and @xmath147 for every @xmath148 such that @xmath149 an extremal ray @xmath144 $ ] is _ nef _ if @xmath150 for every effective divisor @xmath40 on @xmath1 . an extremal ray
which is not nef is said to be _ non - nef_. let @xmath151 be a nef @xmath152-divisor , @xmath153 .
let @xmath154 where
@xmath155 " means the orthogonal complement of @xmath40 in @xmath131 .
then @xmath156 is called a _ good extremal face _ of @xmath12 and @xmath40 is the _ supporting hyperplane _ of @xmath156 , if @xmath156 is entirely contained in the set @xmath157 .
an extremal ray is a @xmath141-dimensional good extremal face .
indeed , for any extremal ray @xmath158 there exists a nef @xmath151 such that @xmath159 .
( mori cone theorem)[conetheorem ] let @xmath1 be a projective manifold of dimension @xmath44 .
then there exists a countable set of curves @xmath160 , with @xmath161 , such that one has the decomposition @xmath162
+ \overline{ne}(x)_{k_x\geq 0}.\ ] ] the decomposition has the properties : 1 .
the set of curves @xmath163 is minimal , no smaller set is sufficient to generate the cone ; 2 . given any neighborhood @xmath164 of @xmath165 , only finitely many @xmath166 $ ] s do not belong to @xmath164 .
the semi - lines @xmath167 $ ] are the extremal rays of @xmath1
. moreover , the curves @xmath163 are ( possibly singular ) reduced irreducible rational curves which satisfy the condition @xmath168 .
( kawamata reid shokurov base point free theorem)[krs ] let @xmath1 be a projective manifold of dimension @xmath169 .
let @xmath40 be a nef cartier divisor such that @xmath170 is nef and big for some positive integer @xmath171 .
then @xmath172 has no base points for @xmath173 .
it is a standard fact that , for a good extremal face @xmath156 , the line bundle @xmath174 is ample for @xmath175 .
therefore , by theorem [ krs ] , the linear system @xmath172 is base point free for @xmath175 , so that it defines a morphism , say @xmath176 . by taking @xmath93 big enough , we may further assume that @xmath56 is normal and the fibers of @xmath177 are connected .
note that @xmath178 , the pair @xmath179 is unique up to isomorphism and @xmath180 @xmath181 . if @xmath23 is an irreducible curve on @xmath1 , then @xmath182 \in f_d$ ] if and only if @xmath183 , which means @xmath184 , i.e. , @xmath185 contracts the good extremal face @xmath156 .
we will call such a contraction , @xmath185 , the _ contraction of _ @xmath156 . if @xmath186 , @xmath158 an extremal ray , we will denote @xmath187 the contraction morphism .
let @xmath188 note that @xmath189 is the locus of curves whose numerical class is in @xmath158 .
we will refer to @xmath189 simply as the _ locus _ of @xmath158 .
if @xmath1 is smooth we define the _ length of an extremal ray _ , @xmath190 \in r \}.\ ] ] note that the cone theorem yields the bound @xmath191 .
we will also use the notation @xmath192 .
we say that a rational curve @xmath23 generating an extremal ray @xmath193 $ ] is a _ minimal curve _ if @xmath194 .
the following useful inequality is inspired by mori s bend and break ( cf .
* theorem 0.4 ) , and also ( * ? ? ?
* theorem 1.1 ) , ( * ? ? ?
* corollary iv.2.6 ) ) .
[ exrayiw ] let @xmath1 be a projective manifold of dimension @xmath44 .
assume that @xmath34 is not nef and let @xmath158 be an extremal ray on @xmath1 of length @xmath195 .
let @xmath196 be the contraction of @xmath158 and let @xmath189 be any irreducible component of the locus of @xmath158 .
let @xmath197 be any irreducible component of any fiber of the restriction , @xmath198 , of @xmath196 to @xmath189 .
then @xmath199 by combining the theorem above with a result due to ando @xcite , winiewski ( * ? ? ?
* theorem ( 1.2 ) ) showed the following , which plays an important role in the sequel .
[ aw ] ( ando @xcite , winiewski @xcite ) let @xmath1 be a projective manifold of dimension @xmath200 . assume that @xmath34 is not nef .
let @xmath201 be the contraction morphism of an extremal ray @xmath158 . if every fiber of @xmath177 has dimension at most one , then @xmath7 is smooth and either @xmath177 is the blowing - up of a smooth codimension two subvariety of @xmath7 , or @xmath177 is a conic fibration .
ando ( * ? ? ?
* ( 3.10 ) , ( 2.3 ) ) proved the theorem above assuming that the locus @xmath189 of @xmath158 satisfies the condition @xmath202 . from the inequality of theorem [ exrayiw ]
it follows that this is the case . indeed , if @xmath203 , for any irreducible component @xmath197 of any fiber of the restriction of @xmath177 to @xmath189 , we would have @xmath204 contradicting the fibers dimension assumption .
we follow the notation in @xcite , to which we refer for details ; see also @xcite .
let @xmath1 be a projective manifold .
by @xmath205 we denote the scheme parameterizing morphisms from @xmath10 to @xmath1 which are birational onto their image .
we will denote by @xmath206 $ ] the point of @xmath205 determined by such a morphism @xmath207 .
a reduced , irreducible subvariety @xmath208 determines a _ family of rational curves _ on @xmath1 .
we let @xmath209 be the universal family , restricted to @xmath210 , with @xmath211 and @xmath212 the natural projections .
we call the image of @xmath213 the _ locus of the family _ , denoted by @xmath214 .
covering family _ is a family satisfying @xmath215 .
we say that a family @xmath210 , closed under the action of @xmath216 , is _ unsplit _ if the image of @xmath210 in @xmath217 under the natural morphism @xmath206\to [ f({{\mathbb p}^{1}})]$ ] is closed . in general ,
the closure of the image of @xmath210 in @xmath218 determines a _ family of rational 1-cycles _ on @xmath1 .
if @xmath219 is a fixed ( closed ) point , we denote by @xmath220 the closed subfamily of @xmath210 consisting of morphisms sending a fixed point @xmath221 to @xmath222 .
we say that @xmath210 is _ locally unsplit _ if , for @xmath223 a general point , the family @xmath220 is unsplit .
a family @xmath210 of rational 1-cycles on @xmath1 is _ quasi - unsplit _ if any two irreducible components of cycles in @xmath210 are numerically proportional .
such families typically arise from cycles belonging to an extremal ray .
we discuss throughout this section some general results on extending morphisms @xmath81 from ample ( smooth ) divisors @xmath3 of a manifold @xmath1 . to begin with ,
let us prove two early theorems due to sommese @xcite ( see also ( * ? ? ?
* ( 5.2.1 ) , ( 5.2.5 ) ) ) that marked the starting point of the subject .
the first one shows that the morphism @xmath6 is always extendable whenever @xmath224 .
the second one gives the restriction that @xmath225 for a smooth @xmath81 to extend .
[ extensionthm ] ( sommese @xcite ) let @xmath3 be a smooth ample divisor on a projective manifold @xmath1 .
let @xmath5 be a surjective morphism .
if @xmath226 then @xmath6 extends to a surjective morphism @xmath227 .
let @xmath228 .
without loss of generality it can be assumed that @xmath229 .
thus we have that @xmath230 and therefore by lefschetz theorem we see that the restriction map gives an isomorphism , @xmath231 . moreover , by remmert stein factorizing @xmath6 it can be assumed that @xmath7 is normal and @xmath6 has connected fibers .
let @xmath89 be a very ample line bundle on @xmath7 .
since @xmath231 there exists an @xmath232 whose restriction , @xmath233 , to @xmath3 is isomorphic to @xmath234 .
let @xmath235 .
we claim that @xmath236 by serre duality we are reduced to showing @xmath237 for @xmath238 . from the relative form of the kodaira vanishing theorem ( see e.g. , @xcite ) we see that @xmath239 for @xmath240 . using the leray spectral sequence we deduce that @xmath241 the last group is zero , since @xmath242 by our assumption .
this shows ( [ vanishing ] ) .
now consider the exact sequence @xmath243 by ( [ vanishing ] ) we have @xmath244 for @xmath245 .
therefore , by serre duality , @xmath246 for @xmath245 .
thus the exact sequence above gives an injection of @xmath247 into @xmath248 .
by serre s vanishing theorem , @xmath249 for @xmath250 .
therefore we conclude that @xmath251 hence in particular @xmath252 , so that we have a surjection @xmath253 since @xmath254 with @xmath89 very ample on @xmath7 , we infer that there exist @xmath255 divisors @xmath256 in @xmath257 such that @xmath258 .
since @xmath3 is ample it thus follows that @xmath259 .
we claim that @xmath73 is spanned by its global sections .
if @xmath260 , then @xmath261 . if @xmath262 , then @xmath263 this contradicts the above inequality
let @xmath264 be the map associated to @xmath265 .
ampleness of @xmath3 yields that @xmath266 , so @xmath267 and we are done .
q.e.d . plus
1pt [ basere ] ( sommese @xcite ) let @xmath3 be a smooth ample divisor on a projective manifold , @xmath1 .
let @xmath5 be a morphism of maximal rank onto a normal variety , @xmath7 .
if @xmath6 extends to a morphism @xmath227 , then @xmath225 .
we follow the topological argument from ( * ? ? ?
* proposition v ) .
let @xmath268 be the image of the set of points where @xmath59 is not of maximal rank .
by ampleness of @xmath3 the set @xmath268 is finite .
let @xmath269 , @xmath270 be the fibers of @xmath6 , @xmath59 over @xmath271 respectively .
let @xmath272 and let @xmath273 . from standard results in topology
we deduce : @xmath274 for all @xmath107 , @xmath275 for @xmath276 and @xmath277 is onto .
it follows that @xmath278 is an isomorphism for @xmath276 and it is onto for @xmath279 . from theorem [ lefthm ] , we have that @xmath280 is an isomorphism if @xmath281 dim@xmath3 and is onto for @xmath282 dim@xmath3 .
consider the following commutative diagram with exact rows : @xmath283 arguing by contradiction , let us assume @xmath284 , or @xmath285 .
it follows that @xmath286 dim@xmath3 and @xmath287 .
this , the above , and the five lemma show that @xmath288 for @xmath289 and @xmath290 is onto . by whitehead
s generalization of hurewicz s theorem ( * ? ? ?
* theorem 9 , p. 399 ) we get @xmath291 for @xmath289 and a surjection @xmath292 by noting that @xmath293 , this leads to the contradiction @xmath294 .
q.e.d . plus 1pt [ equality ] in the situation of theorem [ basere ] , if @xmath295 , a more refined argument based on results of lanteri and struppa @xcite shows that the general fiber @xmath296 of @xmath59 is isomorphic to @xmath297 .
we refer to ( * ? ? ?
* ( 5.2.6 ) , ( 2.3.9 ) ) for more details and complete references . from now on ,
* _ we are reduced to consider the problem of extending morphisms @xmath298 from ample divisors @xmath3 of a manifold @xmath1 in the hardest case when _ @xmath299 . in this
setting , we shall use ( part of ) mori theory ; we compare the cone of curves of @xmath3 and @xmath1 and give results on extending contractions of extremal rays ( cf .
@xcite ) . as noted in (
* section 3 ) , the following useful fact holds true .
it is an easy consequence of theorem [ aw ] and a lemma due to kollr @xcite .
[ kollarprop ] let @xmath1 be a projective manifold of dimension @xmath300 .
assume that @xmath34 is not nef and let @xmath193 $ ] be an extremal ray on @xmath1 .
let @xmath3 be a smooth ample divisor on @xmath1 .
if @xmath301 , then @xmath302 . by the lefschetz theorem , the embedding @xmath303 gives an isomorphism @xmath304 , under which we get a natural inclusion @xmath305 .
let @xmath201 be the contraction of the extremal ray @xmath158 and let @xmath189 be the locus where @xmath177 is not an isomorphism , i.e. , the locus of curves whose numerical class is in @xmath158 .
if there is a fiber @xmath306 of @xmath177 whose dimension is at least two , then @xmath307 contains a curve @xmath140 which generates @xmath158 in @xmath12 , and hence @xmath302 .
thus we can assume that every fiber of @xmath177 has dimension at most one , so that theorem [ aw ] applies .
therefore we are done after showing that in each case of [ aw ] the divisor @xmath3 contains a fiber of @xmath177 . in the birational case , @xmath189 is a @xmath10-bundle over @xmath308 .
let @xmath309 be a fiber of the bundle projection @xmath310 .
then @xmath311 , so that @xmath312 and the ampleness of @xmath3 give @xmath313
. therefore lemma [ kollar](i ) below leads to the contradiction @xmath314 , so @xmath315 . in the conic fibration case , for any fiber @xmath316 of @xmath177 , we have @xmath317 , and hence we get @xmath318
. thus lemma [ kollar](ii ) gives the contradiction @xmath319 .
q.e.d . plus
1pt let us point out the following consequence of proposition [ kollarprop ] ( cf .
section [ fano ] ) .
[ reph ] let @xmath1 be a projective manifold of dimension @xmath72 , let @xmath73 be an ample line bundle on @xmath1 , and let @xmath3 be an effective smooth divisor in @xmath257 .
assume that @xmath320 is nef .
then @xmath1 is a fano manifold and @xmath321 .
let @xmath322 , which is nef .
then @xmath323 is ample , so that @xmath1 is a fano manifold .
let @xmath193 $ ] be an extremal ray in the polyhedral cone @xmath12 .
by assumption , @xmath324 .
then proposition [ kollarprop ] applies to give that @xmath158 is contained in @xmath11 .
so @xmath325 .
q.e.d . plus
1pt [ kollar ] ( kollr @xcite ) let @xmath1 be a projective manifold .
* let @xmath96 be a @xmath10-bundle over a normal projective variety @xmath7 .
let @xmath2 be a divisor such that the restriction @xmath81 is finite of degree one .
if @xmath3 is ample then @xmath326 .
* let @xmath96 be a conic fibration over a normal projective variety @xmath7 .
let @xmath2 be a divisor such that the restriction @xmath81 is finite of degree two ( or one ) . if @xmath3 is ample then @xmath327 .
\(i ) note that @xmath328 is an ample rank @xmath329 vector bundle since @xmath3 is ample . on the other hand
, the section @xmath330 gives an extension @xmath331 where @xmath21 is a line bundle .
thus @xmath332 , which contradicts ampleness for @xmath333 .
\(ii ) if @xmath96 has only smooth fibers , then , after a finite base change @xmath334 , we get a @xmath10-bundle @xmath335 .
now , @xmath336 has rank @xmath100 and we get an extension @xmath337 where @xmath46 is a rank @xmath329 vector bundle .
thus @xmath338 , which contradicts ampleness for @xmath339 .
if @xmath96 has singular fibers , then let @xmath335 be the universal family of lines in the fibers .
the pull - back of @xmath3 to @xmath340 intersects every line once and it is ample .
moreover , @xmath341 , thus we are done by ( i ) .
q.e.d . plus
1pt [ kollarcor ] let @xmath1 be a projective manifold of dimension @xmath300 .
let @xmath3 be a smooth ample divisor on @xmath1 .
assume that @xmath342 is not nef and let @xmath343 be an extremal ray on @xmath3 .
further assume that there exists a nef divisor @xmath40 on @xmath1 such that @xmath344 and such that @xmath345 is contained in the set @xmath346 .
then there exists an extremal ray @xmath347 which induces @xmath343 .
note that @xmath345 is a locally polyhedral face of @xmath12 since @xmath348 .
let @xmath349 , @xmath350 extremal rays on @xmath1 , @xmath351 .
we can assume @xmath352 for each @xmath20 since otherwise we are done .
since @xmath344 , it follows that , for some @xmath353 , @xmath354 therefore @xmath355 , @xmath356 , by proposition [ kollarprop ] .
recalling that @xmath352 , we thus conclude that @xmath343 is not extremal on @xmath3 , a contradiction
. q.e.d . plus
1pt the following numerical invariant was introduced in @xcite .
let @xmath1 be a projective manifold of dimension @xmath200 .
let @xmath3 be a smooth ample divisor on @xmath1 .
let @xmath158 be an extremal ray on @xmath3 and let @xmath73 be an ample divisor on @xmath1 .
then define @xmath357 and @xmath358 the following slightly improves the main result of @xcite .
[ six ] let @xmath1 be a projective manifold of dimension @xmath72 .
let @xmath3 be a smooth ample divisor on @xmath1 .
assume that @xmath342 is not nef and let @xmath158 be an extremal ray on @xmath3 .
let @xmath359 be the contraction of @xmath158 .
the following conditions are equivalent : 1 .
there exists an extremal ray @xmath360 on @xmath1 which induces @xmath158 on @xmath3 ; 2
. there exists a nef divisor @xmath361 on @xmath1 such that @xmath362 ; 3 .
@xmath6 is extendable ; 4 .
@xmath6 has a lifting ; 5 .
there exists an ample line bundle @xmath73 on @xmath1 such that @xmath363 .
\(i ) @xmath364 ( ii ) we have @xmath365 for some nef divisor @xmath361 on @xmath1 .
restricting to @xmath3 , gives @xmath362 .
\(ii ) @xmath364 ( iii ) there exists an extremal ray @xmath366 . then by proposition [ kollarprop ] , @xmath367 ,
that is , @xmath360 induces @xmath158 on @xmath3 . replacing @xmath361 by some other nef divisor on @xmath1
, we may assume that @xmath368 .
it follows that @xmath369 is ample for @xmath173 and cont@xmath370 is given by @xmath371 . by kodaira vanishing
we have @xmath372 , showing that cont@xmath370 extends cont@xmath373 .
\(iii ) @xmath374 ( iv ) is obvious .
\(iv ) @xmath374 ( i ) the morphism @xmath375 which lifts @xmath6 is associated to a complete linear system @xmath37 , for some nef line bundle @xmath21 on @xmath1 .
clearly , @xmath376 and @xmath377 .
hence , in particular , @xmath378 . therefore ,
since @xmath379 is locally polyhedral in @xmath12 ( see also the proof of [ kollarcor ] ) , there exists an extremal ray @xmath360 on @xmath1 such that @xmath380 .
we also know by proposition [ kollarprop ] that @xmath381 .
it thus follows that @xmath6 contracts @xmath360 .
since @xmath6 is the contraction of an extremal ray , @xmath158 , we conclude that @xmath382 in @xmath383 .
\(i ) @xmath364 ( v ) as noted in the proof of ( ii)@xmath374 ( iii ) , if @xmath365 , we have that @xmath384 is ample for @xmath173 .
nefness of @xmath361 yields for each extremal ray @xmath385 $ ] on @xmath3 , @xmath386 minimal curve , @xmath387 . moreover , since @xmath388 , we get @xmath389 .
this means that @xmath363 is minimal . finally , to show ( v ) @xmath364 ( i ) , let @xmath390 $ ] , @xmath23 minimal rational curve .
let @xmath391 and let @xmath392 .
assume that @xmath363 .
then @xmath393 is nef on @xmath3 , since otherwise @xmath394 for some extremal ray @xmath395 $ ] on @xmath3 .
this would give @xmath396 , contradicting the minimality of @xmath397 .
we claim that @xmath398 is nef on @xmath1 . assuming the contrary , by mori s cone theorem there exists an extremal ray @xmath395 $ ] , such that @xmath399 .
let @xmath400 be the contraction of @xmath401 .
if there is a fiber @xmath316 of @xmath400 of dimension @xmath68 , then some curve @xmath402 is contained in @xmath3 by the ampleness of @xmath3 . since @xmath403
, this contradicts the nefness of the restriction of @xmath404 to @xmath3 .
thus we conclude that all fibers of @xmath400 are of dimension @xmath405 . by theorem
[ aw ] we deduce that @xmath1 is either a conic fibration , or the blowing - up of a smooth codimension two subvariety . in this latter case one
has @xmath406 , which contradicts @xmath399 . in the first case one has either @xmath407 or @xmath406 according to whether @xmath386 is a conic or a line . if @xmath406 we have the same numerical contradiction .
if @xmath407 we get @xmath408 by using again the above inequality .
it thus follows that @xmath1 is a @xmath10-bundle over a smooth variety @xmath56 and @xmath3 is a smooth section . by pushing forward the exact sequence @xmath409 we get an exact sequence @xmath410 where @xmath46 is an ample rank @xmath329 vector bundle such that @xmath97 and @xmath411 is an ample line bundle . by the kodaira vanishing theorem we have @xmath412 , so that the sequence splits
this contradicts the ampleness of @xmath46 .
thus we conclude that @xmath413 is nef .
now , note that @xmath414 and that @xmath345 is strictly contained in @xmath415 .
thus corollary [ kollarcor ] applies to say that there exists an extremal ray @xmath416 which induces @xmath158 .
q.e.d . plus
1pt [ newcor ] let @xmath1 be a projective manifold of dimension @xmath72 .
let @xmath3 be a smooth ample divisor on @xmath1 .
assume that @xmath342 is not nef .
then there exists an extremal ray on @xmath3 which extends to an extremal ray on @xmath1 .
in particular , when @xmath3 has only one extremal ray , it always extends . for instance , take any extremal ray @xmath158 of @xmath3 attaining the minimal value of the invariant @xmath417 ( i.e. , @xmath418 ) .
q.e.d . plus
1pt the following theorem is a version of a result of occhetta ( * ? ?
* proposition 5 ) ( who states it in the more general case when @xmath3 is the zero locus of a section of an ample vector bundle @xmath46 on @xmath1 of the expected dimension @xmath419 ) .
his argument contains an unclear critical point . with notation as in the proof below , the conclusion in @xcite uses in an essential way the fact that @xmath40 is an adjoint divisor , i.e. , @xmath420 for some integer @xmath421 and some _ ample _ line bundle @xmath89 on @xmath1 ( implying that @xmath345 is strictly contained in @xmath422 ) .
however , we only know that @xmath423 is ample !
see also remark [ r3.12 ] below .
[ occhetta ] ( cf .
occhetta @xcite ) let @xmath1 be a projective manifold of dimension @xmath300 .
let @xmath3 be a smooth ample divisor on @xmath1 .
assume that @xmath342 is not nef and let @xmath158 be an extremal ray on @xmath3 .
let @xmath424 be a rational curve whose numerical class is in @xmath158 .
assume that the deformations of @xmath23 in @xmath1 yield a covering and quasi - unsplit family of rational cycles .
then @xmath158 is an extremal ray of @xmath1 , too . in particular
, the conclusion holds if the deformations of @xmath23 in @xmath3 cover @xmath3 and @xmath425 for some ample divisor @xmath73 on @xmath1 .
we show first the last claim .
let @xmath426 be the normalization of @xmath23 and let @xmath427 and @xmath207 be the induced morphisms to @xmath3 and @xmath1 respectively .
if @xmath23 yields a covering family of @xmath3 , its deformations in @xmath1 cover @xmath1 , too .
indeed , consider the tangent bundle sequence @xmath428 by pulling back to @xmath10 , we get the exact sequence @xmath429 since both @xmath430 and @xmath431 are nef and hence spanned , we conclude that @xmath432 is spanned , and therefore that @xmath23 induces a covering family on @xmath1 , see ( * ? ? ?
* ii , section 3 , iv , ( 1.9 ) ) .
moreover , the condition @xmath425 ensures that the deformations of @xmath23 in @xmath1 yield an unsplit , hence also quasi - unsplit , family .
let @xmath433 be some ( nef ) supporting divisor of @xmath158 , i.e. , @xmath434 .
by the lefschetz theorem , there exists a divisor @xmath435 which restricts to @xmath436 .
let @xmath210 be a covering and quasi - unsplit family of rational cycles on @xmath1 , containing @xmath23 .
[ wclaim ] @xmath40 is nef .
assume that @xmath437 for some irreducible curve @xmath438 on @xmath1 .
clearly , we can assume that @xmath438 is not contained in @xmath3 since the restriction of @xmath40 to @xmath3 is nef .
since @xmath210 is a covering family and @xmath3 is ample , we can find a curve @xmath439 in @xmath210 parameterizing curves meeting both @xmath438 and @xmath3 .
let @xmath440 be the normalization of @xmath439 .
consider the base - change diagram @xmath441 \ar[rd]_{\pi } \ar @/^1.5pc/[rrr]^{\psi } & s \ar[r ] \ar[d ] & { { \mathscr f}}\ar[r]_{q } \ar[d]^p & x \\ & b \ar[r ] & v & & } \ ] ] where @xmath209 is the universal family and @xmath442 is a desingularization of @xmath268 , an irreducible component of @xmath443 , whose locus contains @xmath438 .
note that @xmath442 is a ruled surface over the curve @xmath440 .
let @xmath444 be the trace on @xmath3 of the image in @xmath1 of the surface @xmath442 .
since @xmath445 is an ample divisor on @xmath446 , there exists at least one irreducible component , say @xmath447 , of @xmath445 which is not contracted by @xmath448 .
let @xmath449 , @xmath450 be two irreducible curves on @xmath442 such that @xmath451 , @xmath452 . by the above and the hypothesis that @xmath210 is quasi - unsplit ,
@xmath450 is not a fiber of @xmath453 .
we can write , for some integers @xmath454 , where @xmath455 is a section of @xmath456 and each @xmath457 is contained in a fiber of @xmath456 .
we also have @xmath458 , for some integers @xmath66 , @xmath459 , @xmath460 .
thus @xmath461 that is , in @xmath462 , one has @xmath463 , with @xmath464 .
let @xmath465 . since @xmath210 is quasi - unsplit , @xmath466 , so that @xmath467 since @xmath436 is nef .
this shows the claim . to conclude we have to show that @xmath158 is an extremal ray on @xmath1 ; see also the proof of (
* theorem ( 5.1 ) ) . by lefschetz s theorem
, the embedding @xmath303 gives a natural inclusion @xmath305 .
clearly , @xmath468 is @xmath34-negative by the adjunction formula . since @xmath158 is an extremal ray of @xmath11 , by duality it corresponds to it an extremal face of maximal dimension @xmath469 of the cone of nef divisors @xmath470 .
therefore we can find @xmath469 good supporting divisors of @xmath158 whose numerical classes are linearly independent in @xmath471 . by claim [ wclaim ] , this implies that such good supporting divisors extend to divisors on @xmath1 that are nef , trivial on @xmath360 , and whose numerical classes are linearly independent in @xmath139 .
since there are @xmath469 of them , and @xmath472 by the isomorphism @xmath473 , this implies that @xmath360 is an extremal ray of @xmath12 .
q.e.d . plus
1pt [ r3.12 ] in ( * ? ? ?
* proposition 5 ) , the author states the result assuming that @xmath158 is nef .
however , the theorem also applies to non - nef extremal rays of @xmath3 , see proposition [ palbup2 ] below .
note that , even when @xmath193 $ ] is nef , in general @xmath23 does not define a covering family of @xmath3 .
e.g. , take cont@xmath373 to be a conic fibration , @xmath23 being a line in a degenerate fiber .
first , we recall the following simple observations , due to bdescu .
[ badlemma ] ( ( * ? ? ? * remark 1 ) , p. 170 ) ) on the projective line @xmath10 , consider a line bundle @xmath474 , for some integer @xmath475 .
write @xmath476 , with @xmath477
. then there exists a surjective map @xmath478 consider the global sections @xmath479 as homogeneous polynomials in two variables @xmath480 , @xmath481 .
we have natural inclusions @xmath482 given by multiplication by @xmath483 and @xmath484 respectively .
then we get a surjective map @xmath485 and injections @xmath486 thus @xmath487 gives the requested map .
the surjectivity follows from the commutative square @xmath488_{ev_x}\ar[r ] & h^0({{\mathbb p}^{1 } } , { \mathcal{o}}_{{{\mathbb p}^{1}}}(a ) ) \ar[d]^{ev_x } \\
( { \mathcal{o}}_{{{\mathbb p}^{1}}}(b)\oplus { \mathcal{o}}_{{{\mathbb p}^{1}}}(c))_x \ar[r ] & ( { \mathcal{o}}_{{{\mathbb p}^{1}}}(a))_x } \ ] ] where the vertical arrows are the evaluation maps in a given point @xmath489 and @xmath490 is onto by spannedness of @xmath474 .
q.e.d . plus
1pt [ badprop ] ( @xcite ) given a vector bundle @xmath491 , with @xmath492 , @xmath493 , and @xmath477 , there exists an exact sequence @xmath494 lemma [ badlemma ] yields a surjective map @xmath495 whose kernel is the trivial bundle since @xmath496 .
q.e.d . plus
1pt [ lurem ] note that proposition [ badprop ] gives rise to a method to construct ample divisors which are projective bundles over @xmath10 .
indeed , let @xmath497 and @xmath498 .
as soon as @xmath499 for each index @xmath500 , the exact sequence expresses @xmath3 as a smooth ample divisor of @xmath1 ; it is recovered by pushing forward the exact sequence @xmath501 the following fact is well known .
we include the proof for reader s convenience . [ wk ]
let @xmath502 be a @xmath503-bundle over @xmath10 , for some rank @xmath44 vector bundle @xmath504 .
assume that @xmath210 is a fano manifold , of index @xmath505 .
then , for some integer @xmath171 , either 1
. @xmath506 , @xmath507 and @xmath508 ; or 2 .
@xmath210 is the blowing - up , @xmath509 , along a codimension two linear subspace of @xmath510 , @xmath511 and @xmath512 .
after normalization of the integers @xmath513 , write @xmath514 and consider the section @xmath438 of @xmath515 corresponding to the quotient @xmath516 onto the first summand .
note that the morphism @xmath517 maps the curve @xmath438 to a point . on the other hand , since @xmath210 is a fano manifold with @xmath518 , there are two extremal rays , @xmath519 , corresponding to @xmath6 , and @xmath520 , generating the cone @xmath521
. since the morphism @xmath522 is not finite , it must coincide with the contraction of @xmath523 .
now , setting @xmath524 , the canonical bundle formula yields @xmath525 therefore , dotting with @xmath438 , we get @xmath526 , so that either @xmath527 , or @xmath528 , leading to the two cases as in the statement . note that by the canonical bundle formula , in the first case the index of @xmath210 is @xmath507 , while , in the second case , @xmath511 .
q.e.d . plus 1pt [ only](only known examples of non - extendable extremal rays ) .
let @xmath1 be a projective manifold of dimension @xmath72 .
let @xmath3 be a smooth ample divisor on @xmath1 .
the only known examples of extremal rays @xmath158 of @xmath3 which do not extend to @xmath1 are the following : 1 . @xmath529 , and @xmath158 is the nef extremal ray corresponding to the @xmath10-bundle projection @xmath530 . the manifold @xmath1 is constructed as in remark [ lurem ] ; 2 .
@xmath531 for some integer @xmath475 , @xmath158 is the non - nef extremal ray corresponding to the blowing - up , @xmath532 , along a codimension two linear subspace of @xmath503 .
again , @xmath1 is constructed as in remark [ lurem ] , @xmath533 . in case
( 1 ) , the @xmath534-bundle projection @xmath535 on the first factor extends by construction .
then , if @xmath530 extends too , we would have a surjective map @xmath536 , where @xmath503 is a fiber of the extension of @xmath6 ; a contradiction .
let @xmath3 be as in case ( 2 ) . by proposition [ badprop ]
, @xmath3 embeds as a smooth ample divisor of @xmath537 . note that @xmath538 , where @xmath519 , @xmath523 are the extremal rays corresponding to the bundle projection @xmath539 , and to the blowing - up @xmath532 respectively . moreover , lemma [ wk ] applies to say that @xmath1 is not a fano manifold .
therefore @xmath11 is strictly contained in @xmath12 . since @xmath540 by the lefschetz theorem , and the projection @xmath539 extends by construction , we thus conclude that the extremal ray @xmath523 does not extend to @xmath1 . clearly @xmath533 in the above example .
note that by taking as @xmath3 a hyperplane section of the segre embeding @xmath1 of @xmath541 , the restriction to @xmath3 of the bundle projection @xmath542 is in fact the blowing - up @xmath532 along a codimension two linear subspace .
of course , in this case , the extremal ray defining @xmath543 extends to @xmath1 . in terms of proposition [ badprop ] , this situation corresponds to the case when @xmath544 , that is @xmath545 is an ample divisor of @xmath546 . that those in [ only ] are the only known examples of non - extendable extremal rays looks quite surprising to us .
we propose the following speculations with the hope they may eventually lead to an explanation of this fact .
let @xmath1 be a projective manifold of dimension @xmath72 .
let @xmath3 be a smooth ample divisor on @xmath1 .
consider the following property , for an extremal ray , @xmath158 , of @xmath3 .
@xmath547 _ for any nef divisor @xmath73 on @xmath3 , such that @xmath548 , it follows + that @xmath549 is nef ( here @xmath549 is the unique divisor class on @xmath1 such that @xmath550_. first , note that @xmath551 implies that @xmath158 is a ray of @xmath1 ( cf .
* theorem 4.1 ) and end of proof of theorem [ occhetta ] ) .
[ sexy ] assume that property @xmath551 holds for all extremal rays of @xmath3 such that @xmath552 for some nef divisor @xmath361 on @xmath1 , @xmath553 .
assume , also , that @xmath554 .
then every extremal ray of @xmath3 extends to an extremal ray of @xmath1 .
assume that we have some extremal ray of @xmath3 , say @xmath555 , which is not a ray of @xmath1 .
we may assume that @xmath556 for some @xmath557 .
observe that , by our hypothesis , any ray @xmath158 of @xmath3 which is contained in some face of @xmath12 satisfies property ( @xmath558 ) , and hence , as noted above , @xmath158 is a ray of @xmath1 .
let @xmath559 and let @xmath560 be all the extremal rays of @xmath11 such that @xmath561 , @xmath562 .
next consider the non - degenerate convex cone @xmath563 defined by @xmath564 then @xmath565 is a face of the cone @xmath566 and , for each index @xmath20 , @xmath567 [ percepito ] there exists a face @xmath316 of the cone @xmath566 such that @xmath316 contains two extremal rays of @xmath3 , say @xmath568 , such that @xmath401 is a ray of @xmath1 and @xmath569 is not
. recall that by proposition [ kollarprop ] all extremal rays of @xmath1 contained in @xmath570 are also extremal rays of @xmath3 . therefore , some of the extremal rays @xmath571 , @xmath240 , are rays of @xmath1 .
take one of them , say @xmath519 .
then there exists a face @xmath572 of @xmath566 containing @xmath519 and some of the other rays @xmath573 , for some @xmath574 .
if one of the rays @xmath573 ( say @xmath523 ) , is not a ray of @xmath1 , then take @xmath575 , @xmath576 and @xmath577 .
if all the extremal rays @xmath573 , @xmath574 , are rays of @xmath1 , we apply the same argument to conclude that either we prove the claim , or every extremal ray of @xmath3 contained in @xmath566 lifts to an extremal ray of @xmath1 , contradicting the assumption that @xmath555 does not .
thus we may assume to be in the situation described in claim [ percepito ] .
take a nef divisor @xmath40 on @xmath3 such that @xmath578 .
if the unique divisor class @xmath579 on @xmath1 which restricts to @xmath40 is nef , then by our assumption @xmath569 is a ray of @xmath12 .
this contradicts the claim , so @xmath579 is not nef on @xmath1 .
now , take a nef divisor @xmath580 on @xmath1 such that @xmath581 . for @xmath582
, consider @xmath583 and let @xmath584 . then @xmath585 . -1.5 cm [ nuco ] ( 2,2 ) ( -1,0)198 ( -0.42,0,7)@xmath586 ( -0.4,0,6)@xmath587 ( -0.38,0,5)@xmath588 ( -0.28,0,47)@xmath588 ( 0.1,0.2)@xmath589 ( 0.06,-0.64)@xmath590 ( -0.24,0.36)@xmath591 ( -0.18,0.28)@xmath592 ( -0.12,0.22)@xmath592 ( 0.3,-0.2)@xmath587 ( 0.32,-0.28)@xmath593 ( 0.26,-0.24)@xmath594 ( 0.23,-0.15)@xmath595 ( 0.3,0.5)@xmath586 ( 0.92,0.22)@xmath596 ( 1,0.25)@xmath597 ( 0.28,0.87)@xmath596 ( 0.28,0.9)@xmath598 ( -0.55,0.87)@xmath596 ( -0.5,0.88)@xmath599 ( -0.5,0.95)@xmath600 ( -0.4,0.43)@xmath596 ( -0.35,0.47)@xmath601 ( -2,0)@xmath602 ( 1.6,-0.03)@xmath603 ( -0.6,1.03)@xmath401 ( 0.8,1.5)@xmath604 ( -1.5,-0.9)@xmath605 ( -1.2,-1.4)@xmath606 ( 0.2,-1.8)@xmath607 ( -0.5,0.88)@xmath608 ( -1.3,0.52)@xmath609 ( 0.3,-0.4)@xmath596 ( 0.32,-0.38)@xmath610 3.4 cm [ notprop ] the extremal ray @xmath401 does not satisfy property @xmath551 . for some @xmath611 , the divisor @xmath612 is not nef and its restriction @xmath613 is nef on @xmath3 . in particular
one has @xmath614 thus , since @xmath581 , equalities hold above , so that @xmath615 .
thus @xmath401 does not satisfy property @xmath551 ; see figure 1 .
the claim above leads to a contradiction , and hence concludes the proof of the proposition .
q.e.d . plus
1pt \(1 ) the examples of non - extending extremal rays in example [ only ] were known to the experts since the early 80 s .
the resulting observation that , in general , @xmath616 when @xmath2 is ample , was rediscovered in @xcite .
\(2 ) the ( open ) problem of deciding whether or not the property ( @xmath558 ) holds for all extremal rays @xmath158 of @xmath3 such that @xmath617 for some nef divisor @xmath40 on @xmath1 , @xmath153 , looks hard but very interesting . a positive answer to it would imply , via proposition [ sexy ] , that non - extendable rays may only occur when @xmath618 .
moreover , a positive answer would give a proof for the claim made in ( * ? ? ?
* theorem 4.3 ) .
see also remark [ beautyfulmind ] for further applications .
let us start by recalling some useful preliminary facts .
the following result goes back to goren @xcite and kobayashi
ochiai @xcite .
we refer also to fujita ( * ? ? ?
* chapter i , ( 1.1 ) , ( 1.2 ) ) where the cohen macaulay assumption was removed .
[ kof ] let @xmath21 be an ample line bundle on an @xmath44-dimensional irreducible projective variety @xmath1 .
if @xmath619 and @xmath620 , then @xmath621 .
[ co ] if @xmath1 is a fano manifold of dimension @xmath44 and index @xmath20 , we have @xmath622 and @xmath623 if equality holds .
use the hilbert polynomial and kodaira vanishing to check the hypothesis of theorem [ kof ] .
see @xcite for details .
q.e.d . plus
1pt [ re ] in a completely similar way one proves that if @xmath1 is as above and @xmath624 , then @xmath625 , see again @xcite .
the smooth version of the following fact was proved by ramanujam @xcite and sommese @xcite .
the general case is due to bdescu @xcite .
we also refer to ( * ? ? ?
* section 2.6 and ( 5.4.10 ) ) for a different argument based on rossi s extension theorem .
[ pn ] let @xmath626 be an ample cartier divisor on a normal projective variety @xmath1 of dimension @xmath200 .
then @xmath1 is the cone @xmath627 , where @xmath628 . if @xmath629 , the assertion is true for @xmath630 as well .
let us recall the argument in the smooth case .
first , note that , by lefschetz theorem , @xmath631 via the embedding @xmath632 .
this is clear if @xmath72 . if @xmath633 , then @xmath634 is injective with torsion free cokernel .
therefore @xmath635 is an isomorphism , since @xmath78 .
let @xmath21 be the ample generator of @xmath636 and assume @xmath637 for some @xmath638 .
note that @xmath639 .
it then follows that @xmath640 therefore @xmath544 and @xmath619 .
next , kodaira vanishing and the exact sequence @xmath641 give @xmath642 . hence the exact sequence @xmath643 yields @xmath644 .
then theorem [ kof ] applies to give @xmath645 .
q.e.d . plus
1pt for the first statement of the theorem below , see fujita ( * ? ? ?
* ( 2.12 ) ) or ionescu @xcite ; the second point follows from theorem [ pn ] and some of bdescu s arguments in @xcite ; the third point was noticed in ( * ? ? ? * section 2 ) and @xcite .
[ fb ] let @xmath1 be an @xmath44-dimensional projective manifold . 1 .
let @xmath646 be a surjective morphism from @xmath1 onto a normal variety @xmath7 .
let @xmath21 be an ample line bundle on @xmath1 .
assume that @xmath647 for a general fiber , @xmath316 , of @xmath456 and that all fibers of @xmath456 are @xmath648-dimensional .
then @xmath646 is a linear @xmath9-bundle with @xmath649 .
2 . let @xmath81 be a @xmath9-bundle over a projective manifold @xmath7 .
assume that @xmath3 is an ample divisor on @xmath1 .
furthermore assume that @xmath6 extends to a morphism @xmath83 .
then there exists a non - splitting exact sequence @xmath650 where @xmath651 , @xmath46 are ample vector bundles on @xmath7 such that @xmath652 , @xmath653 , @xmath6 , @xmath59 are the bundle projections on @xmath7 , and the inclusion @xmath2 is induced by @xmath480 .
3 . let @xmath646 be a linear @xmath82-bundle over a projective manifold @xmath7 .
assume that @xmath654 and the tautological line bundle of @xmath1 , say @xmath21 , is ample .
then @xmath655 is nef .
moreover , the bundle projection @xmath456 is associated to the linear system @xmath656 for @xmath175 ( i.e. , @xmath90 is a scroll over @xmath7 ) .
\(i ) following the argument as in @xcite , let us first show that @xmath7 is smooth .
indeed , let @xmath657 be a closed point and denote by @xmath197 the fiber over @xmath658 .
consider the embedding of @xmath1 given by @xmath659 for @xmath175 .
let @xmath660 be the smooth @xmath661-fold got by intersecting @xmath648 general members @xmath662 of @xmath659 .
since @xmath647 for a general fiber @xmath316 , the restriction @xmath102 of @xmath6 to @xmath660 has degree @xmath663 . since @xmath6 has equidimensional fibers , @xmath664 is a @xmath665-dimensional scheme , and its length is given by @xmath666 .
since @xmath7 is normal , it follows by a well known criterion ( see e.g. , ( * ? ? ?
* chapter ii , theorem 6 ) ) that the above inequality is in fact an equality .
hence in particular @xmath667 , so that @xmath197 is irreducible and generically reduced .
therefore , by the generality of @xmath662 , we may assume that @xmath664 is a reduced @xmath665-cycle consisting of @xmath668 distinct points .
it thus follows that @xmath102 is tale over @xmath658 .
therefore @xmath7 is smooth at @xmath658 since @xmath660 is smooth .
since all fibers of @xmath456 are equidimensional and @xmath1 and @xmath7 are smooth , the morphism @xmath456 is flat . let now @xmath197 be any fiber of @xmath456 .
we have seen above that @xmath197 is irreducible and generically reduced .
moreover , @xmath197 is cohen
macaulay since every fiber is defined by exactly @xmath669 coordinate functions .
it thus follows that @xmath197 is in fact reduced . by the semicontinuity theorem for dimensions of spaces of sections on fibers of a flat morphism ( *
* chapter iii , theorem 12.8 ) , @xmath670 .
then by theorem [ kof ] we conclude that @xmath671 for every fiber @xmath197 of @xmath456 .
then @xmath672 , where @xmath673 .
\(ii ) note that , by ampleness of @xmath3 , @xmath6 equidimensional implies @xmath59 equidimensional .
let @xmath316 be a general fiber of @xmath59 and let @xmath674 be the corresponding fiber of @xmath6 . by ( i ) , it is enough to show that @xmath675 and @xmath676 , where @xmath677 . if @xmath678 we conclude by theorem [ pn ] .
thus we can assume @xmath679 . by taking general hyperplane sections of @xmath7 and by base change
, we can also assume that @xmath7 is a smooth curve ( and hence @xmath316 is a divisor ) .
here we follow bdescu s argument . by the lefschetz theorem
, we have an exact sequence @xmath680 where @xmath635 is the morphism induced by the embedding @xmath303 and @xmath681 is torsion free .
first note that @xmath682 , since otherwise @xmath316 would be an ample divisor . since @xmath683
, we thus conclude that @xmath684 , and therefore that @xmath685 since it is torsion free .
let @xmath686 be a section of the bundle @xmath81 .
then @xmath687 and @xmath688 for some line bundle @xmath73 on @xmath1 inducing @xmath686 on @xmath3 .
write @xmath689 for some integers @xmath171 , @xmath690 . since @xmath691 , we get @xmath692 and therefore @xmath693
. then @xmath694 .
by ampleness of @xmath3 , it must be @xmath695 .
thus @xmath696 has self - intersection @xmath697 on @xmath316 .
this implies @xmath698 and @xmath699 by using theorem [ pn ] .
applying the first part of the statement and pushing forward under @xmath59 the exact sequence @xmath700 we get the desired conclusion .
\(iii ) if @xmath701 is not nef , there exists an extremal ray @xmath158 of @xmath1 such that @xmath702 .
it follows that @xmath703 .
therefore , if @xmath197 is a positive dimensional fiber of the contraction @xmath704 , we have @xmath705 ( see theorem [ exrayiw ] ) .
then , for a fiber @xmath316 of @xmath456 , we have @xmath706 ( where the last inequality follows from the assumption @xmath707 , which is equivalent to saying that @xmath708 ) .
hence @xmath709 .
thus there exists a curve @xmath710 such that @xmath711 , contradicting @xmath712 .
therefore we conclude that @xmath655 is nef , and hence , by theorem [ krs ] , the linear system @xmath656 defines a morphism , say @xmath177 , for @xmath175 .
let now @xmath713 be an extremal ray .
we have @xmath714 , so that @xmath715 , and hence , as above , @xmath716 . since @xmath717 , it follows , again by theorem [ exrayiw ] , that @xmath718 . then @xmath719 .
this implies that @xmath456 is the contraction @xmath720 of the extremal ray @xmath158 .
since this is true for each extremal ray as above , we conclude that the face @xmath721 is in fact @xmath141-dimensional and that @xmath6 coincides with the morphism @xmath177 .
q.e.d . plus
1pt [ inpart ] ( 1 ) in the boundary case @xmath722 of theorem [ fb](iii ) , the same argument gives the nefness of @xmath655 ; moreover , further considerations show that the bundle projection @xmath456 is associated to @xmath656 for @xmath175 unless @xmath723 .
we refer for this to ( * ? ? ?
* ( 3.1 ) ) .
\(2 ) note that by theorem [ basere ] or lemma [ easy ] below , statement ( iii ) of theorem [ fb ] applies under the conditions in [ fb](ii ) . in the case of @xmath9-bundles ,
sommese s theorem [ basere ] admits the following simple alternative proof .
[ easy ] let @xmath3 be a smooth ample divisor on a projective manifold , @xmath1 .
assume that @xmath3 is a @xmath9-bundle over a manifold @xmath7 .
further assume that @xmath6 has an extension @xmath83 .
then @xmath724 . by theorem [ fb ]
we get the exact sequence @xmath725 where @xmath651 , @xmath46 are ample vector bundles on @xmath7 . arguing by contradiction , assume @xmath726 , that is @xmath727 . by le potier s vanishing theorem @xcite we have @xmath728 for @xmath729 .
therefore @xmath730 , so we conclude that ( [ e ] ) splits , contradicting ampleness of @xmath651 .
q.e.d . plus
1pt [ key1 ] let @xmath21 be an ample line bundle on a projective manifold , @xmath1 , of dimension @xmath731 .
assume that there is a smooth @xmath732 such that @xmath3 is a @xmath9-bundle , @xmath81 , over a manifold @xmath7 .
let @xmath733 be a line in a fiber @xmath9 of @xmath6 .
further assume that @xmath734 for some ample line bundle @xmath73 on @xmath1 .
then @xmath735 for an ample rank @xmath736 vector bundle , @xmath46 , on @xmath7 with @xmath6 equal to the restriction to @xmath3 of the induced projection @xmath737 .
lines in the fibers of @xmath6 define a covering family of @xmath3 . by our assumptions ,
the induced family on @xmath1 is unsplit .
therefore theorem [ occhetta ] applies to give that @xmath6 extends .
we conclude by theorem [ fb](ii ) .
q.e.d . plus
1pt the following gives a precise answer to question [ zeroq ] .
[ qnot ] let @xmath1 be an @xmath44-dimensional projective manifold embedded in @xmath70 , @xmath72 .
* assume that @xmath1 has a smooth hyperplane section @xmath738 which is a @xmath9-bundle over a manifold @xmath7 , say @xmath739 , such that the fibers of @xmath6 are linear subspaces of @xmath70 .
then @xmath6 lifts to a linear @xmath82-bundle @xmath740 .
moreover , this is possible only if @xmath741 . *
conversely , assume that @xmath742 is a @xmath82-bundle with linear fibers and @xmath743
. then there exists a smooth hyperplane section @xmath738 which is a @xmath9-bundle .
\(i ) if @xmath733 is a line contained in some fiber of @xmath6 , we have @xmath734 .
thus the first assertion follows from lemma [ key1 ] and the second from theorem [ basere ] ( or lemma [ easy ] ) .
\(ii ) consider the incidence relation @xmath744 where @xmath745 is the fiber @xmath82 over a point @xmath657 and @xmath746 is the point corresponding to the hyperplane @xmath73 in @xmath70 .
then @xmath747 , so that @xmath748 gives @xmath749
. therefore there exists a hyperplane @xmath73 in @xmath70 not containing fibers of @xmath456 and we are done .
q.e.d . plus
1pt consider the setting as in diagram ( [ diagram ] ) from section 2 with @xmath750 .
let us discuss here some applications under the assumption that the canonical bundle @xmath751 of @xmath7 is nef .
we follow the exposition in ( * ? ? ?
* section 4 ) , where the results are proved for a strictly nef and big divisor @xmath3 on @xmath1 . as a first application
, we consider the case when the morphism @xmath81 is a @xmath9-bundle .
an analogous result , assuming @xmath752 instead of @xmath751 to be nef , was proved in @xcite in a completely different way .
the following result is essentially due to winiewski ( * ? ?
* ( 3.3 ) ) .
[ palunique ] let @xmath3 be a @xmath9-bundle over a smooth projective variety and let @xmath81 be the bundle projection . if @xmath751 is nef , then @xmath3 admits a unique extremal ray , and @xmath6 is its contraction .
assume by contradiction that there exists an extremal rational curve , @xmath23 , which is not contracted by @xmath6 .
let @xmath753 be the normalization of @xmath754 .
let @xmath755 be the induced morphism and consider the base change diagram @xmath756^(.50)g \ar[d]_{p ' } & y \ar[d]^p \\
\gamma \ar[r]_{f } & z.}\ ] ] since @xmath438 is a smooth curve , we have a vector bundle @xmath46 on @xmath438 , of rank @xmath757 , such that @xmath758 .
let @xmath759 be the fiber of the bundle projection @xmath760 and let @xmath761 be a curve mapped onto @xmath23 under @xmath762 .
let @xmath763 be the tautological line bundle on @xmath764 .
then we get @xmath765 since @xmath751 is nef , it thus follows @xmath766 . by the grothendieck theorem
, we have @xmath767 , where @xmath768 . then the inequality above yields @xmath769 .
thus @xmath770 let @xmath193 $ ] be the extremal ray generated by @xmath23 .
the composition @xmath771 is a morphism defined by a linear sub - system of @xmath772 , for some @xmath460 . since @xmath773 , we get @xmath774
. therefore ( [ p1 ] ) gives @xmath775 let @xmath776 be the section of @xmath760 corresponding to the surjection @xmath777 .
then @xmath778 , which contradicts ( [ p2 ] ) .
q.e.d . plus
1pt [ lucian1 ] let @xmath81 be a @xmath9-bundle over a projective manifold @xmath7 .
assume that @xmath751 is nef .
if @xmath3 is an ample divisor on a manifold @xmath1 , then @xmath6 extends to a morphism @xmath83 .
furthermore there exists a non - splitting exact sequence @xmath779 such that @xmath652 , @xmath780 , and @xmath6 , @xmath59 are the bundle projections on @xmath7 . by lemma [ palunique ] ,
the bundle projection @xmath6 is the contraction of the unique extremal ray on @xmath3 .
then , by corollary [ newcor ] , @xmath6 extends . thus theorem [ fb](ii ) applies to give the result .
q.e.d . plus
1pt next , we consider the case when the morphism @xmath81 is a blowing - up .
the following general fact is a direct consequence of lemma [ palunique ] .
[ palbup1 ] let @xmath7 be a projective manifold .
let @xmath81 be the blowing - up along a smooth subvariety @xmath781 of codimension @xmath68 .
assume that the canonical bundles @xmath782 and @xmath751 are nef .
then @xmath3 has only one extremal ray .
let @xmath783 be the codimension of @xmath781 in @xmath7 , and let @xmath189 be the exceptional divisor of @xmath6 .
let @xmath193 $ ] be any extremal ray on @xmath3 .
since @xmath784 , we conclude that @xmath785 .
therefore @xmath23 is contained in @xmath189 and @xmath786 .
then apply the proof of lemma [ palunique ] to the @xmath787-bundle @xmath788.q.e.d .
plus 1pt the following generalizes a result due to sommese concerning the reduction map in the case of threefolds ( see @xcite , ( * ? ? ?
* theorem i ) , and also @xcite ) and is closely related to fujita s results in @xcite . [ palbup2 ]
let @xmath7 be a projective manifold .
assume that @xmath751 is nef .
let @xmath81 be the blowing - up along a smooth subvariety @xmath781 of codimension @xmath789 , such that @xmath782 is nef .
if @xmath3 is an ample divisor on a manifold @xmath1 , then there exists a commutative diagram , where @xmath56 is a smooth projective variety and either 1 .
@xmath375 is the blowing - up of @xmath56 along the image of @xmath781 .
moreover , @xmath790 is an ample divisor on @xmath56 whenever @xmath781is @xmath665-dimensional ; or 2 .
@xmath1 is generically a @xmath10-bundle over @xmath7 and @xmath3 is a rational section of it .
moreover , @xmath791 , @xmath792 and fibers of @xmath59 are at most two - dimensional . by lemma [ palbup1 ]
, @xmath3 contains a unique extremal ray and @xmath6 is its contraction . then , by corollary [ newcor ] , @xmath6 extends to a contraction @xmath375 of an extremal ray on @xmath1 , which gives rise to a commutative diagram ( [ diagram ] ) .
assume first that @xmath76 is birational .
let @xmath189 , @xmath793 be the exceptional loci of @xmath6 , @xmath59 respectively , so that @xmath794 . as the restriction @xmath795 of @xmath6 to @xmath189 is a @xmath787-bundle , it follows from theorem [ fb](ii ) that @xmath796 is a @xmath797-bundle .
it is now standard to see that @xmath56 is smooth , @xmath7 is contained in @xmath56 as a divisor and @xmath76 is the blowing - up of @xmath56 along @xmath781 , cf .
also ( * ? ? ?
* section 5 ) . in (
* section 5 ) it is also proved that @xmath798 is an ample divisor on @xmath56 under the extra assumption that the restriction to @xmath781 of the line bundle @xmath799 is ample . in particular
, @xmath790 is ample on @xmath56 if @xmath781 is @xmath665-dimensional .
now , assume that @xmath76 is not birational .
then @xmath800 from ( [ diagram ] ) is an isomorphism and @xmath3 is a rational section for @xmath76 , which is generically a @xmath10-bundle .
let @xmath801 be a general point and let @xmath802 be a line .
we put @xmath803 and we denote by @xmath85 a general fiber of @xmath76 . since @xmath804 is contracted by @xmath76 , it is numerically proportional to @xmath85 .
it follows easily that @xmath805 as 1-cycles . as @xmath806 , we get @xmath807 .
so @xmath808 .
assume that @xmath809 .
we obtain that @xmath810 , a contradiction .
so @xmath791 .
now , denote by @xmath210 the family of all deformations of @xmath804 in @xmath1 .
we claim that @xmath811 . assuming the contrary
, we would have @xmath812 . from the exact sequence @xmath813 using standard facts from deformation theory of rational curves , we find that @xmath814 .
thus , if @xmath222 is a point on @xmath804 , @xmath815 but the same exact sequence gives @xmath816 this is a contradiction and the claim is proved . since @xmath791 , the claim implies that some deformation of @xmath804 equals a fiber of @xmath76 . in particular , @xmath544
, so @xmath792 and we are done .
q.e.d . plus
1pt [ furem ] let us explicitly point out that an analogous result was proved by fujita ( * ? ? ?
* section 5 ) , under the assumption that @xmath817 , but with no nefness condition on @xmath782 and @xmath751 .
however , the contraction morphism @xmath818 obtained in @xcite is in general analytic , not necessarily projective .
next , let us consider the case when @xmath3 admits a pluricanonical fibration .
recall that a _ calabi yau manifold _ @xmath3 is a projective variety with trivial canonical bundle and @xmath819 for @xmath820 .
[ iocy ] let @xmath1 be a projective manifold of dimension @xmath821 .
let @xmath3 be a smooth ample divisor on @xmath1 .
assume that @xmath342 is nef .
then the linear system @xmath822 is base points free for @xmath175 .
if @xmath342 is numerically trivial , then @xmath3 is a calabi
yau variety and @xmath1 is a fano manifold .
if @xmath823 is a trivial cycle and @xmath824 is non - trivial for @xmath825 , then @xmath1 is a fano fibration over @xmath7 and @xmath826 .
the proof runs parallel to that of theorem [ six ] .
since @xmath342 is nef and @xmath3 is ample , we conclude that @xmath827 is nef . thus @xmath828 is spanned for @xmath175 by the kawamata
shokurov base point free theorem , and it defines a morphism @xmath829 . by restricting to @xmath3 we find that @xmath830 is base points free for @xmath175 .
then @xmath3 admits a pluricanonical map , say @xmath831 . if @xmath832 , the morphism @xmath177 is a fibration . if @xmath342 is numerically trivial , then @xmath827 is also , and thus @xmath1 is a fano manifold .
hence in particular @xmath833 for @xmath834 , so that @xmath3 is a calabi yau manifold .
the remaining part of the statement is clear .
plus 1pt we say that a projective manifold @xmath3 is _ extendable _ if there exists a projective manifold @xmath1 such that @xmath2 is an ample divisor .
proposition [ iocy ] shows that a manifold @xmath3 such that @xmath342 is numerically trivial and either @xmath342 is not linearly trivial or @xmath835 , e.g. , @xmath3 an abelian variety , is not extendable .
it is worth noting that proposition [ iocy ] shows that the abundance conjecture @xcite holds true for extendable manifolds .
let us recall what the conjecture says in the smooth case .
_ let @xmath3 be a projective manifold with @xmath342 nef .
then @xmath836 is spanned by its global sections for @xmath175_. we refer to @xcite for a further discussion in the case when @xmath3 is a strictly nef divisor on @xmath1 .
throughout this section we assume that @xmath1 is a smooth projective three fold and @xmath837 is a smooth ample divisor .
the following theorem implies a number of results from @xcite and @xcite .
we follow the arguments in @xcite .
note that @xcite contains a precise description of all types of extremal rays of @xmath1 .
[ 6.1 ] ( cf .
@xcite ) a _ reduction _ of the pair @xmath838 is another pair @xmath839 , with @xmath840 a smooth ample divisor , such that @xmath340 is got by contracting all @xmath841 planes @xmath842 contained in @xmath1 such that @xmath843 and @xmath764 is just the image of @xmath3 in @xmath340 .
[ 6.2 ] let @xmath844 be as above and assume that @xmath342 is not nef .
then one of the following holds .
* @xmath845 , @xmath1 is fano , of index @xmath846 and either : * * @xmath847 , @xmath848 , @xmath849 ; or * * @xmath850 , @xmath851 , @xmath852 ; or * * @xmath1 is a del pezzo three fold , @xmath853 , cf .
@xcite or @xcite for a complete list .
* @xmath1 is a linear @xmath854-bundle over a curve and , for each fiber @xmath316 , either @xmath855 or @xmath856 ; * @xmath1 admits a contraction of an extremal ray , @xmath857 , such that @xmath56 is a ( smooth ) curve , we have @xmath858 for a general fiber of @xmath185 and @xmath859 ( we call @xmath185 a _ quadric fibration _ ) ; * @xmath1 is a linear @xmath10-bundle over a surface and @xmath3 is a rational section ; * a reduction @xmath839 of @xmath838 exists .
since @xmath827 is not nef , there exists an extremal ray @xmath860 $ ] of @xmath1 such that @xmath861 .
consider the length @xmath195 of @xmath158 and observe that we have @xmath862 .
let @xmath863 be the contraction of @xmath158 and let @xmath316 be a general fiber of @xmath185 .
if @xmath864 , we fall in case ( i ) .
so , from now on , we may assume @xmath865 . if @xmath866 , by theorem [ exrayiw ] , @xmath867 . if @xmath868 , by theorem [ exrayiw ] , @xmath56 is a curve .
moreover , @xmath869 or @xmath329 . by corollary [ co ] ,
@xmath870 . if @xmath871 , we get case ( ii ) , @xmath872 by theorem [ fb ] .
assume now that @xmath873 ( and @xmath874 ) .
let @xmath875 and let @xmath876 be an arbitrary fiber of @xmath185 .
remark that @xmath877 , therefore @xmath878 for any curve @xmath879 .
we have that @xmath880 and @xmath881 is ample by the nakai moishezon criterion . by theorem [ kof ]
, @xmath185 makes @xmath1 a @xmath854-bundle and @xmath856 .
thus , when @xmath874 and @xmath56 is a curve , we get case ( ii ) . next
, suppose that @xmath882 , so @xmath869 . if @xmath56 is a curve we get @xmath883 and we deduce from remark [ re ] that @xmath884 , leading to case ( iii )
. if @xmath56 is a surface , we get case ( iv ) .
indeed , @xmath185 is generically a @xmath10-bundle , @xmath3 being a rational section .
so it is enough to see that all fibers of @xmath185 are one - dimensional .
let @xmath268 be an irreducible surface contracted by @xmath185 and let @xmath885 .
we obtain @xmath886 and @xmath887 since @xmath268 is contracted by @xmath185 .
this contradiction shows that @xmath185 is a @xmath10-bundle .
finally , assume that @xmath185 is birational .
for such a ray , it follows from theorem [ exrayiw ] that @xmath189 , the locus of @xmath158 , is an irreducible surface , contracted to a point .
moreover , @xmath888 since @xmath158 is not nef .
we get @xmath889 ; as above , using suitable vanishings , ( see @xcite for details ) we deduce that @xmath890 , @xmath891 and @xmath892 .
this leads to the reduction from case ( v ) .
[ 6.3](@xcite ) let @xmath838 be as above and assume that @xmath893 is a @xmath10-bundle .
then @xmath6 extends to a linear @xmath854-bundle @xmath894 , unless either @xmath895 , @xmath896 , or @xmath850 , @xmath897 , @xmath6 is one of the projections and the other projection extends .
assume first that @xmath845 .
the conclusion follows by looking at the list in theorem [ 6.2](i ) , using the classification of del pezzo three folds .
next , suppose @xmath898 . by lefschetz s theorem , we get an isomorphism @xmath899
. then corollary [ newcor ] applies to give that some extremal ray of @xmath3 extends to @xmath1 .
if the genus @xmath900 , such a ray is unique and its contraction , @xmath6 , extends . apply theorem [ fb ] to conclude .
assume @xmath901 . unless either @xmath902 or @xmath903 , @xmath3 has only one extremal ray , so the previous argument applies .
to conclude , we only have to examine the case when the contraction of the @xmath841 curve of @xmath904 , say @xmath905 , extends to a morphism @xmath906 .
_ case _ 1 .
@xmath907 is the contraction of a @xmath841 plane @xmath189 .
the diagram @xmath908_{\pi}\ar@{^{(}->}[r ] & x\ar[d]^{\overline\pi } \\ { { \mathbb p}^{2}}\cong z\ar@{^{(}->}[r ] & w}\end{xy}\ ] ] shows that @xmath909 , @xmath910 ( see theorem [ pn ] ) .
let @xmath911 .
we get @xmath912 , contradicting ampleness of @xmath3 .
_ case _ 2 .
@xmath913 is a @xmath10-bundle , @xmath3 is a rational section ( see the argument from the proof of theorem [ 6.2 ] , case ( iv ) ) .
let @xmath85 be a fiber of @xmath6 and let @xmath455 be the ( @xmath914 ) curve contracted by @xmath456 . write @xmath915 , for some @xmath421 . from @xmath916
it follows @xmath917 .
let @xmath918 , @xmath919 a line .
since @xmath122 , there is some @xmath920 such that @xmath921 .
we find easily that @xmath922 . now consider the exact sequence @xmath923 we have @xmath924 .
therefore , using also the ampleness of @xmath3 , it follows that the linear system @xmath925 gives a morphism @xmath926 which extends @xmath6 . clearly , @xmath927 is a @xmath854-bundle and , in fact , @xmath928 .
a classification of all cases when @xmath3 is birationally ruled also follows from theorem [ 6.2 ] .
@xcite ) let @xmath838 be as above and assume that @xmath3 is not birationally ruled . then
, either @xmath1 is a ( linear ) @xmath10-bundle and @xmath3 is a rational section , or there is a reduction @xmath929 such that @xmath930 is nef .
looking over the cases ( i)(v ) in theorem [ 6.2 ] and using the hypothesis that @xmath3 is not ruled , we see that only cases ( iv ) and ( v ) are possible .
( @xcite ) let @xmath838 be as above .
assume that @xmath1 is not a @xmath10-bundle , @xmath3 being a rational section .
* if @xmath931 , there is a reduction @xmath929 of @xmath838 such that @xmath932 is a @xmath933 surface and @xmath934 is fano . * if @xmath935 , there is a reduction @xmath936 of @xmath838 such that @xmath934 fibers over a curve , with general fiber a del pezzo surface . use the preceding corollary and proposition [ iocy ] .
we start with the following proposition .
[ conicbdlethm ] let @xmath1 be a projective manifold of dimension @xmath72 .
let @xmath3 be a smooth ample divisor on @xmath1 .
assume that @xmath3 is a conic fibration , with general fiber @xmath85 .
let @xmath210 be the family of rational curves induced by @xmath85 on @xmath1 .
then the following conditions are equivalent : 1 .
@xmath937 ; 2 .
@xmath210 is unsplit ; 3 .
@xmath210 is locally unsplit .
if ( i)(iii ) hold , then @xmath6 is a @xmath10-bundle which extends to @xmath83 .
moreover , @xmath938 and @xmath59 is a @xmath854-bundle .
conversely , if @xmath6 is smooth and extends , conditions ( i)(iii ) hold .
since @xmath939 are clear , it is enough to show @xmath940 .
let @xmath941 , let @xmath942 be a fixed general point and consider the standard exact sequence @xmath943 since @xmath944 , general facts from deformation theory of rational curves say that @xmath945 , and hence @xmath946 , see [ families ] . by semicontinuity ,
the same holds at a general point @xmath219 .
fix such a general point @xmath219 and take another point @xmath947 . since @xmath210 is locally unsplit , we know that each curve from @xmath220 is irreducible . by the non - breaking lemma
, we thus conclude that there is a finite number of curves in @xmath220 passing through @xmath948 .
that is the projection @xmath949 is a finite map .
therefore @xmath950 .
thus we obtain @xmath951 .
assume by contradiction that @xmath475 .
then there exists a curve @xmath952 such that @xmath754 is a curve in @xmath7 . in this case , a variant of the non - breaking lemma ( see ( * ? ? ?
* ( 1.14 ) ) and also ( * ? ? ?
* ( 1.4.5 ) ) ) implies that the curve @xmath23 is numerically equivalent in @xmath1 to @xmath953 for some positive rational number @xmath954 .
now , take a hyperplane section @xmath955 of @xmath7 and let @xmath956 be the extension of @xmath957 on @xmath1 via the isomorphism @xmath231 .
in particular , @xmath958 , this leading to the numerical contradiction @xmath959 .
if ( i)(iii ) hold , @xmath6 extends to a @xmath854-bundle by lemma [ key1 ] .
moreover , @xmath938 by lemma [ easy ] .
conversely , if @xmath6 extends , we have ( i ) by theorem [ fb ] .
q.e.d . plus
1pt we consider now the extension problem for @xmath10-bundles . for perspective
we also recall the ( much easier ) case of @xmath9-bundles , for @xmath960 .
the following major conjecture on the topic ( * ? ? ?
* section 5.5 ) describes all known examples .
we refer to ( * ? ? ?
* section 5.5 ) for the more general case when @xmath1 is a local complete intersection .
[ pdbdleconj ] let @xmath21 be an ample line bundle on a projective manifold , @xmath1 , of dimension @xmath961 .
assume that there is a smooth @xmath732 such that @xmath3 is a @xmath9-bundle , @xmath81 , over a manifold , @xmath7 , of dimension @xmath690 .
then @xmath962 and @xmath735 for an ample vector bundle , @xmath46 , on @xmath7 with @xmath6 equal to the restriction to @xmath3 of the induced projection @xmath737 , except if either : 1 .
@xmath963 ; or 2 .
@xmath964 ; or 3 .
@xmath965 , @xmath6 is the product projection onto the second factor , @xmath966 for an ample vector bundle , @xmath46 , on @xmath10 with the product projection of @xmath3 onto the first factor equal to the induced projection @xmath967 .
the inequality @xmath968 is a necessary condition for @xmath81 to extend by lemma [ easy ] .
the conjecture has been shown except when @xmath679 , @xmath969 , and the base @xmath7 does not map finite - to - one into its albanese variety .
the case when either @xmath970 or @xmath7 is a submanifold of an abelian variety follows from sommese s extension theorems @xcite ( see also fujita @xcite ) .
this argument works also in the case when @xmath7 maps finite - to - one into its albanese variety ( see ( * ? ? ?
* ( 5.2.3 ) ) ) .
[ d2 ] ( sommese ) conjecture [ pdbdleconj ] is true for @xmath971 . since the result is trivial if @xmath7 is a point we can assume without loss of generality that @xmath972 and thus that @xmath973 . from theorem [ extensionthm ] we know that @xmath974 extends to a morphism , @xmath975 .
the result follows from theorem [ fb](ii ) .
q.e.d . plus
1pt the conjecture is also known when @xmath679 and @xmath976 .
if @xmath695 , the result is due to bdescu @xcite ; we have seen a proof in corollary [ 6.3 ] .
the case @xmath977 is due to the contribution of several authors : fania and sommese @xcite , fania , sato and sommese @xcite , sato and spindler @xcite and also @xcite .
below we propose a shorter proof .
the basic ideas are those in @xcite and @xcite , but we do not use the difficult papers @xcite and @xcite .
[ mainpd ] conjecture [ pdbdleconj ] is true when @xmath679 and @xmath977 .
assume that @xmath6 does not extend .
_ step _ 1 .
@xmath7 is ruled . from corollary [ newcor ]
it follows that @xmath3 has some extremal ray , say @xmath158 , which extends to an extremal ray @xmath978 on @xmath1 .
using @xcite , we consider the possible type of @xmath158 . if @xmath158 is nef , @xmath7 is covered by rational curves in the fibers of @xmath704 , so it is ruled . if @xmath158 is not nef , @xmath189 , the locus of @xmath158 , covers @xmath7 ( so again @xmath7 is ruled ) , unless @xmath979 and @xmath980 is a curve .
standard computations ( cf . also poposition [ palbup2 ] ) show that @xmath23 is a @xmath841 curve and we may construct a commutative diagram @xmath981_{{\rm cont}_c } & \ar[l]_p y \ar[d]^{{\rm cont}_r } \ar@{^{(}->}[r ] & x\ar[d]^{{\rm cont}_{\overline r}}\\ z'&\ar[l]^{p ' } y ' \ar @{^{(}->}[r ] & x'}\end{xy}\ ] ] where @xmath764 is ample on @xmath340 ( cf .
@xcite ) and @xmath760 is again a @xmath10-bundle .
so , after finitely many steps , we conclude that @xmath7 is ruled .
_ step _ 2 .
assume the contrary .
as @xmath7 is ruled , there is a morphism @xmath983 which is generically a @xmath10-bundle .
apply theorem [ extensionthm ] to the map @xmath984 to get an extension @xmath985 .
next we use corollary [ 6.3 ] fiberwise .
let @xmath986 , @xmath987 be the general fibers of @xmath988 respectively .
denote by @xmath85 , @xmath455 a fiber and a minimal section of @xmath316 , respectively .
note that the classes of @xmath85 and @xmath455 are not proportional in @xmath989 .
then the diagram @xmath990\ar[d]&n_1(\overline f)\ar[d]\\n_1(y)\ar[r]^\sim & n_1(x)}\end{xy}\ ] ] shows that @xmath991 .
so , from corollary [ 6.3 ] we infer that either @xmath937 , or @xmath992 . in the first case
, @xmath6 extends by lemma [ key1 ] .
so we may assume that @xmath992 from now on .
we refer to @xcite for details concerning the next few arguments . using standard properties of hilbert schemes ,
one shows : * @xmath993 is a @xmath10-bundle ; * any fiber of @xmath456 is isomorphic to @xmath994 ; * the family of curves on @xmath3 determined by minimal sections of the map @xmath995 yields another @xmath10-bundle @xmath996 .
we deduce a cartesian diagram @xmath997_p y\ar[dr]^{p'}&\\z\ar[dr]_\varphi & & z'\ar[dl]^{\varphi'}\\&b&}\end{xy}\ ] ] where @xmath998 is also a @xmath10-bundle .
next , from the above construction , corollary [ 6.3 ] and theorem [ fb ] , we find that @xmath760 extends to a linear @xmath854-bundle @xmath999 . *
this yields an exact sequence @xmath1000 where @xmath1001 are ample vector bundles on @xmath1002 .
if @xmath1003 , one finds @xmath1004 , where @xmath46 is a rank @xmath329 vector bundle on @xmath440 , @xmath1005 and @xmath421 .
moreover , from ( c ) it follows that @xmath1006 . *
if we assume @xmath46 unstable , it is now standard to see that the exact sequence from ( d ) splits .
this is a contradiction , since @xmath15 is ample .
finally , see @xcite for a proof that the case @xmath46 stable also leads to a contradiction .
_ step _ 3 . conclusion .
we know that @xmath982 , so proposition [ fspb ] below applies to give the result .
let us also explicitly note that in the relevant case @xmath679 , @xmath1007 ( by lemma [ easy ] the bundle @xmath81 does not extend in this case ) conjecture [ pdbdleconj ] is equivalent to saying that * _ a @xmath10-bundle @xmath3 , @xmath81 , over a manifold @xmath7 can not be an ample divisor in an @xmath44-dimensional manifold @xmath1 unless @xmath1008 , @xmath1009 , and @xmath1 is a @xmath503-bundle over @xmath10 whose restriction to @xmath3 is the projection @xmath539 on the first factor_. the following proposition from @xcite ensures that , to prove conjecture [ pdbdleconj ] , it is enough to show that @xmath1009 , assuming that @xmath6 does not extend .
[ fspb ] ( ( * ? ? ?
* section 2 ) ) let @xmath3 be a smooth ample divisor on an @xmath44-dimensional projective manifold @xmath1 .
assume that @xmath81 is a @xmath10-bundle over @xmath1010 , @xmath1011 .
if @xmath6 does not extend to @xmath1 , then @xmath1012 and @xmath1 is a @xmath1013-bundle over @xmath10 .
let @xmath1014 and let @xmath1015 .
looking over the proof of theorem [ extensionthm ] we see that , once we have the vanishings in , the argument works also in the case @xmath1016 .
therefore , the assumption that @xmath6 does not extend translates into : there exists some @xmath1017 such that @xmath1018 . using serre duality , kodaira vanishing , and the exact sequence @xmath1019 it follows that @xmath1020 iterating this construction , we may assume that @xmath977 ; in this case @xmath1021 , for some @xmath1022 .
we write @xmath1023 for some rank @xmath329 vector bundle on @xmath1024 .
we also may assume that , if @xmath1025 is a line , we have @xmath1026 .
we shall prove that @xmath1027 , so that @xmath46 is trivial ( see ( * ? ? ?
* section 3.2 ) ) and the conclusion follows .
so , assume that @xmath977 , @xmath1021 and write @xmath1028 , using the notation from ( * ? ? ?
* chapter v.2 ) . since @xmath21 is ample , @xmath421 and @xmath1029 .
hence , for @xmath1017 , @xmath1030 . therefore , either @xmath1031 and @xmath1032 is ample , or @xmath1033 and @xmath1034 .
now , if @xmath1035 , @xmath1036 is nef and big . using kawamata
viehweg vanishing this contradicts .
so @xmath1027 and we are done . further evidence for conjecture [ pdbdleconj ] is given by the following result ( see @xcite for a proof ) .
[ flemma ] ( @xcite ) let @xmath81 be a @xmath10-bundle over a smooth projective three fold @xmath7
. then @xmath3 can not be a very ample divisor in any projective manifold @xmath1 , unless @xmath1037 and @xmath1038 .
[ beautyfulmind ] the following discussion gives some further support to conjecture [ pdbdleconj ] , in relation to the content of section 4 .
let @xmath81 be a smooth @xmath10-bundle with @xmath3 ample divisor in a projective manifold @xmath1 of dimension @xmath72 . if @xmath6 does not extend to @xmath1 , and we assume the hypothesis of proposition [ sexy ] is fulfilled , then the following three conditions hold true .
1 . @xmath1039 ; 2 .
@xmath3 is a fano manifold ; 3 .
@xmath7 is a fano manifold ( and @xmath1040 ) .
condition ( 1 ) directly follows from proposition [ sexy ] . to show ( 2 ) , let @xmath519 be the extremal ray corresponding to the bundle projection @xmath6 .
by corollary [ newcor ] we conclude that there exists an extremal ray @xmath523 on @xmath3 which extends to @xmath1 . by ( 1 )
, @xmath1041 and @xmath3 is a fano manifold .
since @xmath3 is a fano manifold and @xmath6 is smooth , @xmath7 is also a fano manifold , see @xcite . in our special case
, we can give the following alternative argument .
assume that @xmath7 is not a fano manifold .
then @xmath751 would be nef ( since @xmath1040 ) .
therefore proposition [ lucian1 ] applies to say that @xmath6 extends ; a contradiction .
whence ( 3 ) holds .
we already observed that , in view of proposition [ fspb ] , to prove conjecture [ pdbdleconj ] it would be enough to show that @xmath1009 .
this is in agreement with the fact that assuming the hypothesis of proposition [ sexy ] to hold , we get condition ( 3 ) above . for instance , when @xmath1042 , the only fano surface @xmath7 with @xmath1040 is @xmath854 , yielding a very short proof of theorem [ mainpd ] .
throughout this section , let @xmath1 be a projective manifold of dimension @xmath72 , let @xmath73 be an ample line bundle on @xmath1 , and let @xmath3 be a smooth divisor in @xmath257 .
in this general setting , it is natural to ask the following questions .
[ q1 ] if @xmath3 is a fano manifold , when is @xmath1 a fano manifold ? [ q2 ] if @xmath1 is a fano manifold , when do we have @xmath1043 ?
question [ q2 ] has been solved in @xcite and @xcite , by using theorem [ aw ] , in the special case described in the following theorem .
[ ne ] ( kollr , winiewski ) let @xmath1 be a fano manifold of dimension @xmath72 and index @xmath1044 , @xmath1045 , for some ample line bundle @xmath21 on @xmath1 .
assume that we are given a smooth member @xmath1046 , for some integer @xmath93 , @xmath1047 .
then there is an isomorphism @xmath1048 .
the result follows from proposition [ reph ] , since @xmath1049 is nef .
q.e.d . plus
1pt let us come back now to question [ q1 ] .
this question is motivated by the problem of classifying polarized pairs @xmath838 as above , when @xmath3 is a fano manifold of large index . set @xmath1050 , where @xmath32 is an ample line bundle on @xmath3 . the first cases to consider , cf .
corollary [ co ] and remark [ re ] , are @xmath1051 , so @xmath3 is a projective space , and @xmath1052 , so @xmath3 is a quadric ; for a solution , see @xcite and also @xcite , where @xmath1 is only assumed to be normal .
the case when @xmath1053 is a classical del pezzo variety , i.e. , @xmath1054 with @xmath32 very ample , has been completely worked out by lanteri , palleschi and sommese @xcite . in @xcite the next case when @xmath1053 is a mukai variety , i.e. , @xmath1055 , is considered .
the results of @xcite have been refined and strengthened in @xcite under the assumption that @xmath32 is merely ample , as a consequence of a comparing cones result which holds true in the range @xmath1056 .
in @xcite the classification is extended to the next case .
we will work under the _ extra assumption _ that the line bundle @xmath32 is spanned .
note that this is in fact the case when @xmath1053 is either a del pezzo variety of degree at least two , or a mukai variety .
this follows from fujita s classification @xcite of del pezzo manifolds and from mukai s classification , see @xcite and @xcite .
we have the following result ( compare with ( * ? ? ? * ( 4.2 ) ) ) . [ bi ]
let @xmath1 be a projective manifold of dimension @xmath72 , let @xmath73 be an ample line bundle on @xmath1 , and let @xmath3 be an effective divisor in @xmath257 .
assume that @xmath3 is a fano manifold of index @xmath1057 , @xmath1050 .
further assume that @xmath32 is spanned .
then either 1 .
there exists an extremal ray @xmath158 on @xmath1 of length @xmath1058 ; or 2 .
@xmath3 contains an extremal ray of length @xmath1059 ; or 3 .
@xmath1 is a fano manifold and @xmath321 .
by the lefschetz theorem , there exists a unique line bundle @xmath21 on @xmath1 such that @xmath1060 .
first , suppose that @xmath1061 is not nef .
then , by the cone theorem , there exists an extremal ray @xmath1062 $ ] on @xmath3 such that @xmath1063 we can also assume that @xmath23 is a minimal curve , that is @xmath1064 .
therefore @xmath1065 .
set @xmath1066 , @xmath1067 .
since @xmath1050 , the last inequality yields @xmath1068 . then @xmath475 with equality only if @xmath679 .
thus @xmath1069 , so that either @xmath1070 , or @xmath1071 and hence @xmath1072 . from @xmath1050
we get @xmath1073 , that is , @xmath20 divides @xmath1074 .
hence in the first case above it must be @xmath1075 , as in the case ( ii ) of the statement .
thus we can assume that @xmath1071 , @xmath425 and the adjunction formula gives @xmath1076 , or @xmath1077 .
therefore @xmath1078 , i.e. , @xmath1079 is not nef .
then there exists a rational curve @xmath140 generating a ray @xmath1080 $ ] on @xmath1 such that @xmath1081 . since we can assume @xmath1082 , it follows that @xmath1083 and we are in case ( i ) of the statement .
assume now that @xmath1061 is nef .
then by the ascent of nefness ( see the proof of theorem [ six ] ) we infer that @xmath1084 is nef and hence by kawamata
shokurov base point free theorem we conclude that @xmath1085 is spanned for @xmath175 .
we proceed by cases , according to the iitaka dimension of @xmath1084 .
let @xmath1086 be the map with normal image and connected fibers associated to @xmath1087 for @xmath175 . if @xmath1088 , then @xmath1089 , so that @xmath1 is a fano manifold . from @xmath1090
we conclude by lefschetz that @xmath1091 .
therefore theorem [ ne ] applies to give @xmath325 .
assume @xmath1092 . since @xmath3 is ample , the restriction @xmath1093 of @xmath1094 to @xmath3 maps onto @xmath56 , so that @xmath1095 since @xmath3 is a fano manifold .
note that @xmath1093 is not the constant map by ampleness of @xmath3 . recalling that @xmath11 is polyhedral
, we conclude that there exists an extremal ray @xmath158 on @xmath3 which is not contracted by @xmath1093 .
let @xmath1096 be the contraction of @xmath158 .
we claim that all fibers of @xmath177 are of dimension @xmath405 .
otherwise , let @xmath197 be a fiber of dimension @xmath68 .
any fiber @xmath316 of @xmath1093 is a divisor on @xmath3 .
then we can find a curve @xmath1097 .
therefore @xmath23 generates @xmath158 and @xmath1098 , contradicting the fact that @xmath158 is not contracted by @xmath1093 . thus by theorem [ aw ] we know that either @xmath177 is a blowing - up of a smooth codimension two subvariety of @xmath7 and @xmath1099 ; or @xmath177 is a conic fibration and @xmath1100
. in each case
, the equality @xmath1101 contradicts the assumption that @xmath1057 .
assume now @xmath1102 .
we follow here the argument from @xcite . from @xmath1103
we get by lefschetz @xmath1104 that is @xmath1105 .
thus we conclude that @xmath1106 and that @xmath1107 is spanned for @xmath175 . therefore the mumford vanishing theorem @xcite ( see also ( * ? ? ?
* ( 7.65 ) ) ) applies to give @xmath1108 now consider the exact sequence @xmath1109 since @xmath32 is spanned on @xmath3 , by ( [ ssvan ] ) we see that sections of @xmath1110 lift to span @xmath21 in a neighborhood of @xmath3 ; but since @xmath3 is ample we conclude that @xmath21 is spanned off a finite set of points .
hence @xmath21 is nef and therefore @xmath87 is ample by ( [ basic ] ) , i.e. , @xmath1 is a fano manifold .
we conclude that either @xmath321 and hence we are done , or there exists an extremal ray @xmath1111 $ ] on @xmath1 , @xmath1112 such that every fiber of the contraction @xmath201 of @xmath158 has dimension at most one . then theorem [ aw ] applies again to say that either : 1 .
@xmath177 is a blowing - up along a smooth codimension two center @xmath440 and @xmath1113 , or 2 .
@xmath177 is a conic fibration and either @xmath1114 or @xmath1113 . in case
( 1 ) , from ( [ basic ] ) and @xmath1113 we get @xmath1115 . since @xmath1116 and @xmath22 it must be @xmath869 , @xmath1117 . note that @xmath1118 and apply proposition [ kollarprop ] to contradict our present assumption that @xmath1119 .
let us consider case ( 2 ) .
if @xmath1114 we have @xmath1120 , giving @xmath873 , @xmath1117 .
if @xmath1113 we get @xmath1115 , giving @xmath869 , @xmath1117 . in both cases , proposition [ kollarprop ] applies again to give the same contradiction as above .
q.e.d . plus
1pt [ anocor ] ( ( * ? ? ?
* ( 4.2 ) ) ) let @xmath1 be an @xmath44-dimensional projective manifold .
let @xmath73 be an ample line bundle on @xmath1 and let @xmath3 be a divisor in @xmath257 .
assume that @xmath3 is a fano manifold of index @xmath1121 ( hence @xmath1122 ) , @xmath1050 and @xmath32 is spanned .
then @xmath1 is a fano manifold and @xmath1043 . by the proof of theorem [ bi ] , either we are done , or @xmath3 contains an extremal ray @xmath1062 $ ] such that either @xmath1075 , or @xmath1123 and @xmath425 . in the first case
we get the numerical contradiction @xmath1124 thus we may assume @xmath1123 and @xmath425 .
therefore , if @xmath197 is a positive dimensional fiber of the contraction @xmath1125 , we have @xmath1126 ( see theorem [ exrayiw ] ) .
thus , either @xmath1127 or @xmath1128 . in the first case ,
the contraction @xmath6 is the constant map , so that @xmath1129 and the conclusion is clear . in the latter case , since @xmath1130 , we know from theorem [ exrayiw ] that @xmath1131 . if @xmath867 we conclude as above .
assume that @xmath56 is a curve and let @xmath316 be a general fiber of @xmath6 . since @xmath3 is a fano manifold , @xmath1095 .
moreover , since @xmath425 , we get @xmath1132 .
corollary [ co ] applies to give @xmath1133 , @xmath1134 .
therefore , by theorem [ fb](i ) , @xmath1135 for some vector bundle @xmath46 on @xmath10 . using lemma [ wk ] and the assumption @xmath1057
, we see that this case does not occur .
q.e.d . plus
let @xmath1 be a projective @xmath44-dimensional manifold and let @xmath2 be a smooth ample divisor .
here is a list of general facts concerning ascent properties from @xmath3 to @xmath1 .
e.g. , @xmath596 @xmath342 not ample @xmath374 @xmath34 not nef
. it immediately follows from the adjunction formula .
@xmath596 @xmath1136 @xmath374 @xmath1137 . here is the argument from ( * ? ? ?
* proposition 5 ) .
assume by contradiction that @xmath1138 for some @xmath1139 and let @xmath1140 .
write @xmath1141 , with @xmath1142 and @xmath1143 . since @xmath1144 is very ample for @xmath1145
, we can find @xmath1145 and @xmath1146 such that @xmath1147 .
since we have @xmath1148 , it follows that @xmath1149 is very ample outside of @xmath1150 , so that @xmath1151 .
this contradiction proves the assertion .
uniruled manifolds are birationally fano fibrations .
this fact follows from campana s construction ( see e.g. , the preface and chapters @xmath100 , @xmath1152 of debarre s text @xcite ) .
many results in our paper are concerned with the case in which @xmath3 carries a special fano fibration structure .
@xmath596 @xmath3 uniruled @xmath374 @xmath1 uniruled .
it immediately follows from the uniruledness criterion ( * ? ? ?
* ii , section 3 , iv , ( 1.9 ) ) . saying that @xmath3 is uniruled means that there is a morphism @xmath1153 such that @xmath1154 is spanned .
consider the tangent bundle sequence @xmath428 let @xmath1155 be the induced morphism to @xmath1 . by pulling back to @xmath10
, we get the exact sequence @xmath1156 since both @xmath1157 and @xmath1154 are nef , we conclude that @xmath1158 is nef and hence spanned ; this is equivalent to say that @xmath1 is uniruled .
@xmath596 @xmath3 rationally connected @xmath374 @xmath1 rationally connected . saying that @xmath3 is rationally connected is equivalent to the existence of a curve @xmath1159 with ample normal bundle @xmath1160 ( see e.g. , @xcite ) .
therefore the exact sequence of normal bundles @xmath1161 and the ampleness of @xmath3 give the ampleness of @xmath1162 .
@xmath596 @xmath3 unirational @xmath374 @xmath1 unirational ?
this is a hard question and no answer is known .
it is interesting to point out that , since unirationality implies rational connectedness , to find examples of @xmath3 unirational with @xmath1 not unirational would give examples of rationally connected manifolds @xmath1 which are not unirational .
quoting kollr ( * ? ? ?
* section 7 , problem 55 ) , the latter is `` one of the most vexing open problems '' in the theory .
@xmath596 in general , @xmath3 rational does not imply that @xmath1 is rational .
we present below a few results about this problem .
in particular , we obtain a proof of the following classical statement ( ( * ? ? ? * chapter iv ) ) : _ for a very ample smooth divisor on a three fold @xmath1 , the ascent of rationality holds true with the only exception when @xmath1 is the cubic hypersurface of @xmath1163_. the case of the cubic hypersurface is indeed an exception , see @xcite .
[ rat ] ( cf .
also @xcite ) let @xmath21 be an ample line bundle on a smooth projective three fold @xmath1 .
assume that there is a smooth @xmath732 such that @xmath3 is rational .
then @xmath1 is rational unless either : 1 .
@xmath1164 and @xmath90 is a weighted hypersurface of degree @xmath1165 in the weighted projective space @xmath1166 , @xmath1167 ; or 2 .
@xmath1168 and @xmath90 is the double covering of @xmath1169 branched along a smooth surface of degree @xmath1152 , @xmath1167 and @xmath21 is the pull - back of @xmath1170 ; or 3 .
@xmath1 is the hypercubic in @xmath1163 and @xmath1171 .
since @xmath3 is a rational surface , @xmath342 is not nef .
we follow the cases ( i)(v ) from theorem [ 6.2 ] . in case ( i ) , we apply the well - known classification of fano three folds of index @xmath68 ( and @xmath1172 ) , see @xcite .
we either get one of the exceptional cases in the statement , or @xmath1 is the complete intersection of two quadrics in @xmath1173 , or @xmath1174 is a linear section of the grassmannian of lines in @xmath1163 , embedded in @xmath1175 by the plcker embedding . in the last two cases @xmath1 is rational ( see e.g. , @xcite ) .
a simple argument is given in example [ sad ] below .
assume now that we are in case ( ii ) or ( iii ) from theorem [ 6.2 ] .
for such a fibration the base curve is @xmath10 and the general fiber is rational .
moreover , a section exists by tsen s theorem , see e.g. , ( * ? ? ?
so @xmath1 is rational , too .
if we are in case ( iv ) , the base surface is birational to @xmath3 , so it is rational . we conclude that @xmath1 is rational .
finally , case ( v ) leads to one of the previously discussed situations .
q.e.d . plus
1pt the following result , contained in ( * ? ? ?
* theorem 1.3 ) , concerns the ascent of rationality from a suitable rational submanifold .
the proof relies on hironaka s desingularization theory @xcite and on basic properties of rationally connected manifolds @xcite .
[ trei - sase ] ( ( * ? ? ?
* theorem 1.3 ) ) let @xmath1 be a projective variety and @xmath53 a complete linear system of cartier divisors on it .
let @xmath1176 and put @xmath1177 for @xmath1178 .
assume that @xmath1179 is smooth , irreducible of dimension @xmath1180 , for all @xmath20 .
assume moreover that there is a divisor @xmath189 on @xmath1181 and a linear system @xmath1182 such that : 1 .
@xmath1183 is birational , and 2 .
@xmath1184 .
then @xmath1 is rational .
we proceed by induction on @xmath1185 .
let us explain the case @xmath629 , the general case being completely similar .
so , let @xmath1186 be a smooth , irreducible cartier divisor such that @xmath1187 is birational for @xmath1182 , @xmath1188 and @xmath1189 . replacing @xmath1 by its desingularization
, we may assume that @xmath1 is smooth .
as @xmath56 is rational , it is rationally connected , so we may find some smooth rational curve @xmath1190 with @xmath1191 ample .
we have @xmath1192 and from ( ii ) we deduce @xmath1193 . from the exact sequence of normal bundles we get that @xmath1194 is ample ,
so @xmath1 is rationally connected .
in particular , @xmath1195 .
the exact sequence @xmath1196 shows that @xmath1197 .
we may choose a pencil @xmath1198 , containing @xmath56 , such that @xmath1199 , with @xmath1200 and @xmath1201 .
by hironaka s theory @xcite , we may use blowing - ups with smooth centers contained in @xmath1202 , such that after taking the proper transforms of the elements of our pencil , to get : 1 . @xmath1203 has normal crossing ; 2 .
@xmath1204 is base points free ( so @xmath1205 is a birational morphism ) .
further blowing - up of the components of @xmath1206 allows to assume @xmath1207 so @xmath1208 is linearly equivalent to @xmath189 . using the previous exact sequence and the fact that @xmath1209 , it follows that @xmath1210 .
finally , @xmath1211 , so @xmath185 is a birational morphism to @xmath510 .
q.e.d . plus
1pt [ sad ] ( ( * ? ? ?
* example 1.4 ) ) let @xmath1212 be a non - degenerate projective variety of dimension @xmath1213 and degree @xmath1214 , which is not a cone .
then @xmath1 is rational , unless it is a smooth cubic hypersurface , @xmath200 .
if @xmath1 is singular , by projecting from a singular point we get a variety of minimal degree , birational to @xmath1 .
so @xmath1 is rational . if @xmath1 is not linearly normal , @xmath1 is isomorphic to a variety of minimal degree .
hence we may assume @xmath1 to be smooth and linearly normal .
one sees easily that such a linearly normal , non - degenerate manifold @xmath1215 has anticanonical divisor linearly equivalent to @xmath1216 times the hyperplane section , i.e. , they are exactly the so - called `` classical del pezzo manifolds '' .
they were classified by fujita in a series of papers ; see @xcite or @xcite . independently of their classification , the following simple argument shows that such manifolds are rational when @xmath1217 .
consider the surface @xmath56 obtained by intersecting @xmath1 with @xmath1218 general hyperplanes .
note that @xmath56 is a non - degenerate , linearly normal surface of degree @xmath648 in @xmath9 , so it is a del pezzo surface .
as such , @xmath56 is known to admit a representation @xmath1219 as the blowing - up of @xmath1220 points .
let @xmath1221 be the pull - back via @xmath185 of a general line in @xmath854 .
it is easy to see that @xmath21 is a cubic rational curve in the embedding of @xmath56 into @xmath9 .
so , for @xmath1222 , @xmath21 is contained in a hyperplane of @xmath9 .
this shows that the conditions of the theorem [ trei - sase ] are fulfilled for @xmath1 , @xmath53 being the system of hyperplane sections .
we also see that theorem [ trei - sase ] is sharp , as the previous argument fails exactly for the case of cubics .
[ closing ] in closing , we mention three possible generalizations of the problem of extending morphisms from ample divisors on @xmath1 . 1 .
the smoothness assumption on @xmath1 may be relaxed by allowing normal singularities .
let @xmath3 be a smooth divisor in @xmath1 ( @xmath1 is smooth ) , and let us only suppose that @xmath3 has ample normal bundle . then
a well - known result ( @xcite ) shows that there is a birational map @xmath1223 , which is an isomorphism along @xmath3 , such that @xmath1224 is ample and @xmath340 is normal .
see e.g. , @xcite and @xcite for results in this direction .
2 . consider a smooth section @xmath2 of the appropriate expected dimension @xmath1225 of an ample vector bundle @xmath46 on an @xmath44-fold @xmath1 .
note that a lefschetz type theorem for ample vector bundles , due to sommese @xcite , implies that the restriction to @xmath3 gives an isomorphism @xmath1226 .
see e.g. , @xcite and @xcite for results of this type .
3 . in the same spirit
, let us consider a smooth subvariety @xmath3 of a manifold @xmath1 such that @xmath1227 , and @xmath3 has ample normal bundle .
further , let us add the lefschetz type assumption that @xmath1228
. then one can study extensions of rationally connected fibrations @xmath81 onto a normal projective variety @xmath7 .
see @xcite and @xcite for results in this direction .
the very recent paper @xcite classifies pairs @xmath838 , when @xmath2 is an ample divisor which is a homogeneous manifold .
we thank the referee for several useful comments .
l. bdescu , _ hyperplane sections and deformations _ , proceedings of the week of algebraic geometry , bucharest , 1982 , ed . by l. bdescu and d. popescu ,
lecture notes in math .
, vol . 1056 , springer - verlag , new york , 1984 , pp .
l. bdescu , _ infinitesimal deformations of negative weights and hyperplane sections _ , algebraic geometry , proceedings of conference on hyperplane sections , laquila , italy , 1988 , ed .
sommese , a. biancofiore , and e.l .
livorni , lecture notes in math .
, vol . 1417 , springer - verlag , new york , 1990 , pp .
122 .
beltrametti and m.l .
fania , _ fano threefolds as hyperplane sections _ , projective varieties with unexpected properties a volume in memory of giuseppe veronese , proceedings of the international conference `` varieties with unexpected properties '' siena , italy , june 813 , 2004 , ed .
by c. ciliberto et al . ,
w. de gruyter , 2005 , pp .
beltrametti , m.l .
fania , and a.j .
sommese , _ mukai varieties as hyperplane sections _ , proceedings of the fano conference , torino , italy , 2002 , ed . by a. collino , a. conte and m. marchisio , universit degli studi di torino , dipartimento di matematica , torino , 2004 , pp .
185208 .
beltrametti and a.j .
sommese , _ comparing the classical and the adjunction theoretic definition of scrolls _ , geometry of complex projective varieties , cetraro , italy 1990 , ed . by a. lanteri , m. palleschi , and d. struppa , seminars and conferences , vol . 9 , mediterranean press , 1993 , pp . 5574 . m.c .
beltrametti , a.j .
sommese , and j.a .
winiewski , _ results on varieties with many lines and their applications to adjunction theory _
( with an appendix by m.c .
beltrametti and a.j .
sommese ) , complex algebraic varieties , bayreuth 1990 , ed . by k. hulek et al . ,
lecture notes in math .
, vol . 1507 , springer - verlag , new york , 1992 , pp .
1638 .
j. kollr , _ appendix to the paper of c. borcea : homogeneous vector bundles and families of calabi
yau threefolds_. ii , several complex variables and complex geometry , 1989 , ed .
by e. bedford et al . , proc .
pure math .
52 , part 2 , 1991 , 8391 .
a. lanteri , m. palleschi , and a.j .
sommese , _ on triple covers of @xmath510 as very ample divisors _ , classification of algebraic varieties , proceedings laquila , 1992 , ed .
by c. ciliberto , e.l .
livorni , and a.j .
sommese , contemp . math . , vol . 162 , 1994 , pp .
277292 .
s. mori , _ threefolds whose canonical bundles are not numerically effective _ , algebraic threefolds , proceedings varenna , 1981 , ed . by a. conte , lecture notes in math . , vol .
947 , springer - verlag , new york , 1982 , pp .
125189 .
e. sato , _ a variety which contains a @xmath10-fiber space as an ample divisor _
, algebraic geometry and commutative algebra , in honor of masayoshi nagata , vol .
ii , ed . by h. hijikata et al . , kinokuniya , tokyo , 1988 , pp .
665691 .
e. sato and h. spindler , _ on the structure of @xmath1152-folds with a hyperplane section which is a @xmath10 bundle over a ruled surface _ , proceedings of gttingen conference , 1985 , ed .
by h. grauert , lecture notes in math .
, vol . 1194 , springer - verlag , new york , 1986 , pp . | the philosophy that a projective manifold is more special than any of its smooth hyperplane sections " was one of the classical principles of projective geometry .
lefschetz type results and related vanishing theorems were among the typically used techniques .
we shall survey most of the problems , results and conjectures in this area , using the modern setting of ample divisors , and ( some aspects of ) mori theory . |
a unital embedding of a matrix algebra @xmath3 into a unital @xmath0-algebra @xmath4 can exist only if the equation @xmath5 $ ] has a solution @xmath6 .
thus , only @xmath0-algebras in which the class of the unit in @xmath7 is @xmath8-divisible admit a unital embedding of @xmath3 . whereas all von neumann algebras ( with no central summand of type i@xmath9 for @xmath10 finite ) have this divisibility property for all @xmath8 , the same is not true for @xmath0-algebras , even for the simple ones .
@xmath0-algebras can fail to have non - trivial projections .
even if they have many projections , as in the real rank zero case , one can not expect to solve the equation @xmath5 $ ] exactly in @xmath11 .
this paper is concerned with different weaker notions of divisibility , phrased in terms of the cuntz semigroup of the @xmath0-algebra , and with how they relate to embeddability properties of the @xmath0-algebra . instead of solving the equation
@xmath5 $ ] for @xmath6 , one should look for less restrictive notions of divisibility .
one can try , for example , to solve the inequalities @xmath12 in the cuntz semigroup of @xmath4 for fixed positive integers @xmath8 and @xmath10 ( typically with @xmath13 ) .
we say that @xmath4 is @xmath14-divisible if one can solve this inequality .
this is one of three divisibility properties we shall consider in this paper .
we show that there is a full @xmath15-homomorphism from @xmath16 , the cone over @xmath3 , into @xmath4 if and only if @xmath4 is @xmath14-divisible for some @xmath10 .
let us mention three embedding problems that served as motivation for this paper .
let @xmath4 be a unital @xmath0-algebra with no non - zero finite dimensional representations .
can one always find an embedding of some unital simple infinite dimensional @xmath0-algebra into @xmath4 ?
can one always find an embedding of @xmath17 into @xmath4 whose image is full in @xmath4 ?
can one always find two positive mutually orthogonal full elements in @xmath4 ?
an affirmative answer to the former problem will imply an affirmative anwer to the second problem , which is known as the global glimm halving problem " .
an affirmative answer to the global glimm halving problem will imply an affirmative answer to the last mentioned problem .
we suspect that all three problems may have negative answers .
the second and the third problem led us to consider two new notions of divisibility properties . in more detail , we say that @xmath4 is weakly @xmath14-divisible if there are elements @xmath18 in @xmath19 such that @xmath20 . weak divisibility measures the rank of @xmath4 in the sense that @xmath4 is weakly @xmath14-divisible for some @xmath10 if and only if @xmath4 has no non - zero representations of dimension @xmath21 .
in particular , @xmath4 has no non - zero finite dimensional representations if and only if for every @xmath8 there is @xmath10 such that @xmath4 is weakly @xmath14-divisible .
we say that @xmath4 is @xmath14-decomposable if there are elements @xmath22 in @xmath19 such that @xmath23 .
for a given @xmath8 , @xmath4 is @xmath14-decomposable for some @xmath10 if and only if @xmath4 contains @xmath8 pairwise orthogonal , pairwise equivalent full positive elements .
it was shown in @xcite that there exists a simple unital infinite dimensional @xmath0-algebra which does not admit a unital embedding of the jiang - su algebra @xmath24 .
this answered in the negative a question posed by the second named author .
it is implicit in @xcite that this simple @xmath0-algebra has bad divisibility properties , cf .
remark [ rem : j - s ] .
this leads us to a useful observation , which loosely can be formulated as follows : if @xmath4 and @xmath25 are unital @xmath0-algebras , and if there is a unital @xmath15-homomorphism from @xmath4 to @xmath25 , then the divisibility properties of @xmath25 are no worse than those of @xmath4 .
in other words , if @xmath4 has better divisibility properties than @xmath25 , then you can not unitally embed @xmath4 into @xmath25 .
comparability in the cuntz semigroup is concerned with the extent to which one can conclude that @xmath26 if the size " of @xmath27 ( e.g. , measured in terms of states ) is ( much ) smaller than the size " of @xmath28 .
comparability and divisibility are probably the two most fundamental properties of the cuntz semigroup .
good comparability and divisibility properties are necessary and sufficient conditions in winter s theorem , @xcite , to conclude that a simple , separable , unital @xmath0-algebra with locally finite nuclear dimension tensorially absorbs the jiang - su algebra . also , good comparability and divisibility properties are both necessary and sufficient conditions to ensure that the cuntz semigroup of a simple , separable , unital , exact @xmath0-algebra @xmath4 is ( naturally ) isomorphic to @xmath29 , cf .
@xcite , @xcite , and @xcite .
the existence of simple @xmath0-algebras with bad comparability properties was discovered by villadsen , @xcite , in the mid 1990 s .
this discovery was the first indication that the elliott conjecture could be false ( in general ) , and it was also the first example of a simple @xmath0-algebra exhibiting infinite dimensional " behaviour .
villadsen s example in @xcite has been generalized extensively by several authors ( including villadsen himself ) to exhibit simple @xmath0-algebras with various kinds of unexpected behaviour , including many ways of failing to have good comparability properties .
however , little work has been done to construct simple @xmath0-algebras with bad divisibility behaviour , and the literature does not contain systematic ways of producing such examples . in this paper
we show that there is a duality between comparability and divisibility ( lemma [ lm : z ] ) , and we use this duality to construct examples of simple and non - simple @xmath0-algebras with bad divisibility behaviour .
we use lemma [ lm : z ] to obtain a result that concerns the structure of @xmath0-algebras that arise as the tensor product of a sequence of unital ( simple non - elementary ) @xmath0-algebras .
each such @xmath0-algebra will of course have non - trivial central sequences .
dadarlat and toms proved in @xcite that if the infinite tensor power @xmath30 of a fixed unital @xmath0-algebra @xmath4 contains a unital copy of an ahs - algebra without characters , then it automatically absorbs the jiang - su algebra .
it is not known if this condition always is satisfied , even when @xmath4 is simple and non - elementary .
we show in section [ sec : divcomp ] that @xmath30 has the corona factorization property for every unital @xmath4 without characters ( and in particular for every unital simple @xmath0-algebra @xmath31 ) . in the other direction
we give , in section [ sec : obstruct ] , an example of a sequence of simple unital infinite dimensional @xmath0-algebras whose tensor power , @xmath32 , does not absorb ( or admit an embedding of ) the jiang - su algebra .
non - divisibility of a @xmath0-algebra can be interpreted as a degree of inhomogeneity ( or lumpiness " ) of the @xmath0-algebra .
simple @xmath0-algebras are sometimes thought of as being very homogeneous , as for example in @xcite . from this point of view it may at first be surprising that a simple infinite dimensional @xmath0-algebra can fail to have good divisibility properties .
we show that there exists a sequence @xmath1 of simple , unital , infinite dimensional @xmath0-algebras such that @xmath2 ( and also the associated ultrapowers of @xmath1 ) has a character .
none of the @xmath0-algebras @xmath33 can have a character ( being simple and not equal to @xmath34 ) , however we can show that they posses almost characters " as defined in section [ sec : ultrapowers ] .
in section [ sec : infinite ] we consider what one might call super - divisibility " , which leads to a ( new ) notion of infiniteness of positive elements ( and which implies that a multiple of the given element is properly infinite ) .
we use this to reformulate the corona factorization property of semigroups considered in @xcite .
we study variations of examples , originally due to dixmier and douady , and answer in this way two questions from @xcite in the negative : the sum of two properly infinite positive elements need not be properly infinite , and the multiplier algebra of a @xmath0-algebra which has a properly infinite strictly positive element need not be properly infinite .
let @xmath4 be a @xmath0-algebra and let @xmath19 denote the cuntz semigroup of @xmath4 , i.e. , the set of cuntz equivalence classes of positive elements in @xmath35 endowed with suitable order and addition structures .
given a positive element @xmath36 in @xmath35 , we shall denote by @xmath37 the cuntz class of @xmath36 . in @xcite , coward , elliott and ivanescu
give an alternative picture of the cuntz semigroup where @xmath19 consists of suitable equivalence classes of countably generated hilbert @xmath0-modules over @xmath4 .
when using the hilbert modules picture of @xmath19 , we shall denote the equivalence class of a countably generated hilbert module @xmath38 by @xmath39 $ ] .
we present here some well - known definitions and facts about the cuntz semigroup .
first of all , we shall frequently use the axioms of the category @xmath40 , of which @xmath19 is always an object ( see @xcite ) .
an ordered abelian semigroup @xmath41 is an object in the category @xmath40 if 1 . every increasing sequence in @xmath41 has a supremum , 2 . for every @xmath42 there exists a sequence @xmath43 in @xmath41 such that @xmath44 and @xmath45 , 3 .
if @xmath46 and @xmath47 , then @xmath48 , 4 . if @xmath43 and @xmath49 are increasing sequences then @xmath50 .
recall that @xmath51 in @xmath41 if whenever @xmath52 for some increasing sequence @xmath49 in @xmath41 , then @xmath53 for some @xmath54 .
an element @xmath55 is called _ compact _ if @xmath56 .
we also note the following two additional properties of the cuntz semigroup of a @xmath0-algebrawhich are note listed among the axioms of @xmath40 .
the first of them asserts that the cuntz semigroup of a @xmath0-algebra almost has the riesz decomposition property , and the second states that its order relation is almost the algebraic order . *
if @xmath57 , then there exist @xmath58 and @xmath59 , with @xmath60 and @xmath61 , and such that @xmath62 .
* if @xmath63 , then there exists @xmath64 such that @xmath65 . for the proofs of these facts , see ( * ? ? ?
* proposition 5.1.1 ) for the first and ( * ? ? ?
* lemma 7.1 ( i ) ) for the second .
we will also make use of the sequential continuity with respect to inductive limits of the functor @xmath66 proved in @xcite ( see also the proof of ( * ? ? ? * theorem 4.8 ) ) .
it can be stated as follows : [ prop : culimits ] let @xmath67 be a sequential inductive limit of @xmath0-algebras . 1 . for every @xmath68 there exists an increasing sequence @xmath69 with supremum @xmath70 and such that each @xmath71 belongs to @xmath72 .
2 . if @xmath73 are such that @xmath74 , then for every @xmath46 there exists @xmath75 such that @xmath76 .
[ rem : cm ] let @xmath8 be a positive integer , and let @xmath16 denote the cone over @xmath3 , i.e. , the @xmath0-algebra of all continuous functions @xmath77 \to m_m({{\mathbb c}})$ ] that vanish at @xmath78
. for each @xmath79 , let @xmath80 denote the @xmath81th matrix unit in @xmath3 , and denote by @xmath82 the function @xmath83 in @xmath16
. then @xmath84 are positive contractions in @xmath16 which are pairwise equivalent and orthogonal .
( we say here that two positive elements @xmath36 and @xmath85 are equivalent , denoted @xmath86 , if @xmath87 and @xmath88 for some element @xmath27 in the ambient @xmath0-algebra . )
we recall the following well - known universal property of @xmath16 ( see for example ( * ? ? ?
* propositions 2.3 and 2.4 ) ) : let @xmath4 be any @xmath0-algebra and let @xmath89 be positive contractions in @xmath4 .
then there exists a @xmath15-homomorphism @xmath90 satisfying @xmath91 if and only if @xmath89 are pairwise orthogonal and pairwise equivalent in @xmath4 .
the following lemma is well - known : [ lm : comp ] let @xmath4 be a @xmath0-algebra and let @xmath92 be positive elements in @xmath4 .
then : 1 .
@xmath93 if and only if for each @xmath94 there exist @xmath95 such that @xmath96 .
2 . @xmath97 if and only if for each @xmath98 there exist mutually orthogonal positive elements @xmath99 in @xmath100 such that @xmath101 for all @xmath54 .
( i ) . if @xmath93 , then @xmath102 , whence @xmath103 for some @xmath104 .
the converse statement is trivial .
suppose that @xmath97 .
then @xmath105 for some @xmath106 in @xmath107 .
thus @xmath108 if @xmath109 and @xmath110 for all @xmath54 . put @xmath111
. it is now straightforward to verify that the @xmath112 s are as desired .
the converse statement is trivial .
here is another lemma that we will use frequently : [ lm : cone - comp ] let @xmath36 and @xmath85 be positive elements in @xmath35 with @xmath113 , and let @xmath114 .
the following are equivalent : 1 .
@xmath115 , 2 . for each @xmath94 there exist mutually orthogonal positive elements @xmath116 in @xmath117 such that @xmath118 for all @xmath54 .
3 . for each @xmath94 there exists a @xmath15-homomorphism @xmath119 such that @xmath120 .
the implications ( iii ) @xmath121 ( ii ) @xmath121 ( i ) are clear , cf .
remark [ rem : cm ] and lemma [ lm : comp ] .
let us show that ( i ) implies ( iii ) .
let @xmath98 be given . by lemma [ lm : comp ] ( ii )
there are mutually orthogonal positive elements @xmath122 in @xmath117 such that each @xmath123 is equivalent to @xmath124 . by the universal property of the cone @xmath16 ,
see remark [ rem : cm ] , there is a @xmath15-homomorphism @xmath125 satisfying @xmath126 .
hence ( iii ) holds .
[ def : divisibility ] let @xmath4 be a @xmath0-algebra and fix @xmath68 .
let @xmath127 be integers .
then : 1 . @xmath70 is _
@xmath14-divisible _ if for every @xmath128 with @xmath129 there exists @xmath130 such that @xmath131 and @xmath132 . + the least @xmath10 such that @xmath70 is @xmath14-divisible is denoted by @xmath133 , with @xmath134 if no such @xmath10 exists .
2 . @xmath70 is _
@xmath14-decomposable _ if for every @xmath128 with @xmath129 there exist elements @xmath135 such that @xmath136 and @xmath137 for all @xmath138 . + the least @xmath10 such that @xmath70 is @xmath14-decomposable is denoted by @xmath139 , with @xmath140 if no such @xmath10 exists .
3 . @xmath70 is _ weakly @xmath14-divisible _ if for every @xmath128 with @xmath129 there exist elements @xmath141 such that @xmath142 for all @xmath138 and @xmath143 .
+ the least @xmath10 such that @xmath70 is weakly @xmath14-divisible is denoted by @xmath144 , with @xmath145 if no such @xmath10 exists .
[ rem : divisible - unital ] in the case that @xmath70 in definition [ def : divisibility ] is compact ( e.g. , when @xmath4 is unital and @xmath146 ) , the conditions above read a little easier : 1 .
@xmath70 is _
@xmath14-divisible _ if there exists @xmath130 such that @xmath147 .
2 . @xmath70 is _
@xmath14-decomposable _ if there exist elements @xmath135 such that @xmath148 for all @xmath138 .
3 . @xmath70 is _
weakly @xmath14-divisible _ if there exist elements @xmath141 such that @xmath149 .
the three divisibility properties above are related as follows : [ prop : a ] let @xmath150 and @xmath68
. then @xmath151 the two first inequalities are clear ( take @xmath152 in both cases ) . to prove the last inequality , suppose that @xmath70 is @xmath14-decomposable .
we show that @xmath70 is weakly @xmath153-divisible .
let @xmath46 and find @xmath154 such that @xmath155 .
there exist elements @xmath156 in @xmath19 such that @xmath157 and @xmath158 for all @xmath54 .
we proceed to find elements @xmath159 satisfying * @xmath160 , * @xmath161 if @xmath162 , and @xmath163 , * @xmath164 if @xmath162 , and @xmath165 , * @xmath166 , * @xmath167 if @xmath168 , and @xmath169 .
the elements above are constructed inductively after @xmath170 using the following fact : * if @xmath171 in @xmath19 , then there exist @xmath172 such that @xmath173 , @xmath174 , and @xmath175 , which follows from property ( p1 ) of the cuntz semigroup stated in the previous section .
take first @xmath176 . the existence of @xmath177 , with @xmath178 , satisfying ( b ) , ( d ) and ( e ) follows from ( @xmath179 ) applied to @xmath180 . the existence of @xmath181 satisfying ( a ) and ( c ) then follows from axiom ( a2 ) of the cuntz semigroup from the previous section .
assume that @xmath182 and that @xmath183 and @xmath184 have been found .
the existence of @xmath185 , with @xmath178 , satisfying ( b ) , ( d ) and ( e ) follows from ( @xmath179 ) applied to @xmath186 ( to see that the latter inequality holds , note that @xmath187 , which follows by repeated use of ( e ) . ) the existence of @xmath188 satisfying ( a ) and ( c ) follows from axiom ( a2 ) .
we claim that the @xmath189 elements @xmath190 witness the weak @xmath153-divisibility of @xmath70 .
indeed , it follows from ( d ) and ( e ) that @xmath191 for all @xmath192 , whence @xmath193 it follows from ( b ) and ( c ) that the sum of the elements @xmath194 is greater than or equal to @xmath195 .
if any of the divisibility numbers @xmath133 , @xmath139 , and @xmath196 is less than @xmath8 , then @xmath70 ( or a multiple of @xmath70 ) must be properly infinite , as shown below .
we shall pursue this and related questions in more detail in section [ sec : infinite ] .
[ prop : propinf ] let @xmath4 be a @xmath0-algebra and let @xmath68 . 1
. if @xmath70 is properly infinite , then @xmath197 for all integers @xmath198 .
if @xmath199 are integers and if @xmath70 is either @xmath14-divisible , @xmath14-decomposable or weakly @xmath14-divisible , then @xmath200 is properly infinite , i.e. , @xmath201 .
3 . if @xmath199 are integers and if @xmath70 is compact and @xmath14-divisible , then @xmath70 is properly infinite .
( i ) . if @xmath70 is properly infinite , then @xmath202 for all @xmath8 , whence @xmath197 .
assume that @xmath70 is weakly @xmath14-divisible and take @xmath46 .
then there exist @xmath18 such that @xmath203 for all @xmath54 , and @xmath204 .
thus , @xmath205 as this holds for all @xmath46 , we get @xmath206 .
this entails that @xmath207 for all positive integers @xmath170 , whence @xmath208 for all @xmath209 .
in particular , @xmath210 , which implies that @xmath200 is properly infinite .
next , suppose that @xmath70 is @xmath14-decomposable and let @xmath46 .
then there exist @xmath156 such that @xmath157 and @xmath211 for all @xmath54 .
thus , @xmath212 arguing as before , we conclude that @xmath200 is properly infinite .
finally note that if @xmath70 is @xmath14-divisible , then it is both @xmath14-decomposable and weakly @xmath14-divisible , whence @xmath200 is properly infinite .
( iii ) . since @xmath213 and @xmath214
, there exists @xmath27 such that @xmath147 .
arguing as above this implies that @xmath215 is properly infinite .
[ rem : image ] by functoriality , each @xmath15-homomorphism @xmath216 between @xmath0-algebras @xmath4 and @xmath25 induces a morphism @xmath217 which preserves order , addition , and the relation of compact containment .
thus , for each @xmath68 , and with @xmath218 , we have : @xmath219 in particular , if @xmath4 and @xmath25 are unital @xmath0-algebras , and if @xmath220 for some @xmath8 ( or if the corresponding inequality holds for one of the other two divisibility numbers ) , then one can not find a unital embedding of @xmath4 into @xmath25 .
divisibility numbers thus serve as an obstruction for embedding a unital @xmath0-algebra with nice divisibility properties into a unital @xmath0-algebra with less nice divisibility properties .
the three divisibility properties behave well with respect to inductive limits thanks to the sequential continuity of the functor @xmath66 : [ prop : inductive - limits ] let @xmath67 be a sequential inductive limit of @xmath0-algebra .
let @xmath221 and , for each @xmath54 , denote by @xmath222 and @xmath223 the images of @xmath70 in @xmath224 and @xmath19 , respectively . then : @xmath225 @xmath226 if @xmath70 is compact ( i.e. , if @xmath214 ) , then the above inequalities are equalities .
we will only prove the statements above in the former case ; the proofs for the two other cases are similar .
the inequalities @xmath227 , with @xmath228 , follow from remark [ rem : image ] .
suppose now that @xmath70 is compact .
set @xmath229 .
then there exists @xmath230 such that @xmath231 . by proposition [ prop : culimits ]
( i ) and compactness of @xmath232 , it follows that @xmath27 is the image of some @xmath233 for some @xmath54 . by axiom ( a2 ) of the cuntz semigroup and by compactness of @xmath232 there exists @xmath234 in @xmath224 such that @xmath235 . since the @xmath71 s are compact , proposition [ prop : culimits ] ( ii ) implies that there exists @xmath236 such that @xmath237 thus @xmath238 is @xmath14-divisible in @xmath239 .
[ def : diva ] let @xmath4 be a @xmath240-unital @xmath0-algebra .
then @xmath4 contains a strictly positive element .
this element represents a class in @xmath19 , which is independent of the choice of the strictly positive element , and which we shall denote by @xmath241 .
if @xmath4 is unital , then @xmath242 . we shall write @xmath196 , @xmath243 , and @xmath244 for @xmath245 , @xmath246 , and @xmath247 , respectively . if @xmath4 and @xmath25 are unital @xmath0-algebras such that there exist unital @xmath15-homomorphisms @xmath248 and @xmath249 , then , by remark [ rem : image ] , we must have @xmath250 for all @xmath8 .
this applies in particular to the situation where @xmath4 is any unital @xmath0-algebra and @xmath251 for some unital @xmath0-algebra @xmath252 which has a character . in general ,
if @xmath252 is any unital @xmath0-algebra , possibly without characters , the divisibility numbers associated with @xmath253 are smaller than or equal to those of @xmath4 .
let us first examine the divisibility numbers for matrix algebras : [ rem : matrix ] let @xmath254 and @xmath162 be integers .
using that @xmath255 an elementary algebraic argument yields that @xmath256 and @xmath257 here @xmath258 and @xmath259 are the ceiling " and floor " functions .
in particular , @xmath260 if and only if @xmath261 , and @xmath262 if @xmath263 and @xmath264 .
[ def : rank ] let @xmath4 be a @xmath0-algebra .
let @xmath265 denote the smallest positive integer @xmath10 for which @xmath4 has an irreducible representation on a hilbert space of dimension @xmath10 , and set @xmath266 if @xmath4 has no finite dimensional ( irreducible ) representation .
note that @xmath267 if and only if @xmath4 has a character .
we remind the reader about the following classical result due to glimm : [ prop : glimm ] let @xmath4 be a ( not necessarily unital ) @xmath0-algebra . then there is a non - zero @xmath15-homomorphism @xmath268 if and only if @xmath4 admits at least one irreducible representation on a hilbert of dimension @xmath269 .
it follows from remark [ rem : image ] that @xmath196 , @xmath243 , and @xmath244 are greater than or equal to @xmath270 if @xmath271 . in particular , these three quantities are infinite when @xmath272 .
[ ex : simple ] if @xmath4 is a simple , unital , infinite dimensional @xmath0-algebra , then @xmath244 , @xmath243 , and @xmath196 are finite for all positive integers @xmath8 . indeed , by the assumption that @xmath4 is infinite dimensional
, it follows that there is a non - zero @xmath130 such that @xmath273 .
as every simple unital @xmath0-algebra is algebraically simple , it follows that @xmath274 for some positive integer @xmath10 , i.e. , @xmath275 is @xmath14-divisible .
hence @xmath276 , which entails that also @xmath277 and @xmath278 .
the dimension drop @xmath0-algebra @xmath279 , associated with the positive integers @xmath280 and @xmath281 , is defined to be @xmath282,m_p \otimes m_q \mid f(0 ) \in m_p \otimes { { \mathbb c}}1_q , \ ; \ ; f(1 ) \in { { \mathbb c}}1_p \otimes m_q \}.\ ] ] note that @xmath283 .
it was shown in ( * ? ? ?
* lemma 4.2 ) ( and its proof ) that @xmath284 . by remark
[ rem : image ] , it follows that if @xmath285 maps unitally into @xmath4 , then @xmath286 .
moreover , as shown in ( * ? ? ?
* proposition 5.1 ) , if @xmath4 is a unital @xmath0-algebra of stable rank one , then @xmath286 if and only if @xmath285 maps unitally into @xmath4 .
the property almost divisibility " of a @xmath0-algebra is expressed by saying that @xmath286 for all integers @xmath198 . if every dimension drop algebra @xmath285 maps unitally into @xmath4 , or if the jiang - su algebra maps unitally into @xmath4 , then @xmath4 is almost divisible .
[ rem : j - s ] it was shown in @xcite that there is a simple unital infinite dimensional nuclear @xmath0-algebra @xmath4 such that the dimension drop @xmath0-algebra @xmath287 , and hence the jiang - su algebra @xmath24 , do not embed unitally into @xmath4 .
the divisibility properties of @xmath4 were not explicitly mentioned in @xcite , but it is easily seen ( using lemma [ lm : z ] , that is paraphrased from ( * ? ? ?
* lemma 4.3 ) ) that @xmath288 .
we shall in section [ sec : obstruct ] give further examples of simple unital infinite dimensional @xmath0-algebras where the divisibility numbers attain non - trivial values .
[ ex : rr0 ] it was shown in ( * ? ? ? * proposition 5.7 ) that if @xmath4 is a unital @xmath0-algebra of real rank zero then @xmath4 admits a unital embedding of a finite dimensional @xmath0-algebra of rank at least @xmath10 if and only if @xmath289 . combining this with remark [ rem : matrix ]
we see that @xmath286 whenever @xmath4 is a unital @xmath0-algebra of real rank zero and with @xmath290 .
in particular , every unital @xmath0-algebra @xmath4 of real rank zero and with @xmath266 is almost divisible .
kirchberg considered in @xcite a covering number of a unital @xmath0-algebra @xmath25 .
let us recall the definition : [ def : cov ] let @xmath114 .
the covering number of a unital @xmath0-algebra @xmath25 , denoted by @xmath291 , is the least positive integer @xmath10 such that there exist finite dimensional @xmath0-algebras @xmath292 with @xmath293 , @xmath15-homomorphisms @xmath294 , and @xmath295 such that @xmath296 .
kirchberg s covering number @xmath291 relates to our @xmath297 as follows .
[ prop : cov ] let @xmath25 be a unital @xmath0-algebra and let @xmath8 be a positive integer . 1 . @xmath291 is the least @xmath10 for which there exist @xmath298 such that @xmath299 for some integers @xmath300 , some positive integers @xmath301 , and some @xmath302 .
2 . @xmath303 .
assume that @xmath304 and let @xmath305 , @xmath306 , and @xmath307 be as in definition [ def : cov ] .
write @xmath308 with @xmath309 .
let @xmath310 be a one - dimensional projection in @xmath311 .
it then follows from lemma [ lm : comp ] that the elements @xmath312 satisfy the relations in .
suppose , conversely , that @xmath313 is chosen such that there are elements @xmath314 and @xmath315 in @xmath316 satisfying . put @xmath308 . by the assumption that @xmath317 it follows from lemma [ lm : comp ] ( ii ) that there are mutually orthogonal positive elements @xmath318 in @xmath25 , where @xmath319 , @xmath320 , @xmath321 , such that @xmath322 .
we can further assume that the @xmath323 positive elements @xmath324 are pairwise equivalent .
it then follows from the universal property of the cone over a matrix algebra ( see remark [ rem : cm ] ) that there are @xmath15-homomorphisms @xmath306 such that @xmath325 , where @xmath310 is a one - dimensional projection in the summand @xmath326 of @xmath305 .
the existence of @xmath307 with @xmath296 follows from lemma [ lm : comp ] ( i ) .
thus @xmath327 .
( ii ) . to prove the first inequality ,
assume that @xmath328 and take @xmath329 such that @xmath330 .
then holds with @xmath331 and @xmath332 .
assume next that @xmath333 , and find elements @xmath314 and @xmath315 satisfying the relations in . upon replacing @xmath315 with an integral multiple of @xmath315 we can assume that @xmath334 for all @xmath54 and @xmath75 .
let @xmath335 , @xmath336 , be the sum of a suitable subset of the @xmath315 s such that @xmath337 .
the @xmath338 elements @xmath339 will then witness that @xmath340 .
one can collect the sequence of divisibility numbers @xmath341 of a unital @xmath0-algebra @xmath4 into a single divisibility number as follows : @xmath342 in a similar way one can define @xmath343 and @xmath344
. propositions [ prop : almdiv ] and [ lm : dv - stable ] below hold verbatim for those quantities as well .
however , to keep the exposition bounded , we only treat the case of @xmath345 " .
it follows from proposition [ prop : propinf ] that @xmath346 if and only if @xmath4 is properly infinite and that @xmath347 if @xmath4 is not properly infinite .
[ prop : almdiv ] let @xmath4 be a unital @xmath0-algebra . 1
. @xmath348 for all integers @xmath254 .
2 . @xmath349 ( the limit always exists , but is possibly equal to @xmath350 ) .
if @xmath4 is not properly infinite , then @xmath351 if and only if @xmath286 for all integers @xmath254 . it follows from proposition [ prop : propinf ] and from ( iii ) above , that @xmath4 is almost divisible if and only if @xmath352 ( i.e. , if and only if @xmath346 or @xmath351 ) . ( i ) . if @xmath353 there is nothing to prove .
assume that @xmath354 .
let @xmath254 be given .
let @xmath355 be the smallest integer strictly greater than @xmath356 .
we show that @xmath357 .
choose @xmath358 and a positive integer @xmath359 such that @xmath360 by the definition of @xmath361 there is @xmath362 such that @xmath363 .
take @xmath130 such that @xmath364 .
write @xmath365 , with @xmath366 and @xmath367 . also , write @xmath368 , with @xmath369 and @xmath370
. put @xmath371 . then @xmath372 . with @xmath373 denoting the ceiling function , we have @xmath374 ( ii ) .
it follows from ( i ) that @xmath375 and so the claims follows .
if " part is trivial , and the only if " part follows from ( i ) . we proceed to discuss how @xmath376 behaves under forming matrix algebras : [ lm : dv - stable ] let @xmath4 be a unital @xmath0-algebra . 1 .
@xmath377 for all integers @xmath378 .
2 . if @xmath19 is almost unperforated , then @xmath379 for all integers @xmath378 .
\(i ) follows from remark [ rem : image ] ( as @xmath4 embeds unitally into @xmath380 ) .
assume that @xmath19 is almost unperforated .
we show first that @xmath381 for all @xmath378 . to see this
take any integer @xmath254 , and use proposition [ prop : almdiv ] ( i ) to see that @xmath382 .
write @xmath383 and @xmath384 , where @xmath323 and @xmath385 are positive integers , @xmath386 , and @xmath387 .
identify @xmath388 with @xmath19 in the canonical way , where @xmath389 is identified with @xmath390 . under this identification
we can find @xmath130 such that @xmath391 .
in particular , @xmath392 which by the assumption that @xmath19 is almost unperforated implies that @xmath393 .
this shows that @xmath394 now , @xmath395 as @xmath396 , and so follows by letting @xmath8 tend to infinity . to complete the proof of (
ii ) , take @xmath378 . by ( i )
and we have : @xmath397 for all @xmath398 , which shows that @xmath399 .
we have previously remarked that @xmath400 whenever @xmath272 .
it follows that @xmath353 whenever @xmath401 , i.e. , whenever @xmath4 admits a non - zero finite dimensional representation .
it can happen that @xmath402 .
take for example @xmath4 such that @xmath380 is properly infinite , but @xmath4 itself is not properly infinite , cf .
. then @xmath403 and @xmath347 .
it is an important open problem if @xmath352 ( i.e. , if @xmath4 is almost divisible ) for every ( simple ) unital infinite dimensional @xmath0-algebra @xmath4 for which @xmath19 is almost unperforated .
the property that any of the divisibility numbers @xmath244 , @xmath243 , and @xmath196 is finite , when @xmath4 is a unital @xmath0-algebra , has interpretations in terms of structural properties of the @xmath0-algebra @xmath4 .
we have already noted that the divisibility numbers always are finite when @xmath4 is a simple @xmath0-algebra , and the corresponding structural properties of the @xmath0-algebra are , as we shall see , trivially satisfied for simple @xmath0-algebras .
the correct definition of finite divisibility " in the non - unital case is what we call @xmath405-divisibility as defined below .
[ def : omega - divisibility ] let @xmath4 be a @xmath0-algebra , let @xmath68 , and let @xmath8 be a positive integer .
then : 1 .
@xmath70 is _
@xmath405-divisible _ if for all @xmath128 with @xmath129 there exists @xmath130 such that @xmath131 and @xmath406 for some positive integer @xmath10 .
2 . @xmath70 is _
@xmath405-decomposable _ if for all @xmath128 with @xmath129 there exist elements @xmath135 such that @xmath136 and @xmath407 for some positive integer @xmath10 and for all @xmath75 .
3 . @xmath70 is _ weakly @xmath405-divisible _ if for all @xmath128 with @xmath129 there exist elements @xmath408 in @xmath19 such that @xmath142 for all @xmath75 and @xmath409 .
[ rem : omega - div ] if @xmath70 in definition [ def : omega - divisibility ] is compact , then @xmath70 is @xmath405-divisible , @xmath405-decomposable , respectively , weakly @xmath405-divisible if and only if @xmath410 , @xmath411 , respectively , @xmath412 , cf . remark [ rem : divisible - unital ] . in the next result
we express @xmath405-divisibility in terms of structural properties of the @xmath0-algebra . part ( iii ) is almost contained in @xcite ( see ( * ? ? ?
* definition 3.1 ) and ( * ? ? ?
* remark 3.3 ( 7 ) ) and compare with definition [ def : cov ] and proposition [ prop : cov ] ) .
recall the definition of the rank of a @xmath0-algebra from definition [ def : rank ] .
[ thm : finite - div ] let @xmath4 be a @xmath240-unital @xmath0-algebra and let @xmath413 be a strictly positive element of @xmath4 .
( if @xmath4 is unital , we can take @xmath413 to be the unit of @xmath4 . )
put @xmath414 . 1 .
@xmath70 is @xmath405-divisible if and only if for every @xmath94 there exists a @xmath15-homomorphism @xmath415 such that @xmath416 belongs to the closed two - sided ideal generated by the image of @xmath417 .
@xmath70 is @xmath405-decomposable if and only if for every @xmath94 there exist mutually orthogonal positive elements @xmath116 in @xmath4 such that @xmath416 belongs to the closed two - sided ideal generated by @xmath418 for each @xmath54 .
3 . the following are equivalent : 1 .
@xmath70 is weakly @xmath405-divisible , 2 .
@xmath419 , 3 .
there exist @xmath15-homomorphisms @xmath420 , @xmath421 , for some @xmath10 , such that @xmath416 belongs to the closed two - sided ideal generated by the union of the images of the @xmath422 s .
let us assume that @xmath70 is @xmath405-divisible .
let @xmath94 .
find @xmath230 and a positive integer @xmath10 such that @xmath423 and @xmath424 .
choose a positive element @xmath36 in @xmath425 such that @xmath426 , and choose @xmath427 such that @xmath428 .
by lemma [ lm : cone - comp ] there exists @xmath415 such that @xmath429 . then @xmath430 which implies that @xmath416 belongs to the closed two - sided ideal generated by the image of @xmath417 .
suppose conversely that for every @xmath94 there exists @xmath431 such that @xmath416 is in the closed two - sided ideal generated by @xmath432 .
. then @xmath423 by lemma [ lm : cone - comp ] , while @xmath434 for some positive integer @xmath10 .
this shows that @xmath70 is @xmath405-divisible .
only if " .
let @xmath94 and suppose that @xmath435 in @xmath4 exist with the stipulated properties .
set @xmath436 .
then @xmath437 since @xmath416 belongs to the closed two - sided ideal generated by @xmath123 , @xmath438 for some integer @xmath313 .
it follows that @xmath70 is @xmath405-decomposable . if " . if @xmath439 is @xmath405-decomposable and if @xmath98 , then there are positive elements @xmath440 in @xmath425 such that @xmath441 and @xmath442 for some positive integer @xmath10 .
choose @xmath427 such that @xmath443 for all @xmath75 .
by lemma [ lm : comp ] ( ii ) there are pairwise orthogonal positive elements @xmath122 in @xmath4 such that @xmath444 .
then the closed two - sided ideal generated by @xmath123 contains @xmath416 for each @xmath75 .
( a ) @xmath121 ( b ) .
assume that @xmath70 is weakly @xmath405-divisible .
suppose that @xmath4 has an irreducible representation @xmath445 of finite positive dimension @xmath170 .
then @xmath446 is necessarily surjective . since @xmath405-divisibility is preserved by @xmath15-homomorphisms ( cf .
remark [ rem : image ] ) , we conclude that @xmath447 is weakly @xmath405-divisible .
but then @xmath448 , cf .
example [ rem : matrix ] .
hence ( b ) holds .
\(b ) @xmath121 ( c ) .
assume that ( b ) holds .
let @xmath449 be the family of all non - zero @xmath15-homomorphisms @xmath450 and let @xmath451 be the closed two - sided ideal in @xmath4 generated by the images of all @xmath422 s .
thus each @xmath422 maps @xmath16 into @xmath451 .
we claim that @xmath452 .
assume , to reach a contradiction , that @xmath453 . by the assumption that @xmath454 , all irreducible representations of @xmath455 have dimension at least @xmath8 .
it follows from glimm s lemma ( proposition [ prop : glimm ] ) that there is a non - zero @xmath15-homomorphism @xmath456 , which by projectivity lifts to a @xmath15-homomorphism @xmath90 .
but the image of @xmath417 is not contained in @xmath451 , which is a contradiction . for each finite subset @xmath457 of @xmath458
consider the closed two - sided ideal @xmath459 of @xmath4 generated by @xmath460 .
then @xmath4 is the closure of the union of the upwards directed family of ideals @xmath461 . hence , for each @xmath98 , there is a finite subset @xmath457 of @xmath458 such that @xmath416 belongs to @xmath462 . thus ( c ) holds .
\(c ) @xmath121 ( a ) .
assume that ( c ) holds .
set @xmath463 for @xmath421 .
then @xmath464 for all @xmath54 .
moreover , @xmath124 belongs to the algebraic ideal generated by the @xmath10 elements @xmath465 , whence @xmath466 for suitable positive integers @xmath467 . put @xmath468 and let @xmath469 be a listing of the elements @xmath470 , with @xmath471 repeated @xmath467 times . then @xmath142 and @xmath472
this shows that @xmath70 is weakly @xmath405-divisible .
the theorem above can be simplified in the case where @xmath70 is compact , and in particular in the case where @xmath4 is unital : [ cor : finite - div ] let @xmath4 be a unital @xmath0-algebra , and let @xmath8 be a positive integer . then : 1 . @xmath473 if and only if there exists a @xmath15-homomorphism @xmath415 whose image is full in @xmath4 .
2 . @xmath474 if and only if there exist full , pairwise orthogonal positive elements @xmath116 in @xmath4 .
3 . the following are equivalent : 1 .
@xmath475 , 2 .
@xmath419 , 3 .
there exist @xmath15-homomorphisms @xmath420 , @xmath421 for some @xmath10 , such that the union of their images is full in @xmath4 . propositions [ prop : glimm ] ( glimm ) , proposition [ prop : a ] , and corollary [ cor : finite - div ] ( i ) immediately imply : [ cor : simple ] if @xmath4 is a unital , infinite dimensional , simple @xmath0-algebra , then the three divisibility numbers @xmath244 , @xmath243 , and @xmath196 are finite for every integer @xmath198 .
let us also note what it means to have infinite @xmath476 numbers : [ cor : char ] let @xmath4 be a unital @xmath0-algebra . 1 .
@xmath4 admits a character if and only if @xmath477 .
2 . @xmath4 admits a finite dimensional representations if and only if @xmath478 for some integer @xmath254 .
glimm s lemma ( proposition [ prop : glimm ] ) says that there exists a non - zero @xmath15-homomorphism from @xmath479 into a @xmath0-algebra @xmath4 if and only if @xmath4 admits an irreducible representation of dimension at least @xmath10 .
it is not known how
large " one can make the image of such a @xmath15-homomorphism . in particular , it is not known for which @xmath0-algebras @xmath4 one can find a @xmath15-homomorphism @xmath268 whose image is full in @xmath4 ( i.e. , the image is not contained in any proper closed two - sided ideal in @xmath4 ) . for @xmath480 this problem is known as the global glimm halving problem " ( see @xcite , @xcite and @xcite ) .
a unital @xmath0-algebra @xmath4 is said to have the _ global glimm halving property _ if there is a @xmath15-homomorphism @xmath481 with full image .
more specifically , one can ask if any ( unital ) @xmath0-algebra , which admits no finite dimensional representation , satisfies the global glimm halving property . in view of corollary [ cor : finite - div ] , this problem for unital @xmath0-algebras @xmath4 may be restated as follows : does @xmath482 imply that @xmath478 for some positive integer @xmath8 ? for a non - unital @xmath0-algebra @xmath4 , the one can restate the problem in the following way : does @xmath482 imply that @xmath241 fails to be @xmath405-divisible for some positive integer @xmath8 .
it is shown in @xcite that if @xmath4 is a weakly purely infinite @xmath0-algebra , then @xmath4 is purely infinite if and only if all hereditary sub-@xmath0-algebras of @xmath4 have the global glimm halving property .
( it is easy to see that the rank of any weakly purely infinite @xmath0-algebra is infinite . )
it is an open problem if all weakly purely infinite @xmath0-algebras are purely infinite .
let @xmath4 be a unital @xmath0-algebra .
it follows from proposition [ prop : a ] ( and also from corollary [ cor : finite - div ] ) that @xmath483 for all positive integers @xmath8 .
none of the two reverse implications hold in general .
for each integer @xmath254 consider the bott projection @xmath280 in @xmath484 .
this projection has ( complex ) dimension @xmath8 and it has no non - trivial sub - projections .
the unital @xmath0-algebra @xmath485 is a homogeneous @xmath0-algebra of rank @xmath8 .
hence @xmath475 .
suppose that @xmath474 . then , by corollary [ cor : finite - div ] ( ii ) , there would exist full , pairwise orthogonal , positive elements @xmath486 in @xmath4 .
this would entail that each @xmath123 is one - dimensional in each fiber of @xmath4 , and hence that @xmath487 is a one - dimensional projection for some @xmath488 .
but this contradicts the fact that @xmath280 has no proper subprojections . to see that @xmath474 does not imply @xmath473 ,
consider the @xmath0-algebra @xmath489 , and take the ( one - dimensional ) bott projection @xmath490 and a trivial @xmath491-dimensional projection @xmath492 such that @xmath280 and @xmath281 are orthogonal .
put @xmath493 .
it follows from a @xmath494-theoretical argument that @xmath495 can not be written as the sum of @xmath8 pairwise orthogonal and equivalent projections ( because @xmath496 $ ] is not divisible by @xmath8 in @xmath11 ) .
the unit of @xmath4 can be written as the sum of @xmath8 ( necessarily full ) projections , so @xmath474 .
assume that @xmath473 .
then , by corollary [ cor : finite - div ] ( i ) , there is a @xmath15-homomorphism @xmath90 whose image is full in @xmath4 .
as explained in remark [ rem : cm ] , this entails that there exist full , pairwise orthogonal , pairwise equivalent , positive elements @xmath122 in @xmath4 . arguing as in the paragraph above , we can assume that each @xmath123 is in fact a projection .
but that contradicts the fact that @xmath497 is not the sum of @xmath8 pairwise equivalent projections .
let @xmath4 and @xmath25 be @xmath0-algebras . then there is a natural bi - additive map @xmath498 defined as follows : if @xmath426 and @xmath499 with @xmath36 a positive element in @xmath425 and @xmath85 a positive element in @xmath500 , then @xmath501 , where we identify @xmath502 with @xmath503 .
note that @xmath504 if @xmath505 and @xmath506 . part ( i ) of the following result was ( implicitly ) proved in ( * ? ? ?
* lemma 4.3 ) , and was used to prove that @xmath507 is almost unperforated for all unital @xmath0-algebras @xmath4 .
[ lm : z ] let @xmath4 and @xmath25 be unital @xmath0-algebras and let @xmath508 be integers . 1 .
let @xmath509 be such that @xmath510 .
if @xmath25 is @xmath14-divisible , then @xmath511 .
2 . let @xmath512 be such that @xmath513 for all @xmath75 . if @xmath25 is @xmath14-decomposable , then @xmath514 3 .
let @xmath515 be such that @xmath516 for all @xmath75 . if @xmath25 is weakly @xmath14-divisible , then @xmath517 ( i ) .
take @xmath518 such that @xmath519 . then @xmath520 ( ii ) .
take @xmath521 such that @xmath522 . then @xmath523 ( iii ) .
take @xmath524 such that @xmath525 .
then @xmath526 the lemma above can loosely be paraphrased as follows : good divisibility properties of @xmath25 ensure good comparability properties of @xmath527 , and bad comparability properties of @xmath527 entail bad divisibility properties of @xmath25 .
we proceed to show that infinite tensor products of ( suitable ) unital @xmath0-algebras can not have very bad comparability properties .
[ lm : limit1 ] let @xmath528 be a sequence of unital @xmath0-algebras such that @xmath529 .
then @xmath530 for all integers @xmath254 , where @xmath10 is any integer satisfying @xmath531 .
take @xmath10 such that @xmath531 .
for each @xmath170 , find @xmath532 , @xmath533 , such that @xmath534 .
given a multi - index @xmath535 , put @xmath536 then @xmath537 recall that a @xmath0-algebra @xmath4 has the corona factorization property if and only if all full projections in @xmath538 are properly infinite .
it was shown in @xcite that if @xmath4 is a separable @xmath0-algebra , then @xmath4 and all its closed two - sided ideals have the corona factorization property if and only if for every integer @xmath254 , and for all @xmath539 in @xmath19 such that @xmath540 and @xmath541 for all @xmath75 , one has @xmath542 for some integer @xmath543 .
let @xmath528 be a sequence of unital @xmath0-algebras such that @xmath544 it follows that the @xmath0-algebra @xmath545 and all its closed two - sided ideals have the corona factorization property .
in particular , if @xmath4 is a unital @xmath0-algebra without characters then @xmath546 and all its closed two - sided ideals have the corona factorization property .
put @xmath547 , and for each @xmath313 put @xmath548 and @xmath549 .
we shall view @xmath550 as an increasing sequence of sub-@xmath0-algebras of @xmath25 such that @xmath551 is dense in @xmath25 , and we shall identify @xmath25 with @xmath552 for all @xmath10 .
let @xmath198 be an integer and let @xmath553 in @xmath316 be such that @xmath540 and @xmath541 for all @xmath75 .
by lemma [ lm : limit1 ] there is a positive integer @xmath554 such that @xmath555 for all @xmath10 .
we show that @xmath542 .
this will prove that @xmath25 has the corona factorization property .
repeated use of proposition [ prop : culimits ] ( i ) and ( ii ) shows that there exists a positive integer @xmath10 , and elements @xmath556 in @xmath557 such that @xmath558 where @xmath559 denotes the canonical embedding @xmath560 .
we can now apply lemma [ lm : z ] ( iii ) to deduce that @xmath561 as desired .
a trivial obstruction to ( weak ) divisibility of a @xmath0-algebra is its rank : @xmath562 if and only if @xmath563 ( by corollary [ cor : finite - div ] ( iii ) ) . in this section we shall discuss ways of obtaining homogeneous @xmath0-algebras with large rank and large weak divisibility constant .
we use these techniques to construct unital simple @xmath0-algebras with large weak divisibility constants .
we remark first that lemma [ lm : z ] provides non - trivial obstructions to divisibility in @xmath25 .
indeed , it follows by that lemma that if there exists a unital @xmath0-algebra @xmath4 and @xmath509 such that @xmath510 but @xmath564 , then @xmath565 .
similarly , if there exist @xmath566 in @xmath19 such that @xmath513 for all @xmath75 while @xmath567 then @xmath568 . finally , if there exist @xmath569 such that @xmath541 for all @xmath75 while @xmath570 then @xmath571 . we introduce below another way to obtain bad divisibility behavior : [ lm:2-div ]
let @xmath572 be compact elements .
if @xmath573 then there exist @xmath574 in @xmath19 such that @xmath575 for all @xmath54 and @xmath576 by assumption there exists @xmath27 such that @xmath577 . by compactness of @xmath578
we can find @xmath579 such that @xmath580 .
since @xmath581 , it follows from property ( p1 ) of the cuntz semigroup ( see section [ sec : preliminary ] ) ( leaving @xmath70 unchanged ) that there exists @xmath582 such that @xmath583 as @xmath584 there is @xmath585 such that @xmath586 .
apply ( p2 ) to the relation @xmath587 to obtain @xmath588 satisfying @xmath589 by compactness of @xmath590 we can find @xmath591 such that @xmath592 . now ,
@xmath593 and so we can use ( p1 ) ( leaving @xmath594 unchanged ) to find @xmath595 such that @xmath596 it follows that @xmath597 and that @xmath598 as desired .
the corollary below illustrates how the preceding lemma can be used to find elements with bad divisibility properties : [ cor:2-div - obstruct ] let @xmath599 be a compact hausdorff space and suppose that @xmath600 is a projection such that @xmath601\nleq ( 2n+1)[p]$ ] in @xmath602 , where @xmath603 denotes the unit of @xmath604
. then @xmath605 .
consider the compact elements @xmath606 and @xmath607 of @xmath608 .
note that @xmath609 implies @xmath610 and that @xmath611 .
the desired conclusion now follows from lemma [ lm:2-div ] . the relation @xmath601\nleq ( 2n+1)[p]$ ] in @xmath602 is satisfied whenever the @xmath612-fold direct sum of @xmath280 with itself is a projection with non - trivial euler class ( as explained in more detail below ) .
it is known that for each integer @xmath613 and for each positive integer @xmath554 there exist @xmath599 and @xmath614 such that @xmath615 $ ] has non - trivial euler class and @xmath280 has rank @xmath616 .
the unital @xmath0-algebra @xmath617 with this choice of @xmath599 and @xmath280 will then satisfy @xmath618 and @xmath619
. we will now give a different method for constructing elements with large divisibility constants and large rank , where we get upper bounds and where we also can give sharper lower bounds for the weak divisibility constant .
let @xmath620 denote the 2-dimensional sphere .
let @xmath280 denote the bott - projection " in @xmath621 , i.e. , the projection associated to the hopf line bundle over @xmath620 .
for each @xmath622 , let @xmath623 be given by @xmath624 since @xmath625 in @xmath626 , we have @xmath627 as another obstruction to weak divisibility , we shall use the following corollary of lemma [ lm : z ] ( iii ) , cf .
the remarks at the beginning of this section , applied to the relations @xmath628 , @xmath629 .
[ cor : key ] let @xmath599 be a locally compact hausdorff space , and let @xmath630 be a projection .
let @xmath631 be the projections in @xmath632 defined in the preceding paragraph .
suppose that @xmath633 then @xmath634 .
let us now give examples of projections to which the corollary above can be applied .
we will make use of characteristic classes of vector bundles . recall that projections in @xmath635 , with @xmath599 compact and hausdorff , give rise to vector bundles over @xmath599 :
if @xmath280 is a projection , then @xmath636 , with @xmath637 is the vector bundle associated to @xmath280 . up
to murray - von neumann equivalence of projections and isomorphism of vector bundles , this correspondence is a bijection .
we denote by @xmath638 , or simply @xmath639 , the euler class of @xmath640 . for the cartesian product of spheres
@xmath641 we have ( e.g. , by the knneth formula ) that @xmath642/(z_1 ^ 2,z_2 ^ 2,\dots , z_n^2).\ ] ] with this identification , the euler classes of the projections @xmath643 defined earlier can be shown to be @xmath644 .
[ prop : euler - obstruct ] let @xmath599 be a compact hausdorff space and let @xmath645 be a projection such that @xmath646
. then @xmath647 by corollary [ cor : key ] it suffices to show that @xmath648 in @xmath649 , where @xmath650 denotes the unit in @xmath604 .
( in the formulation of the proposition above , we denoted @xmath650 simply by @xmath603 . )
observe that the trivial rank 1 projection is a subprojection of the projection on the left - hand side of .
thus , it suffices to show that the right side of has non - zero euler class .
set @xmath651 . for each positive integer @xmath54 ,
let @xmath652 denote the @xmath54th characteristic class of @xmath281 ( so that @xmath653 ) . by the knneth theorem ( ( * ?
* theorem a.6 ) ) , we can identify @xmath654 with @xmath655 .
then @xmath656 in the above computation we have used that @xmath657 , for @xmath281 a projection of rank @xmath170 and @xmath280 a projection of rank 1 . to obtain the last equality we have used that @xmath658 for all @xmath54 .
let us now give examples of families of projections to which the above proposition can be applied .
we shall here and in the following , whenever @xmath280 is a projection ( in a @xmath0-algebra ) and @xmath10 is a positive integer , let @xmath659 denote the @xmath10-fold direct sum , @xmath660 , ( in a matrix algebra over the given @xmath0-algebra ) of the projection @xmath280 .
let @xmath554 be a positive integer , and let @xmath661 denote the @xmath662-dimensional complex projective space .
let @xmath663 denote the tautological line bundle over @xmath661 and @xmath664 the rank 1 projection associated to it .
it is known that @xmath665/(z^{2n})$ ] , where we have identified @xmath666 with @xmath667/(z^{2n})$ ] .
let @xmath668 be positive integers such that @xmath669 .
then @xmath670 .
it follows that @xmath671 indeed , the first inequality follows from proposition [ prop : euler - obstruct ] and the calculations made above .
the second inequality follows from proposition [ prop : a ] .
the last inequality can be proved as follows : put @xmath672
. then @xmath673 . by a classical result about vector bundles ( see ( * ? ? ?
* chapter 9 , proposition 1.1 ) ) we have that @xmath674 if @xmath675 .
it follows that @xmath676 or , equivalently , that @xmath677 , if @xmath678 . in this and the following two subsections we give examples of unital simple @xmath0-algebras with bad divisibility behaviour .
they use the euler class obstruction described in the following example .
[ ex : spheres ] let @xmath616 be a positive integer . following the notation in @xcite , for each set @xmath679 , let @xmath680 be the one - dimensional projection in @xmath681 given by @xmath682 where @xmath683 is as defined above corollary [ cor : key ] .
it is shown in ( * ? ? ?
* proposition 4.5 ) that if @xmath684 are subsets of @xmath685 that admit a matching ( i.e. , @xmath686 for all subset @xmath457 of @xmath687 ) then the euler class of @xmath688 is non - zero .
the examples constructed in this and the following two subsectons are built on the same template described in the following lemma ( which is a variation of one of villadsen s constructions ) .
we retain the terminology from the example above throughout the rest of this section .
[ lm : template ] let @xmath689 be a sequence of pairwise disjoint subsets of @xmath690 .
choose @xmath691 large enough so that all @xmath692 , @xmath693 , are contained in the set @xmath694 .
( this , of course , can be accomplished by taking @xmath695 . )
consider the projection @xmath696 of rank @xmath697 in @xmath698 given by @xmath699 it follows that there is a simple unital ah - algebra @xmath4 which is the inductive limit of the sequence @xmath700
^ -{\varphi_1 } & q_2 \big ( c((s^2)^{d_2 } ) \otimes { { \cal k}}\big ) q_2 \ar[r]^-{\varphi_2 } & \cdots \ar[r ] & a,}\ ] ] where the connecting mappings @xmath701 are unital .
set @xmath702 and @xmath703 .
write @xmath704 let @xmath705 be the projection mapping , and let @xmath706 denote the composition map @xmath707
. choose @xmath708 for each @xmath10 such that the set @xmath709 is dense in @xmath710 for all @xmath198 .
define a @xmath15-homomorphism @xmath711 by @xmath712 where we in a suitable way have identified @xmath713 with a subalgebra of @xmath714 .
we also identify @xmath635 with @xmath715 .
we make another identification : if @xmath716 then the projection @xmath717 is defined both in @xmath718 and in @xmath719 and @xmath720 ( where the former occurrence of @xmath721 is viewed as an element in the former algebra , and the latter in the latter ) .
we shall use the same notation for the two projections . taking these identification a step further
, we have @xmath722 and that @xmath723 . (
these identification hold , strictly speaking , only up to conjugation with an inner automorphism on @xmath714 . ) in this notation we get @xmath724 for all @xmath725 , i.e. , @xmath726 ( possibly after composing @xmath727 with an inner automorphism on @xmath714 ) .
this shows that @xmath727 maps @xmath33 unitally into @xmath728 .
let @xmath701 denote the unital @xmath15-homomorphism that arises in this way ,
i.e. , @xmath701 is the restriction of @xmath727 to @xmath33 ( and the co - restriction to @xmath728 ) , and let @xmath4 be the inductive limit of the sequence @xmath729
^ -{\varphi_1 } & a_2 \ar[r]^-{\varphi_2 } & a_3 \ar[r]^-{\varphi_3 } & \cdots \ar[r ] & a.}\ ] ] let @xmath730 denote the composition map @xmath731 when @xmath732 .
one can check that @xmath733 is non - zero for all @xmath734 if @xmath735 is a function in @xmath736 which is non - zero on at least one point in the set @xmath737 . by the choice of the points @xmath738
, it follows that for each @xmath8 and for each non - zero @xmath735 in @xmath739 there is @xmath740 such that @xmath741 is full in @xmath33 ( i.e. , that @xmath742 for all @xmath734 ) .
this entails that @xmath4 is simple .
[ lm : simple-1 ] let @xmath554 be a positive integer . in the notation of lemma [ lm : template ] choose the sequence @xmath743 such that @xmath744 .
it then follows that @xmath745 for all @xmath10 .
we use proposition [ prop : euler - obstruct ] to prove the first claim .
it suffices to show that the euler class of the projection @xmath746 is non - zero .
but this follows from ( * ? ? ?
* proposition 4.5 ) , cf .
example [ ex : spheres ] above , and from the choice of the sets @xmath747 . to prove the second claim , put @xmath748 and note that @xmath749 .
it follows from ( * ? ? ?
* proposition 1 ) that @xmath750 if @xmath751 .
this shows that @xmath752 .
[ thm : simple ] for each positive integer @xmath554 there exists a simple unital infinite dimensional ah - algebra @xmath4 such that @xmath753 .
let @xmath4 be the simple @xmath0-algebra constructed in lemma [ lm : template ] based on the choice of @xmath743 made in lemma [ lm : simple-1 ] .
then @xmath4 is the inductive limit of the sequence of @xmath0-algebras @xmath754 with unital connecting mappings .
it follows from lemma [ lm : simple-1 ] that @xmath755 for all @xmath10 , and that @xmath756 . by proposition [ prop : inductive - limits ] and remark [ rem : image ] , @xmath757 and @xmath758 .
suppose that @xmath759 is a class of unital @xmath0-algebra .
an element @xmath4 in @xmath759 is an _
inital object in @xmath759 _ if there exists a unital @xmath15-homomorphism @xmath248 for every @xmath25 in @xmath759 .
it is well - known that the cuntz algebra @xmath760 is an initial object in the class of unital properly infinite @xmath0-algebras .
in fact , a unital @xmath0-algebra is properly infinite if and only if it contains @xmath760 as a unital sub-@xmath0-algebra .
every properly infinite unital sub-@xmath0-algebra of @xmath760 is then also an initial object in the class of unital properly infinite @xmath0-algebras .
hence the cuntz - toeplitz algebras , @xmath761 , @xmath378 , are initial objects and so are all unital kirchberg algebras @xmath4 for which the assignment @xmath762 \mapsto 1 $ ] extends to a homomorphism @xmath763 .
it was shown in @xcite that also the class of unital @xmath0-algebras of real rank zero and of infinite rank has initial objects .
one can even find initial objects to this class which are simple af - algebras ( necessarily with infinite dimensional trace simplex ) .
it follows in particular that the class of unital simple infinite dimensional @xmath0-algebras of real rank zero has initial objects .
clearly , @xmath34 is an initial object in the category of all unital @xmath0-algebras , and so is any unital @xmath0-algebra that admits a character .
( note that we do not require the unital @xmath15-homomorphism @xmath248 to be injective . )
the corollary below shows that initial objects do not exist in the general non - real rank zero case .
the class of unital simple infinite dimensional @xmath0-algebras and the class of unital @xmath0-algebras of infinite rank do not have initial objects .
in fact , there is no unital @xmath0-algebra without characters that maps unitally into every unital simple infinite dimensional @xmath0-algebra . if @xmath4 is a unital @xmath0-algebra that maps unitally into every unital simple infinite dimensional @xmath0-algebra , then @xmath764 for all unital simple infinite dimensional @xmath0-algebras @xmath25 , cf .
remark [ rem : image ] , whence @xmath477 by theorem [ thm : simple ] . on the other hand , if @xmath4 has no character , then @xmath765 by corollary [ cor : char ] .
we can give a lower and an upper bound on the asymptotic divisibility constant ( discussed in section [ sec : stable - div ] ) for the @xmath0-algebra considered above : [ cor : simple - asymp ] let @xmath554 be a positive integer , and let @xmath4 be the simple ah - algebra constructed in theorem [ thm : simple ] associated with @xmath554 .
it follows that @xmath766 by proposition [ prop : almdiv ] we get that @xmath767 . to prove the reserve inequality , take any positive integer @xmath10 and put @xmath768 .
we show that @xmath769 , where @xmath33 is as in the proof of theorem [ thm : simple ] . in the notation of lemma [ lm : simple-1 ] , let @xmath770 , put @xmath771 , and recall that @xmath696 is the unit of the @xmath0-algebra @xmath33 . by the definition of @xmath696 ( in lemma [ lm : simple-1 ] ) it follows that @xmath131 . as @xmath772 if follows from ( * ? ? ?
* proposition 1 ) that @xmath773 , whence @xmath774 .
this proves that @xmath769 .
it follows from remark [ rem : image ] that @xmath775 whenever @xmath8 is a power of 2 , and this entails that @xmath776 as desired .
we can use lemma [ lm : template ] to construct a unital , simple ah - algebra @xmath4 such that @xmath353 .
the proof requires the following sharpening of corollary [ cor : key ] that may have independent interest .
[ cor : key - m ] let @xmath599 be a locally compact hausdorff space , and let @xmath630 be a projection .
let @xmath8 and @xmath554 be positive integers , let @xmath777 be pairwise disjoint subsets of @xmath690 with @xmath778 for all @xmath54 , and let @xmath779 be the associated projections in @xmath780 defined in example [ ex : spheres ] .
suppose that @xmath781 then @xmath782 .
apply lemma [ lm : z ] ( iii ) to @xmath783 and @xmath784 , and note that @xmath785 , cf .
example [ ex : spheres ] .
[ lm : simple-2 ] let @xmath743 be a sequence of pairwise disjoint subsets of @xmath690 with @xmath786 .
then , in the notation of lemma [ lm : template ] , we have @xmath787 if @xmath788 , and that @xmath789 if @xmath790 .
we use corollary [ cor : key - m ] with @xmath791 and @xmath792 to prove the first claim .
as @xmath793 it suffices to show that @xmath794 has non - trivial euler class , when @xmath795 are as in corollary [ cor : key - m ] .
write @xmath796 as explained in example [ ex : spheres ] , to prove non - triviality of the euler class of the projection @xmath794 one needs to verify the combinatorial fact that the family of sets @xmath797 , @xmath798 , @xmath799 , and with the set @xmath800 repeated @xmath801 times , satisfies the marriage lemma condition . by first exhausting the elements in the sets @xmath802 , and using that @xmath803 , it suffices to show that the family of sets @xmath804 , @xmath805 , with each set repeated @xmath806 times , satisfies the marriage lemma condition .
however , this holds because @xmath807 when @xmath808 .
the second claim follows from the fact that the dimension of the projection @xmath696 is @xmath697 and that @xmath809 whenever @xmath810 .
[ thm : simple-2 ] there is a simple unital infinite dimensional ah - algebra @xmath4 which satisfies @xmath353 .
let @xmath4 be the simple ah - algebra contructed in lemma [ lm : template ] with respect to the choice of @xmath804 from lemma [ lm : simple-2 ] . recall that @xmath4 is an inductive limit of a sequence of unital @xmath0-algebras @xmath811 , where @xmath812 it therefore follows from lemma [ lm : simple-2 ] that @xmath813 , when @xmath814 , and @xmath815 when @xmath790 .
this entails that @xmath816 by proposition [ prop : inductive - limits ] .
finally , by proposition [ prop : almdiv ] ( ii ) , @xmath817 as remarked in section [ sec : stable - div ] , if @xmath4 is any unital @xmath0-algebra , then @xmath346 if and only if @xmath4 is properly infinite , and @xmath818 $ ] otherwise .
moreover , @xmath351 if and only if @xmath4 is almost divisible ( and not properly infinite ) . in other words ,
the range of the invariant @xmath376 is contained in the set @xmath819 $ ] , and @xmath352 if and only if @xmath4 is almost divisible .
we can easily produce examples of simple , unital , infinite dimensional @xmath0-algebras @xmath4 such that @xmath346 ( eg .
, @xmath4 could be a cuntz algebra ) , or such that @xmath351 ( eg .
, @xmath4 is any simple , unital , infinite - dimensional @xmath0-algebra of real rank zero , cf .
example [ ex : rr0 ] ) .
the theorem above provides an example of a simple , unital , infinite dimensional @xmath0-algebra @xmath4 where @xmath353 .
it follows from corollary [ cor : simple - asymp ] that @xmath820 attains infinitely many values in the interval @xmath821 , when restricted to the class of unital , simple , infinite dimensional @xmath0-algebras , and that the possible values of @xmath820 in this interval is upwards unbounded .
we do not know if all values in the interval @xmath821 are thus attained .
for that matter we can not exhibit any number in the interval @xmath821 which for sure is the value of @xmath361 for some simple unital infinite dimensional @xmath0-algebra @xmath4 .
we end this section by giving yet another class of examples of simple unital @xmath0-algebras with bad divisibility properties .
the ones we construct below are of the form @xmath822 , where the @xmath823 s are unital simple infinite dimensional @xmath0-algebras .
in particular , such @xmath0-algebras need not absorb the jiang - su algebra tensorially .
it remains an open problem if @xmath824 absorbs the jiang - su algebra whenever @xmath4 is a simple unital infinite dimensional @xmath0-algebra ( or a unital @xmath0-algebra without characters ) , cf .
@xcite .
it was shown in ( * ? ? ?
* example 4.8 ) that there exists a sequence @xmath1 of homogeneous @xmath0-algebras of rank two such that @xmath32 does not absorb the jiang - su algebra tensorially .
( it is an easy conseqence of this that the jiang - su algebra can not embed unitally into @xmath825 for some @xmath170 . ) of course , one can regroup the tensor factors @xmath33 to get a new sequence @xmath826 of unital @xmath0-algebras each of which has infinite rank and where the jiang - su algebra does not embed into @xmath827 .
it is not known if every unital @xmath0-algebra of infinite rank admits an embedding of a unital simple infinite dimensional @xmath0-algebra .
if it were true , then theorem [ thm : inf - tensor ] below would follow from ( * ? ? ?
* example 4.8 ) .
we introduce some notation to keep track of the combinatorics .
define a total order on the set @xmath828 by @xmath829 for each @xmath830 and for each integer @xmath831 let @xmath832 denote the set of all @xmath8-tuples @xmath833 such that @xmath834 and @xmath835 for all @xmath836 .
[ lm : inf - tensor ] let @xmath543 be an integer . for each integer @xmath398 , let @xmath837 be a sequence of subsets of @xmath690 such that @xmath838 whenever @xmath839 , and such that @xmath840 ( the quantity on the right - hand side is finite because @xmath832 is finite for all @xmath841 and all @xmath831 , and @xmath842 when @xmath843
let @xmath844 and @xmath845 be as defined in lemma [ lm : template ] associated with the sequence @xmath837 .
it then follows that @xmath846 for all positive integers @xmath10 and @xmath8 .
let @xmath847 be the set of all non - zero @xmath8-tuples @xmath833 such that @xmath848 for all @xmath849 .
adopt the convention @xmath850 for all @xmath170 and let @xmath851 denote the trivial (= constant ) one - dimensional projection .
we can then express the projection @xmath852 as follows : @xmath853 by proposition [ prop : euler - obstruct ] it suffices to show that the euler class of the projection @xmath854 is non - zero , or , equivalently , that the euler class of the projection @xmath855 is non - zero . by ( * ? ? ?
* proposition 4.5 ) , cf .
example [ ex : spheres ] , it suffices to show that the family of sets @xmath856 , where @xmath857 and where the set @xmath856 is repeated @xmath858 times , admits a matching .
construct a matching by selecting the matching elements for the @xmath858 copies of the set @xmath856 inside the subset @xmath859 , where @xmath841 is the largest of the elements @xmath860 . to check
that this work , i.e. , to see that @xmath861 is large enough , let @xmath862 be the set of those @xmath8-tuples @xmath863 in @xmath847 for which @xmath864 then @xmath865 , and so it follows by the assumption on @xmath866 that @xmath867 the suggested matching is therefore possible .
the theorem below shows that an infinite tensor product of simple unital infinite dimensional @xmath0-algebras does not necessarily have good divisibility properties .
any such @xmath0-algebra @xmath868 will have ( many ) non - trivial central sequences , i.e. , the central sequence algebra @xmath869 with respect to an ultrafilter @xmath404 on @xmath690 is non - trivial .
for example , @xmath16 embeds into @xmath869 for all @xmath8 , albeit , not necessarily with full image .
however , in the example contructed below , one can not embed the jiang - su algebra into @xmath869 .
[ thm : inf - tensor ] for each integer @xmath870 , there exist a sequence @xmath1 of unital simple infinite dimensional @xmath0-algebras ( in fact , ah - algebras ) such that @xmath871 in particular , @xmath32 does not absorb the jiang - su algebra @xmath24 , and it does not even admit a unital embedding of @xmath24 .
let @xmath872 be the simple unital ah - algebra constructed in lemma [ lm : template ] associated with the sequence @xmath873 from lemma [ lm : inf - tensor ] .
then @xmath872 is an inductive limit of a sequence , @xmath874 , of unital homogenous @xmath0-algebras with unital connecting maps , where @xmath875 the infinite tensor product @xmath876 is the inductive limit of the sequence @xmath877 with unital connecting maps .
it therefore suffices to show that @xmath878 for every @xmath8 , cf . proposition [ prop : inductive - limits ] .
now , @xmath879 is the inductive limit of the sequence @xmath880 with unital connecting mappings , and so , again by proposition [ prop : inductive - limits ] , it suffices to show that @xmath881 for every @xmath8 and @xmath10 .
the latter is precisely the content of lemma [ lm : inf - tensor ] .
in this section we show that our divisibility properties behave well with respect to taking direct products and ultrapowers of sequences of unital @xmath0-algebras .
this has the surprising consequence that such products and ultrapowers may admit characters even if all the @xmath0-algebras in the ingoing sequence are unital , simple and infinite dimensional .
we define the notion of almost characters " and show that the existence of such is related to the invariant @xmath882 .
it follows in particular that simple unital infinite dimensional @xmath0-algebras can have almost characters .
first we need some technical lemmas : [ lm : limit ] let @xmath4 be a unital @xmath0-algebra and let @xmath883 be an upward directed family of ideals of @xmath4 .
set @xmath884 .
it follows that @xmath885 @xmath886 for all positive integers @xmath8 .
the inequality @xmath887 " in all three cases follows from remark [ rem : image ] since we have a unital @xmath15-homomorphism @xmath888 for each @xmath889 .
we prove the reverse inequality
@xmath890 " only in the case of @xmath891 ; the proofs of the other two instances are similar .
set @xmath892 , and let us show that @xmath893 for some @xmath889 .
find @xmath894 in @xmath895 be such that @xmath896 for all @xmath75 .
find positive contractions @xmath897 in @xmath425 such that @xmath898 , where @xmath899 is the image of @xmath900 under the quotient mapping @xmath901 .
find @xmath98 such that @xmath902 for all @xmath75 .
it follows from ( * ? ? ?
* lemma 4.12 ) that there are positive elements @xmath903 in @xmath904 such that @xmath905 for all @xmath75 .
there is @xmath906 such that @xmath907 since @xmath908 is dense in @xmath451 , it follows that @xmath909 and @xmath910 all belong to @xmath911 for some @xmath889 .
let @xmath912 be the cuntz class of the image of the element @xmath913 under the quotient mapping @xmath914 .
then @xmath915 , whence @xmath893 .
for each @xmath98 , let @xmath916 $ ] be a continuous functions such that @xmath917 and @xmath918 when @xmath919 .
[ lm : r ] let @xmath4 be a unital @xmath0-algebra .
let @xmath920 be positive elements in @xmath4 such that @xmath921 .
then , for some @xmath98 , there are contractions @xmath922 in @xmath4 such that @xmath923 by assumption , and by compactness of @xmath275 , there are elements @xmath924 such that @xmath925 .
thus @xmath926 is invertible for some @xmath98 , and so there are elements @xmath927 such that @xmath928 . put @xmath929 and notice that @xmath930 for all @xmath75 .
thus @xmath931 which shows that the @xmath922 s are contractions with the desired properties .
[ lm : x ] let @xmath4 be a unital @xmath0-algebra , and let @xmath932 be positive integers . 1 .
@xmath4 is weakly @xmath14-divisible if and only if there exist positive contractions @xmath933 and contractions @xmath922 in @xmath4 , @xmath693 and @xmath934 , such that @xmath935 are pairwise equivalent and orthogonal for all @xmath75 , and such that @xmath936 .
@xmath4 is @xmath14-decomposable if and only if there exist pairwise orthogonal positive contractions @xmath112 and contractions @xmath315 in @xmath4 , @xmath693 and @xmath934 , such that @xmath937 for all @xmath54 .
@xmath4 is @xmath14-divisible if and only if there exist pairwise equivalent and pairwise orthogonal positive contractions @xmath112 and contractions @xmath922 in @xmath4 , @xmath693 and @xmath934 , such that @xmath938
. we view @xmath4 as being a sub-@xmath0-algebra of @xmath425 which is identified with the corner @xmath939 .
if " . put @xmath940 . then @xmath941 only if " .
choose @xmath942 such that @xmath943 .
use lemma [ lm : comp ] ( ii ) to find positive contractions @xmath944 in @xmath4 , @xmath693 and @xmath934 , such that @xmath945 are pairwise orthogonal and equivalent for all @xmath75 and such that @xmath946 .
it then follows that @xmath947 , and so it follows from lemma [ lm : r ] that there are @xmath98 and contractions @xmath922 in @xmath4 such that @xmath948 .
the contractions @xmath922 together with the positive contractions @xmath949 are then as desired .
if " . put @xmath950 .
then @xmath951 for all @xmath54 .
only if "
. choose @xmath952 such that @xmath953 .
use lemma [ lm : comp ] ( ii ) to find pairwise orthogonal and equivalent positive contractions @xmath435 in @xmath4 such that @xmath954 .
proceed as in the proof of only if " in part ( i ) to obtain elements @xmath955 ( for a suitable @xmath98 ) and @xmath315 with the desired properties . ( iii ) . if " . put @xmath956
. then @xmath957 only if " .
choose @xmath540 such that @xmath958 .
by lemma [ lm : comp ] ( ii ) there are pairwise orthogonal and pairwise equivalent positive contractions @xmath959 in @xmath4 such that @xmath960 .
we can now follow the proof of only if " in part ( i ) to obtain elements @xmath961 ( for a suitable @xmath98 ) and @xmath922 with the desired properties .
if @xmath962 is a sequence of @xmath0-algebras , then we denote by @xmath963 the @xmath0-algebra of all bounded sequences @xmath964 , with @xmath965 .
if @xmath404 is a ( free ) filter on @xmath690 , then denote by @xmath966 the closed two - sided ideal in @xmath963 consisting of those sequences @xmath964 for which @xmath967 .
finally , denote the quotient @xmath968 by @xmath969 .
[ prop : ultrapower ] let @xmath1 be a sequence of unital @xmath0-algebras . then , for all integers @xmath254 and for any free filer @xmath404 on @xmath690 we have : 1 .
2 . @xmath971 .
3 . @xmath972 .
we only prove ( i ) .
the proofs of ( ii ) and ( iii ) are very similar .
we have unital @xmath15-homomorphisms @xmath973 for all @xmath10 .
therefore the inequality @xmath890 " holds in the first identity in ( i ) ( and also in ( ii ) and ( iii ) ) , cf .
remark [ rem : image ] .
we show next that @xmath974 .
let @xmath10 be a positive integer such that @xmath975 for all @xmath170 .
then , by lemma [ lm : x ] ( i ) , for each @xmath170 we can find positive contractions @xmath976 and contractions @xmath977 in @xmath872 , for @xmath693 and @xmath934 , such that @xmath978 are pairwise orthogonal and equivalent for all @xmath75 , and such that @xmath979 put @xmath980 then @xmath935 are pairwise orthogonal and equivalent , and @xmath981 is equal to the unit of @xmath963 . by lemma [ lm : x ] ( i ) , this shows that @xmath963 is @xmath14-divisible , whence @xmath982 . to prove the second part of ( i ) note first that we have a natural unital ( surjective ) @xmath15-homomorphism @xmath983 for each @xmath984 .
we can therefore use remark [ rem : image ] and the first identity in ( i ) to conclude that @xmath985 which shows that @xmath986 .
we proceed to prove the reverse inequality . for
each @xmath984 consider the ideal @xmath987 in @xmath963 consisting of those sequences @xmath964 for which @xmath988 for all @xmath989 .
then @xmath990 ( where @xmath404 is ordered by reverse inclusion ) .
we can now use lemma [ lm : limit ] and the first identity in ( i ) to conclude that @xmath991 if we combine the proposition above with corollary [ cor : char ] ( i ) we obtain : [ cor : dv - char ] let @xmath962 be a sequence of unital @xmath0-algebras such that @xmath992 .
then @xmath963 has a character , and so does @xmath969 for each free filter @xmath404 on @xmath690 . if we combine the corollary above with theorem [ thm : simple ]
, then we obtain the following surprising fact : [ cor : simple - char ] there is a sequence @xmath962 of unital simple infinite dimensional @xmath0-algebras such that @xmath963 and @xmath969 have characters for each free filter @xmath404 on @xmath690 .
clearly , none of the @xmath0-algebras @xmath872 in the corollary above can have a character .
however , they have `` almost characters '' in the sense defined below .
this is one way of understanding how the product @xmath0-algebra can have a character when none of the individual @xmath0-algebras has one .
[ def : n - ep - char ] let @xmath993 be an integer and let @xmath98 .
a unital @xmath0-algebra @xmath4 is said to have _
@xmath994-characters _ if for every @xmath554-tuple @xmath995 of unitaries in @xmath4 there exists a state @xmath996 on @xmath4 such that @xmath997 for @xmath998 .
a state @xmath996 on a unital @xmath0-algebra is a character if and only if @xmath999 for all unitary elements @xmath1000 .
most simple @xmath0-algebras that we know of do not have @xmath1001-characters for small @xmath94 .
for example , if @xmath4 is a @xmath0-algebra which contains unitaries @xmath1002 such that @xmath1003 for some @xmath1004 and for some @xmath1005 , then @xmath4 does not admit any @xmath1001-character for some small enough @xmath98 . indeed ,
if @xmath996 is a state on @xmath4 such that @xmath1006 and @xmath1007 are close to @xmath603 , then @xmath1008 is close to @xmath603 .
[ prop : n - ep - char ] a unital @xmath0-algebra has a character if and only if it has @xmath994-characters for all pairs @xmath994 , where @xmath993 is an integer and @xmath98 . the only if " part is trivial .
assume that @xmath4 is a unital @xmath0-algebra that has @xmath994-characters for all pairs @xmath994 .
for each finite subset @xmath457 of the unitary group of @xmath4 and for each @xmath98 , let @xmath1009 denote the set of states @xmath996 on @xmath4 such that @xmath1010 for all @xmath1011 .
then , by assumption , @xmath1009 is non - empty .
it follows that @xmath1012 is non - empty , and any state in this intersection is a character .
[ prop : prod - char ] let @xmath962 be a sequence of unital @xmath0-algebras , and let @xmath404 be a free ultrafilter on @xmath690 .
then @xmath969 has a character if and only if for each integer @xmath993 and for each @xmath98 there exists @xmath984 such that @xmath872 has @xmath994-characters for each @xmath989 .
we prove first the if " part . by proposition [ prop : n - ep - char ]
it suffices to show that @xmath969 has @xmath994-characters for all @xmath994 .
fix @xmath994 and find @xmath984 such that @xmath872 has @xmath994-characters for each @xmath989 .
let @xmath1013 be unitaries in @xmath969 , and let @xmath1014 be a lift of @xmath1015 . then for each @xmath989 there is a state @xmath1016 on @xmath872 such that @xmath1017 for @xmath998 .
choose arbitrary states @xmath1016 on @xmath872 for @xmath1018 and define a state @xmath996 on @xmath969 by @xmath1019 , where @xmath1020 is a lift of @xmath27 .
( a priori , @xmath996 defines a state on @xmath963 , and one checks that it vanishes on the ideal @xmath966 .
) then @xmath1021 for @xmath998 , which shows that @xmath969 has @xmath994-characters .
suppose next that @xmath969 has a character @xmath996 .
fix @xmath994 , and let @xmath1022 be the set of those @xmath1023 for which @xmath872 does not have @xmath994-characters . for each @xmath1024 choose unitaries @xmath1025 in @xmath872 , @xmath998 , such that there is no state @xmath1026 on @xmath872 for which @xmath1027 for all @xmath998 .
choose arbitrary unitaries @xmath1028 for @xmath1029 , and let @xmath1015 be the unitary element @xmath1030 in @xmath963 .
let @xmath25 be the ( separable ) sub-@xmath0-algebra of @xmath969 generated by the unitaries @xmath1031 , where @xmath1032 is the quotient mapping @xmath1033 . by ( * ?
* lemma 2.5 ) there is a sequence @xmath1016 of pure states on @xmath872 such that @xmath1034 for all @xmath1035 with @xmath1036 . now , @xmath1037 it follows that there exists @xmath984 such that @xmath1038 for all @xmath989 and for all @xmath998 .
this entails that @xmath1039 .
hence @xmath872 has @xmath994-characters for all @xmath989 .
we can relate the existence of @xmath994-characters on a @xmath0-algebra @xmath4 to the divisibility quantity @xmath1040 . for each pair
@xmath994 , where @xmath993 is an integer and @xmath94 , there exists an integer @xmath378 such that every unital@xmath0-algebra @xmath4 which satisfies @xmath1041 has @xmath994-characters .
conversely , for every integer @xmath378 there exists a pair @xmath994 , where @xmath993 is an integer and where @xmath98 , such that every unital @xmath0-algebra @xmath4 which has @xmath994-characters satisfies @xmath1041 .
suppose that the first claim were false .
then there would exist a pair @xmath994 and a sequence @xmath1 of unital @xmath0-algebras such that @xmath1042 and none of the @xmath33 s have @xmath994-characters .
however , if @xmath404 is any free ultrafilter @xmath690 , then @xmath1043 has a character by corollary [ cor : dv - char ] , whence @xmath33 has @xmath994-characters for each @xmath10 in some subset @xmath984 , a contradiction .
suppose next that the second statement were false .
then there would exist an integer @xmath378 and a sequence @xmath962 of unital @xmath0-algebras such that @xmath872 has @xmath1044-characters but @xmath1045 . let @xmath404 be a free ultrafilter on @xmath690 .
it then follows from proposition [ prop : prod - char ] that @xmath969 has a character
. hence @xmath1046 , whence @xmath1047 by proposition [ prop : ultrapower ] , a contradiction . for each pair @xmath994 , where @xmath993 is an integer and @xmath94
, there exists a unital simple infinite dimensional @xmath0-algebra which has @xmath994-characters .
we end this section by giving several equivalent formulation of some well - known open problems for @xmath0-algebras .
recall that a @xmath0-algebra @xmath4 has the global glimm halving property if there is a @xmath15-homomorphism @xmath481 whose image is full in @xmath4 .
[ prop : ggh3 ] the following statements are equivalent : 1 . every unital @xmath0-algebra that has no finite dimensional representation has the global glimm halving property .
2 . for all unital @xmath0-algebras
@xmath4 , if @xmath475 for all @xmath254 , then @xmath1048 .
3 . for every sequence @xmath962 of unital @xmath0-algebras ,
if @xmath1049 for all @xmath8 , then @xmath1050 .
\(i ) @xmath1051 ( ii ) .
@xmath4 has no finite dimensional representations if and only if @xmath454 for all @xmath8 , which by corollary [ cor : finite - div ] ( iii ) is equivalent to @xmath475 for all @xmath8 .
it was shown in corollary [ cor : finite - div ] ( i ) that the global glimm halving property holds for @xmath4 if and only if @xmath1048 .
\(ii ) @xmath121 ( iii ) . given a sequence @xmath962 of unital @xmath0-algebras such that @xmath1049 for all @xmath8 .
consider the @xmath0-algebra @xmath1052 .
then @xmath1053 by proposition [ prop : ultrapower ] .
thus @xmath1048 , which implies that @xmath1054 , again by proposition [ prop : ultrapower ] .
\(iii ) @xmath121 ( ii ) is trivial : take @xmath1055 for all @xmath170 .
[ prop : ggh4 ] the following statements are equivalent : 1 .
all unital @xmath0-algebras @xmath4 that have no finite dimensional representation contain two positive full elements that are orthogonal to each other .
2 . for all unital @xmath0-algebras
@xmath4 , if @xmath475 for all @xmath254 , then @xmath1056 .
\(i ) @xmath1051 ( ii ) . as in the proof of proposition [ prop : ggh3 ]
, @xmath4 has no finite dimensional representations if and only if @xmath475 for all @xmath8 .
it was shown in corollary [ cor : finite - div ] ( ii ) that @xmath4 contains two positive full elements that are orthogonal to each other if and only if @xmath1056 .
\(ii ) @xmath121 ( iii ) is similar to the proof of ( ii ) @xmath121 ( iii ) in proposition [ prop : ggh3 ] .
( iii ) @xmath121 ( ii ) is trivial .
following @xcite , a cuntz class @xmath70 in the cuntz semigroup of a @xmath0-algebra @xmath4 is said to be properly infinite if it satisfies @xmath1057 ( whence @xmath1058 ) .
similarly , a countably generated hilbert module over @xmath4 is properly infinite if its cuntz class is properly infinite .
we saw in proposition [ prop : propinf ] how infiniteness of a cuntz class can arise from a certain type of divisibility property . in this section
we shall investigate this and related phenomena further with emphasis on the following property : [ def : omega - n ] let @xmath4 be a @xmath0-algebra , let @xmath313 be an integer , and let @xmath70 be an element in @xmath19 .
we say that @xmath70 is @xmath1059-decomposable if there exist @xmath1060 such that @xmath1061 and @xmath1062 for all @xmath54 . 1 .
@xmath70 is @xmath1059-decomposable , 2 .
there exist @xmath1060 such that @xmath1061 and @xmath1063 for all @xmath54 , 3
. there exist @xmath1060 and @xmath1064 such that @xmath1061 , @xmath1065 for all @xmath54 , and @xmath1066 , 4 .
there exist @xmath1060 such that @xmath1061 and @xmath1067 for all @xmath170 .
\(i ) @xmath121 ( ii ) .
suppose that @xmath1068 satisfy @xmath1069 .
let @xmath1070 be a partition of the natural numbers into infinite sets .
then the elements @xmath1071 witness that condition ( ii ) holds .
( iii)@xmath121 ( iv ) .
let @xmath1073 and @xmath1074 be as in ( iii ) .
then @xmath1075 .
observe that the left side of this inequality decreases as @xmath170 increases .
thus , @xmath1076 for all @xmath1077 . taking the supremum over all
@xmath1077 we get @xmath1078 .
\(iv ) @xmath121 ( i ) .
suppose that @xmath1068 satisfy the condition in ( iv ) .
let @xmath69 be such that @xmath44 for all @xmath54 and @xmath45 .
then there exists a sequence @xmath1079 such that the elements @xmath1080 satisfy @xmath1081 for all @xmath54 .
let @xmath1070 be a partition of the natural numbers into infinite sets .
then the elements @xmath1082 satisfy the condition in definition [ def : omega - n ] .
it was shown in @xcite that the corona factorization property for a @xmath0-algebra is equivalent to a condition for its cuntz semigroup , that we here shall refer to as ( cfp4s ) .
a complete ordered abelian semigroup is said to have ( cfp4s ) if whenever @xmath1073 is a full sequence , @xmath1083 , and @xmath1074 is such that @xmath1084 for all @xmath54 and some @xmath8 , then there exists @xmath10 such that @xmath1085 . recall that a full sequence is one that is increasing and such that @xmath1086 is a full element , i.e. , @xmath1087 is the largest element of the semigroup ( which we shall denote by @xmath350 ) .
the proposition below relates the ( cfp4s ) with the notion of @xmath1088-divisibility .
in fact , it is a consequence of this proposition that a semigroup in the category @xmath1089 has a full elements which is @xmath1088-divisible and not properly infinite if and only if the semigroup does not satisfy ( cfp4s ) . 1
. @xmath41 has property ( cfp4s ) .
2 . for every sequence @xmath1074 in @xmath41 , if there is a full sequence @xmath1073 in @xmath41 such that @xmath1090 for some @xmath8 and for all @xmath54 , then @xmath1091 .
3 . for every sequence @xmath1074 in @xmath41 ,
if @xmath1092 for some @xmath8 and for all @xmath54 , then @xmath1093 .
4 . for every sequence @xmath1074 in @xmath41 ,
if @xmath1094 for some @xmath8 and for all @xmath10 , then @xmath1093
. 5 . for every full element @xmath28 in @xmath41 , if @xmath28 is @xmath1088-decomposable for some @xmath8 , then @xmath28 is properly infinite ( whence @xmath1095 ) .
\(ii ) @xmath121 ( i ) . if @xmath1101 is a full sequence and if @xmath1102 is another sequence such that @xmath1103 , then @xmath1091 by ( ii ) .
in particular , if @xmath1104 , then @xmath1105 for some @xmath10 by the definition of compact containment .
\(iii ) @xmath121 ( v ) .
suppose that ( iii ) holds , and that @xmath1109 is full and @xmath1088-decomposable
. then @xmath28 satisfies condition ( ii ) of lemma [ rem : indexpartition ] , so there exists a sequence @xmath1102 such that @xmath1110 and @xmath1111 for all @xmath54 . but then @xmath1091 because ( iii ) holds , whence @xmath1112 .
\(v ) @xmath121 ( iv ) .
suppose that ( v ) holds and let @xmath1074 be a sequence in @xmath41 such that @xmath1094 for some @xmath8 and for all @xmath10 .
put @xmath1113 .
we must show that @xmath1095 .
we know that @xmath1114 , so @xmath28 is full .
it is easy to see that @xmath28 satisfies condition ( iv ) of lemma [ rem : indexpartition ] , so @xmath28 is @xmath1088-decomposable .
hence @xmath1115 by the assumption that ( v ) holds .
\(iv ) @xmath121 ( ii ) .
let @xmath1073 be a full sequence in @xmath41 , let @xmath198 be a positive integer , and let @xmath1102 be such that @xmath1084 for all @xmath54 .
then @xmath1116 for all @xmath1117 . as @xmath1118 , we conclude that @xmath1094 for all @xmath10 . by
( iv ) this entails that @xmath1093 , and in particular that @xmath1119 . in the following example we describe a cuntz semigroup with an element @xmath70 that is @xmath1120-decomposable but not properly infinite .
in particular , @xmath1121 is properly infinite while @xmath70 is not .
this example is well known in other contexts , and it was discussed in the paragraph preceding corollary [ cor : key ] .
[ ex : infty2decomposable ] let @xmath1122 be a countable cartesian product of 2-dimensional spheres , and let @xmath1123 be the one - dimensional projection arising as the pull back of a non - trivial rank one projection @xmath280 in @xmath1124 along the @xmath54th coordinate projection @xmath1125 , cf . the comments above corollary [ cor : key ] .
let @xmath413 be a trivial one - dimensional projection .
then @xmath1126 for all @xmath554 , because the euler class of the projection on the right - hand side is non - trivial .
[ prop : tensorstability ] let @xmath1134 be a sequence of mutually orthogonal positive elements in a @xmath0-algebra @xmath4 such that @xmath1135 converges to a strictly positive element in @xmath4 .
assume that @xmath1136 for all @xmath54 .
then @xmath1137 is stable for every @xmath240-unital @xmath0-algebra @xmath25 with @xmath1138 .
set @xmath1139 and let @xmath85 be a strictly positive element in @xmath25 .
notice that @xmath1140 is a strictly positive element of @xmath1137 . in order to prove stability of @xmath1137
we will use the stability criterion obtained in @xcite : @xmath1137 is stable if for every @xmath94 there exists a positive element @xmath1141 in @xmath4 which is orthogonal to @xmath1142 and satisfies @xmath1143 .
arguing as in the proof of ( iv ) @xmath121 ( i ) in lemma [ rem : indexpartition ] we may assume that @xmath1144 for all @xmath54 . by theorem [ thm : finite - div ] ( iii )
, @xmath1138 is equivalent to weak @xmath1145-divisibility for @xmath1146 . thus there
exist a sequence @xmath1073 in @xmath316 such that @xmath1147 for all @xmath54 and @xmath1148 .
we can form a new sequence @xmath1149 in which each @xmath314 appears repeated infinitely often . in this way we may assume without loss of generality that @xmath1150 for all @xmath75 .
find positive elements @xmath418 in @xmath25 such that @xmath1151 and @xmath1152 .
then @xmath1153 for some integer @xmath543 .
set @xmath1154 .
then @xmath1141 is orthogonal to @xmath1142 . also , @xmath1155 for each @xmath54 .
hence @xmath1156 thus , @xmath1157 .
this shows that @xmath1137 is stable .
the proposition above can be applied to the @xmath0-algebra @xmath1158 arising from example [ ex : infty2decomposable ] with @xmath1159 .
the @xmath0-algebra @xmath4 is not stable ( because @xmath1160 while @xmath1161 ) , but @xmath527 is stable for every @xmath0-algebra @xmath25 that does not have a character by proposition [ prop : tensorstability ] and example [ ex : infty2decomposable ] . the example obtained in @xcite of a simple @xmath0-algebra @xmath4 of stable rank 1 such that @xmath380 is stable , while @xmath1162 is not stable , likewise satisfies the hypotheses of proposition [ prop : tensorstability ] .
in fact , to the authors knowledge , every example of a @xmath0-algebra that tensored with @xmath1163 becomes stable also has the stronger property of becoming stable after being tensored with any @xmath0-algebra that has no representations of dimension less than @xmath10 .
this raises the following question : 1 .
@xmath37 is @xmath1059-decomposable .
@xmath4 contains a full hereditary subalgebra @xmath25 such that @xmath1165 is stable for any @xmath0-algebra @xmath1166 such that @xmath1167 .
3 . @xmath4 contains a full hereditary subalgebra @xmath25 such that @xmath1168 is stable .
\(i ) @xmath121 ( ii ) .
let @xmath1073 be such that @xmath1169 and @xmath1170 .
let @xmath1171 be strictly positive and let @xmath1172 be mutually orthogonal elements such that @xmath1173 .
we can find mutually orthogonal positive elements @xmath112 in @xmath4 such that @xmath1174 , @xmath1175 , and such that @xmath1176 is convergent .
it then follows from proposition [ prop : tensorstability ] that ( ii ) holds when @xmath25 is the hereditary sub-@xmath0-algebra generated by @xmath1176 .
use ( * ? ? ?
* lemma 5.3 ) to find a sequence @xmath1177 of pairwise orthogonal positive elements in @xmath25 such that @xmath1178 for all @xmath170 . then condition ( iii ) of lemma [ rem : indexpartition ] is satisfied with @xmath1179 , @xmath1180 , and @xmath1181 , whence @xmath1182 is @xmath1059-decomposable
. we will next give an example of a cuntz semigroup element that is weakly @xmath1120-divisible but not properly infinite .
this example needs some preparatory results .
let us first recall an example given by dixmier and doaudy in @xcite .
[ ex : dixdou ] let @xmath1190 denote the closed unit ball of @xmath1191 endowed with the weak topology .
let @xmath1192 denote the @xmath1193-hilbert module of continuous maps from @xmath1194 to @xmath1191 .
we will construct a countably generated @xmath1193-hilbert module @xmath252 such that @xmath1195 but @xmath1196 .
let @xmath1197 be given by @xmath1198 consider the @xmath1193-module @xmath1199 of functions from @xmath1194 to @xmath1200 that have the form @xmath1201 , with @xmath1202 and @xmath1203 .
the module @xmath1199 is a pre - hilbert c*-module over @xmath1193 when endowed with the pointwise inner product .
indeed , if @xmath1204 and @xmath1205 are vectors in @xmath1199 then @xmath1206 let @xmath252 denote the completion of @xmath1199 with respect to the norm induced by its @xmath1193-valued inner product .
observe that @xmath1207 .
since @xmath252 is countably generated , we also have that @xmath1208 by kasparov s stabilization theorem .
let us see that @xmath1196 .
consider @xmath1209 , the orthogonal complement of @xmath1210 .
then @xmath1211 , where @xmath1212 it was implicitly shown by dixmier and douady , and explicitly pointed out by blanchard and kirchberg ( ( * ? ? ?
* proposition 3.6 ) ) , that for any @xmath1213 there exists @xmath1214 such that @xmath1215 .
that is , every section of @xmath1216 vanishes at some point ( we will reprove this fact in proposition [ prop : sectionsvanish ] below ) . notice that @xmath1217 it can be deduced from this that @xmath1196 ( see ( * ? ? ?
* proposition 19 ) ) .
let @xmath1218 denote the unit ball in @xmath1219 .
let @xmath1220 be defined as @xmath1221 ( the function @xmath735 is a homeomorphism from @xmath1218 to the set of positive elements of @xmath1222 with trace 1 . on the boundary 2-sphere of @xmath1218 it agrees with the tautological rank 1 projection .
) consider the @xmath1223-module associated to @xmath735 : @xmath1224 [ prop : sectionsvanish ] let @xmath1194 and @xmath1218 be as before .
let @xmath1225 , where each @xmath1226 is either @xmath1194 or @xmath1218 and the index set @xmath451 is non - empty . for each @xmath54 ,
let @xmath1227 be the pull - back along the projection map @xmath1228 of either the module @xmath1216 defined in example [ ex : dixdou ] or the module @xmath457 defined in .
finally , let @xmath38 be the @xmath604-module defined by @xmath1229 .
then @xmath604 does not embed in @xmath38 as a @xmath604-module ( i.e. , for every @xmath1230 there exists @xmath1231 such that @xmath1215 ) . notice that if every @xmath1226 agrees with @xmath1218 , the above proposition can be proven using standard methods in algebraic topology ( e.g. , characteristic classes ) .
indeed , it suffices to restrict to the boundary 2-sphere of each @xmath1226 and use that on that set @xmath457 is the tautological rank 1 projective module .
it is the inclusion of the spaces @xmath1194 in the definition of @xmath599 that forces us to use a different route in the proof .
let @xmath1230 , and write @xmath1232 , with @xmath1233 .
in order to show @xmath1234 for some @xmath1231 , it suffices to prove this for @xmath590 belonging to a dense submodule of @xmath38 . for suppose that @xmath1235 is a sequence in @xmath38 such that @xmath1236 and @xmath1237 for some @xmath1238 . then by the compactness of @xmath599 there exists a subsequence @xmath1239 such that @xmath1240 , and so @xmath1215 .
thus , we may assume that the index set @xmath451 is finite .
furthermore , for the indices @xmath54 such that @xmath1241 , we may assume that @xmath1242 , where @xmath1243 is the pull back along @xmath1244 of the dense submodule @xmath1245 defined in .
we will argue by contradiction that @xmath1234 for some @xmath1231 .
suppose that @xmath1252 is invertible , and assume without loss of generality that @xmath1253 .
observe that , for each @xmath1254 , @xmath1255 is a function from @xmath599 into the unit ball of @xmath1256 , while for @xmath1257 the entry @xmath1255 is a function from @xmath599 into the unit ball of @xmath1258 ( let us denote it by @xmath1259 ) .
let @xmath1260 denote the projection onto the first direct summand and let @xmath1261 denote the hopf fibration ( extended to the unit ball ) : @xmath1262 let @xmath1263\to [ 0,1]$ ] be such that @xmath1264 , @xmath1265 for @xmath1266 $ ] , and @xmath889 is linear in @xmath1267 $ ] .
define @xmath1268 by @xmath1269 consider the continuous map @xmath1270 given by the vector of functions @xmath1271 since @xmath599 is a compact convex subset of the vector space @xmath1272 , the map @xmath1273 has a fixed point by the schauder fixed point theorem .
let @xmath1274 be a fixed point of @xmath1273 .
since @xmath1275 , we must have @xmath1276 for at least one index @xmath54 .
notice that both @xmath1277 and @xmath1278 map all vectors of norm at least @xmath1279 into the unit sphere of either @xmath1194 or @xmath1218 .
it follows that the fixed point @xmath1280 satisfies @xmath1281 and @xmath1282 for all @xmath1283 .
there are two cases to consider : @xmath1254 and @xmath1284 .
suppose that @xmath1254 .
the general form of @xmath1242 is @xmath1285 , for some @xmath1286 and @xmath1287 . since @xmath1281
, we have @xmath1288 but @xmath1289 .
this contradicts that @xmath1281 .
suppose that @xmath1290 .
since @xmath1291 , with @xmath1292 , is a section of the tautological bundle on @xmath620 , we have @xmath1293 whenever @xmath1292 .
it follows that @xmath1294 . but
@xmath1295 , by the fixed point property of @xmath1280 .
this again contradicts that @xmath1281 .
let @xmath1296 and consider the hilbert module @xmath1297 , described in the statement of the previous proposition .
we have shown that @xmath1298\nleq [ h]$ ] .
in particular , @xmath39 $ ] is not properly infinite ( since it is full ) .
let us show that @xmath39 $ ] is weakly @xmath1120-divisible .
consider the open sets @xmath1299 and @xmath1300 , where @xmath1301 and @xmath1302 are ( open ) upper and lower hemispheres of @xmath1218 that together cover @xmath1218 .
we claim that @xmath1303 and @xmath1304 .
indeed , @xmath1305 but @xmath1306 . therefore , @xmath1307 but @xmath1308 .
thus , @xmath1303 .
symmetrically , we have that @xmath1304 .
it follows that @xmath1309 $ ] and @xmath1310 $ ] are properly infinite , and @xmath39\leq [ hc_0(u)]+[hc_0(v)]$ ] .
thus , @xmath39 $ ] is weakly @xmath1120-divisible .
the previous example answers a question posed in ( * ? ? ?
* question 3.10 ) : if @xmath36 and @xmath85 are properly infinite positive elements , is @xmath1311 properly infinite ? in the language of hilbert modules , this question asks whether @xmath38 is properly infinite if @xmath1312 , and @xmath1313 are properly infinite submodules of @xmath38 .
we obtain a counterexample taking @xmath38 as in the previous example , @xmath1314 and @xmath1315 .
in this example we answer ( in the negative ) the following question , posed in ( * ? ? ?
* question 3.4 ) : if @xmath39 $ ] is properly infinite , is the unit of @xmath1316 a properly infinite projection ?
let @xmath1317 and consider the hilbert @xmath604-module @xmath1318 the module @xmath1319 is the pull back along @xmath1320 of the dixmier - douady module @xmath252 .
since @xmath1321 embeds in @xmath252 , @xmath1322 embeds in @xmath1319 .
thus , @xmath39 $ ] is properly infinite .
also , the direct sum of the module @xmath1323 with itself gives @xmath1322 ( because @xmath1324 contains @xmath1223 as a direct summand ) .
therefore , @xmath1325 .
however , @xmath38 is not isomorphic to @xmath1322 , because every section of @xmath1326 vanishes , and so adding the trivial rank 1 module to it can not yield the trivial hilbert module @xmath1322 ( see the proof of @xmath1327 in ( * ? ? ?
* proposition 19 ) ) .
it follows that @xmath1328 is not a direct summand of @xmath38 , i.e. , the unit of @xmath1316 is not properly infinite . | we consider three notions of divisibility in the cuntz semigroup of a @xmath0-algebra , and show how they reflect properties of the @xmath0-algebra .
we develop methods to construct ( simple and non - simple ) @xmath0-algebras with specific divisibility behaviour . as a byproduct of our investigations , we show that there exists a sequence @xmath1 of simple unital infinite dimensional @xmath0-algebras such that the product @xmath2 has a character . |
according to classical electrodynamics an accelerated charge emits radiation .
if the acceleration is due to an incident laser field this process of emission may be called inelastic scattering of the incident laser field by the charge ( e.g. , an electron ) . this process has been thoroughly investigated and is called either thomson scattering if quantum effects are negligible @xcite or compton scattering if quantum effects are important @xcite . throughout all of the cited works the incident laser field is taken as a plane - wave field with a peak electric field @xmath0 and carrier angular frequency @xmath1 , and its intensity is characterized by the relativistically invariant parameter @xmath2 here @xmath3 is the electron s charge , @xmath4 is its mass , and @xmath5 is the speed of light .
a pulse in this work is indicated as _ intense _ , if the parameter @xmath6 is much larger than unity .
an electron in such an intense laser field is accelerated to relativistic velocities already in one laser period @xcite .
the electric field strengths needed to attain a parameter @xmath6 equal to unity for optical radiation and for x - rays would be @xmath7 and @xmath8 , respectively . here
@xmath9 is the usual planck constant .
these field strengths correspond to laser intensities of @xmath10 in the optical regime ( @xmath11 ev ) , laser intensities of these orders have already been obtained during the last decade @xcite . among others ,
these laser systems have been employed to find extensive experimental proof of nonlinear thomson scattering @xcite . in these experiments ,
laser systems were employed , reaching nonlinearity parameters ( this choice of language is explained below ) of the order of up to @xmath12 . due to lack of sufficiently intense laser systems so far , nonlinear compton scattering in a laser field has been verified in only one experimental setup @xcite , but with the development of laser systems to ever higher peak intensities , more experimental tests for nonlinear compton scattering seem to be in reach .
the record optical intensity of @xmath13 , obtained in 2008 , corresponds to a parameter @xmath6 of the order of @xmath14 @xcite .
the parameter @xmath6 is often referred to as the nonlinearity parameter @xcite .
in fact , for @xmath15 , the interaction of an electron with the laser is no longer linearly dependent on the laser intensity .
this can be understood classically by observing that at electron velocities inside the laser field close to @xmath5 , the magnetic force on the electron is of comparable strength as the electric force , and the total interaction is no longer linear in the external field because the velocity also depends on the field .
another physical interpretation for the nonlinear dependence of the scattering rates on the laser intensity can be given in the photon picture of the laser field .
in fact , the intensity of a radiation field is connected to the photon number density .
for a not too intense laser field , an electron will basically always scatter only one photon from it . on the other hand ,
if the incident radiation is very intense , that is , its photon flux is very high , the electron likely interacts with many photons from the laser field . thus the scattering rate will no longer depend linearly on the laser intensity but will exhibit a more complex dependency . in this picture
, the parameter @xmath6 gives the ratio of energy absorbed by the electron @xmath16 in one compton wavelength @xmath17 in units of the incident laser photon energy @xmath18 . in this sense , if @xmath19 , the electron on average absorbs more than one photon from the laser field during the process , which again yields nonlinear effects .
so in this work , the terms _ multiphoton _ and _ nonlinear _ compton scattering are used interchangeably .
the photon flux in state - of - the - art laser facilities may now become so high that the electron absorbs a large number of photons , resulting in the emission of a single high - energetic photon .
if the energy of this photon is of the order of the electron s energy , then the recoil effect on the electron motion has to be taken into account .
this is an intrinsically quantum effect .
the strength of nonlinear quantum effects is described by the dimensionless parameter @xmath20 , where @xmath21 is the electron s initial four - momentum , where @xmath22 is the electromagnetic field strength tensor [ @xmath23 is the laser photon four - momentum and @xmath24 is the laser four - potential strength , with @xmath25 and the metric @xmath26 is used in this article .
it is always possible to consider the scattering of an electron by a plane wave in a reference frame in which the electron and the laser field are initially counterpropagating , and in that frame the expression of @xmath27 simplifies to @xmath28 where for two four - vectors @xmath29 and @xmath30 we introduced the notation @xmath31 , and @xmath32 is the critical field of qed .
this allows the interpretation of @xmath27 as the laser s electric field amplitude evaluated in the reference frame in which the electron initially is at rest , in units of @xmath33 .
the critical field transfers an energy @xmath34 to an electron over one electron compton wavelength . creating an electric field of that amplitude would demand a laser intensity of @xmath35 all laser fields available today fall short of reaching the critical field strength by at least 3 or 4 orders of magnitude .
the physical relevance of this quantity is that in a constant and uniform electric field of the critical field strength , the probability of creating electron - positron pairs from a vacuum becomes non negligible @xcite .
so if the parameter @xmath27 approaches unity , the electron will feel an electric field strength at which there are nonlinear qed effects expected to happen .
it has been pointed out that for small intensity parameters @xmath36 , quantum effects scale with the laser photon energy ( in the initial rest frame of the electron ) , that is , with the parameter @xmath37 while for the opposite case @xmath38 , they scale with the parameter @xmath27 , that is , rather with the electric field strength @xcite . so in the case @xmath38 , which we put particular emphasis on , we expect for @xmath39 , quantum effects to become important .
most of the theoretical works done so far on nonlinear compton scattering considered a monochromatic laser wave @xcite .
in fact , there has been some work on electron scattering from a laser pulse of duration @xmath40 and frequency @xmath1 @xcite , but there the authors considered a pulse fulfilling the condition @xmath41 that is , a pulse containing many cycles of the carrier field .
however , in order to generate high laser peak intensities and correspondingly high @xmath6 parameters , laser pulses are compressed spatially as well as temporally .
spatial compression is usually neglected in theoretical works , assuming to deal with laser beams which are not tightly focused and which can be safely approximated by a plane wave .
this is also the case of the present article .
concerning temporal compression , there have been works on nonlinear compton scattering not relying on the condition ( [ longpulsecond ] ) @xcite .
the resulting structure of the general scattering matrix element obtained by the authors agrees with the one found in this article . on the other hand , in @xcite ,
the authors perform a more exploratory analysis of the scattering process in the regime @xmath42 , while in this article we present a detailed analytical analysis of high - intensity ( @xmath43 ) nonlinear compton scattering . in this work , next to the analytical part
, particular emphasis is put on the effect on the spectra of varying the laser intensity and the incident electron energy .
also , in @xcite , the authors consider effects of the absolute phase [ the so - called carrier envelope phase ( cep ) ] of few - cycle pulses and its impact on the emitted photon energy spectra . here
we also consider the effect of cep on the angular distribution of the emitted radiation ( see also @xcite ) .
laser pulses lasting only one or two cycles , as investigated in the present work , have become available in different frequency ranges such as in the mid - infrared @xcite , in the near - infrared @xcite , in the optical @xcite , and in the extreme ultraviolet regimes @xcite .
moreover , we point out that all high - field laser facilities operating or under construction employ short pulse durations to generate high field strengths .
for instance , the petawatt field synthesizer ( pfs ) laser system under construction in garching ( germany ) aims at optical laser intensities of the order of @xmath44 ( @xmath45 ) by compressing an energy of @xmath46 to only @xmath47 , corresponding to less than two laser cycles @xcite . at the extreme light infrastructure @xcite and at the high power laser energy research @xcite facilities , which aim at unprecedented laser intensities of the order of @xmath48 , pulse durations of about @xmath49
are envisaged .
this shows the close linking between the generation of large values of @xmath6 and short pulse durations . additionally ,
since all high - field facilities referenced here operate at optical wavelengths , we will also focus on the energy regime @xmath50 ev for the incident laser .
put in quantitative terms , short pulses containing one or only a few cycles of the electric field will be distinguished by the condition @xmath51 in contrast to eq .
( [ longpulsecond ] ) . for optical lasers
, this corresponds to pulse durations on the order of @xmath52 fs .
we will label laser pulses fulfilling the condition ( [ shortpulsecond ] ) as ultrashort .
as it will turn out later , the connection between few - cycle pulses and nonlinear compton scattering is twofold .
not only does one have to incorporate eq .
( [ shortpulsecond ] ) into the framework of compton scattering , but also , nonlinear compton scattering offers a thorough way of determining the precise temporal shape of few - cycle laser pulses @xcite . the main purpose of this article , however , is to investigate nonlinear compton scattering in ultrashort pulses in the framework of strong - field qed . a visualization of the scattering process we are going to consider in the language of feynman diagrams looks like as shown on the left - hand side of fig .
[ feyngraph ] .
the electron enters the laser field with an incoming four - momentum @xmath53 . during the interaction with the laser pulse ,
it may absorb or reemit @xmath54 photons from or into the laser s photon field , all sharing the same wave vector @xmath23 , and at some point , the electron emits a single final photon with wave vector @xmath55 . in fig .
[ feyngraph ] , @xmath54 may be an arbitrarily large natural number ( the case @xmath56 is absent because energy - momentum conservation can not be fulfilled ) , what we represent by the dots inserted into the electron line . after the scattering , the electron will propagate with a changed four - momentum which , after exiting the laser field , is given by @xmath57 . in order to take into account exactly the ultra strong laser field , which can not be treated perturbatively
, one would have to sum over all @xmath54 in the process depicted on the left - hand side of fig .
[ feyngraph ] . instead of doing so , we perform the calculations in the furry picture of quantum dynamics , intrinsically taking the external field into account exactly . here
the electron states employed for computing the process amplitude are obtained by solving the dirac equation in the plane - wave field .
the resulting states are known as volkov states , and they are indicated as a double solid line in the right - hand side of fig .
[ feyngraph ] .
we will restrict to values of the parameter @xmath27 smaller or of the order of 1 .
therefore the main contribution to radiation comes from the diagram on the right - hand side of fig .
[ feyngraph ] , and two - photon emission is roughly @xmath58 times smaller than the process considered here @xcite . the article is organized as follows : in section [ theosec ] , we are going to work out the theory of the considered process .
we will find that the scattering amplitude is determined by three integrals of functions of the laser pulse and of the electron parameters .
these integrals are explicitly evaluated in different parameters regimes . in section [ resultssec ] ,
we present emission spectra of an electron scattered from two model pulses , which we choose to model different temporal shapes of a single - cycle pulse .
finally , the summarizing discussion of section [ conclusionssec ] concludes the article .
units with @xmath59 are used below .
the multiphoton compton scattering process diagrammatically shown in fig .
[ feyngraph ] is described by the matrix element @xmath60 here @xmath61 are the usual dirac matrices , the functions @xmath62 and @xmath63 are the spinor wave functions of the electron in the background plane - wave field and its dirac conjugate , respectively , @xmath64 is a normalization volume and @xmath65 is the imaginary unit .
the electron has four - momenta @xmath66 and @xmath67 before and after scattering , respectively .
the four - vector @xmath68 gives the emitted photon s polarization , while @xmath69 is its four - momentum .
the states @xmath62 are found as solutions of the dirac equation @xmath70-m\right\ } \psi_{p\sigma } = 0,\ ] ] where the vector potential @xmath71 describes the background plane wave and depends only on the phase @xmath72 , with @xmath23 being the wave vector of the laser field .
the solutions of the above equation were found by volkov already in 1935 @xcite and can be found , for example , in @xcite .
@xmath73\,u_{p\sigma}\ , { \text{e}}^{{\mathbbm{i}}\ , s } , \label{volkov}\end{aligned}\ ] ] with the classical action @xmath74}{(kp)}-\frac{e^2 a^2(\phi')}{2\ , ( kp)}\right]\ ] ] and a free electron spinor @xmath75 .
the feynman slash notation @xmath76 is used as throughout this article . for the vector potential @xmath71
we choose the gauge in which @xmath77 with @xmath78 , where the shape function @xmath79 gives the temporal shape of the pulse and @xmath80 is a unit vector pointing along the laser s polarization axis .
the electric field of the wave has an amplitude @xmath81 and can be written as @xmath82 , with the electric field s shape function @xmath83 .
we are going to investigate the scattering process in a coordinate frame where the laser pulse propagates along the positive @xmath84 axis , that is , is described by the wave vector @xmath85 , and where the electron before the scattering process propagates along the negative @xmath84 axis with the initial four - momentum @xmath86 .
the laser is modeled to be linearly polarized along the @xmath87 direction .
this choice of coordinates is visualized in fig .
[ coordsys ] . whenever a particular shape function is needed to obtain definite results for emission spectra , we will model the laser s four - potential by a specific choice .
one possible choice used in this work is @xmath88 corresponding to the electric field s shape function ( see fig . [ sechfieldpic ] ) @xmath89 ) @xmath90 ( solid red line ) and @xmath91 ( dashed blue line ) . ] + the former choice models a single - cycle pulse of the electric field corresponding in the optical domain to a pulse duration of roughly @xmath92 fs for @xmath93 ev . ) @xmath90 ( solid red line ) and @xmath91 ( dashed blue line ) . ] also , it corresponds to a sine - shaped laser pulse ; that is , the maximum of the electric field s envelope coincides with a minimum of the oscillating function @xcite . to also model a cosine - shaped pulse , we consider as a second choice of four - potential @xmath94 corresponding to the electric field s shape function @xmath95 the shape functions are show in fig .
[ sechtanhfieldpic ] and the temporal pulse duration is similar to that for choice ( [ elfield ] ) .
the above expressions of the pulse shape functions describe well a single - cycle laser pulse and have the advantage that a number of exact analytical results can be obtained . in order to closer investigate the two choices for the shape functions ( [ fourpotential ] ) and ( [ fourpotential1 ] ) , the fourier transforms of the two electric fields
are shown in fig .
[ fourierpic ] .
it is clear that the two choices lead to different frequency distributions with central angular frequencies of @xmath96 ev and @xmath97 ev , respectively .
( solid red line ) and for @xmath98 ( dashed blue line ) . ]
even though for the computation of emission spectra we need specific @xmath90 , many of our findings below will be valid for arbitrary shape functions .
plugging now the solutions ( [ volkov ] ) into eq .
( [ matrixelement ] ) we end up with the expression @xmath99 of the matrix element @xmath100 , where the notation @xmath101 is used and @xmath102\,u_{p\sigma } , \end{aligned}\ ] ] where @xmath103 is the complex conjugate of the polarization vector of the outgoing photon and where the functions @xmath104 } , & & i\in\{0,1,2\},\end{aligned}\ ] ] contain all the relevant dynamical information of the process . thus analyzing the scattering process amounts essentially to evaluating the functions ( [ f_i ] ) .
there we have introduced the important parameters @xmath105 with @xmath106 and @xmath107 being the spherical angular coordinates of the emitted photon , assuming the positive @xmath84 axis as the polar axis .
it is convenient to introduce the notation @xmath108 , \label{expfct}\ ] ] for the exponent in eq .
( [ f_i ] ) . also , we mention that the first parameter function @xmath109 is divergent because the complex exponential factor is not tending to zero for @xmath110 .
this problem , however , can be circumvented by expressing @xmath109 as a linear combination of the other two parameter functions , which are well defined due to the preexponential function @xmath90 , which goes to zero in the considered case of a pulsed laser field .
in fact , @xmath111 on the other hand , it is @xmath112 then @xmath113 it is interesting to note that the @xmath114 functions in eq .
( [ finalsfi ] ) state the energy - momentum conservation in the considered process .
we have here a deep difference between the scattering of an electron off a monochromatic laser wave and off an ultrashort laser pulse . in the former case
, it is possible to expand the transition matrix element ( [ finalsfi ] ) into a fourier series in which the @xmath115 term contains a four - dimensional energy momentum - conserving @xmath114 function of the form @xmath116 .
these @xmath114 functions can be interpreted as energy momentum conservation laws for an electron absorbing @xmath117 photons from the laser field and emitting a single photon of wave vector @xmath55 .
the quantities @xmath118 and @xmath119 appearing in the energy - momentum conservation laws are defined as @xmath120k_{\mu}$ ] and @xmath121k_{\mu}$ ] and are the so - called quasi momentum , of the electron before and after scattering ( see @xcite ) .
the square of this quasi momentum is equal to the square of the so - called dressed mass @xmath122 .
from the fact that in our conservation laws , there occurs no quasi momentum , we conclude that an electron scattering off an ultrashort laser pulse inside this pulse will not behave as if it had a dressed mass , unlike in the scattering off a long pulse .
this observation was just recently put into a quantitative frame considering electron - positron pair creation in ultra strong laser pulses by heinzl et al .
there the authors make the interesting suggestion to recover information on the mass dressing by investigating the positions of resonance peaks in pair production spectra which are affected by an electron mass shift . from their theoretical investigations , a reduction of the effective mass inside a few - cycle pulse of an order of magnitude
is obtained as compared to the monochromatic case of the same intensity . the question
if an electron acquires a mass dressing could also be investigated experimentally by utilizing nonlinear compton scattering .
in fact , from the second energy momentum - conserving @xmath114 function in eq .
( [ finalsfi ] ) , we obtain @xmath123 measuring now not only the final photon s but also the electron s momentum after scattering , this equation could be tested . furthermore , from eq .
( [ ergconservation ] ) , one can easily determine the maximum energy which can be emitted at an angle @xmath106 : @xmath124 if , in addition to the above conservation law , we consider that @xmath125 , we find an explicit formula to determine the energy of the electron after the scattering @xmath126}{\epsilon+p- \omega'[1-\cos\left(\vartheta\right)]}. \label{epsdelta}\end{aligned}\ ] ] the first energy momentum - conserving @xmath114 function in eq .
( [ finalsfi ] ) simply states that if the electron is initially counterpropagating with the laser field , then it holds that @xmath127 starting from the above @xmath128-matrix element @xmath100 , one can calculate the emitted energy spectrum @xmath129 ( average energy emitted between @xmath130 and @xmath131 , in the solid angle @xmath132 ) by averaging over the initial electron spin and by summing over the final electron spin and the photon polarization .
the final result reads as follows @xmath133-\epsilon- p\right\}-2m^2\right ) |f_0|^2 \nonumber \\ & & -m\xi\frac{\omega^{\prime\,2}\sin(\vartheta)\left[1-\cos(\vartheta)\right]\cos\left(\varphi\right)}{\epsilon+p- \omega'\left[1-\cos(\vartheta)\right]}\text{re}\left(f_0f_1^*\right)-m^2\,\xi^2 \text{re}\left(f_0f_2^*\right)\nonumber \\ & & \left.+\frac{m^2\xi^2}{2}\left\{\frac{\epsilon+p- \omega'\left[1-\cos(\vartheta)\right]}{\epsilon+p}+\frac{\epsilon+p}{\epsilon+p- \omega'\left[1-\cos(\vartheta)\right]}\right\}|f_1|^2\right\rangle , \label{ergspec}\end{aligned}\ ] ] where the expression of @xmath134 from eq .
( [ epsdelta ] ) has to be substituted .
the parameter functions @xmath135 given above in general are not analytically integrable . however
, their properties can be investigated in limiting cases .
for instance , in the low - intensity regime @xmath136 , it is possible in eq .
( [ f_i ] ) to expand the parameter functions as perturbation series in the small parameter @xmath6 , which makes them integrable . by noting that @xmath137 , @xmath138 , and @xmath139 , and after expanding around @xmath140 ,
one obtains the textbook result of single - photon compton scattering @xcite .
we consider here the case of a highly relativistic electron scattered by a relativistically intense laser pulse , that is , @xmath141 . in this case , since the electron is always ultrarelativistic , one expects from classical considerations that it radiates mostly along its instantaneous velocity @xcite , that is , close to polar angles @xmath142 for @xmath143 , @xmath144 for @xmath145 or essentially within the whole interval @xmath146 $ ] for @xmath147 . on the other hand ,
the expected azimuthal angle range is @xmath148 or @xmath149 .
as we will see , these classical considerations also hold in the quantum case .
the reason is that the motion of an ultrarelativistic electron is essentially quasiclassical in the presence of under critical electromagnetic fields , and quantum effects amount to the recoil due to photon emission @xcite .
then , at large values of @xmath6 and @xmath150 , as we are considering here , we can focus our analysis on the plane of polarization .
moreover , the above estimations on the emission region together with the fact that @xmath151 @xcite help in concluding that in the present regime , the parameters defined in eq .
( [ alphabetafinal ] ) scale as @xmath152 here it is important to note that the asymptotic relation for @xmath153 does not hold if @xmath154 .
thus we will have to restrict our calculations to angles @xmath106 not too close to @xmath155 . from these relations and from eq .
( [ f_i ] ) , it can be seen that the exponential factor of the parameter functions is very large and the integrand is rapidly oscillating .
this allows for an evaluation of the integrals by means of the stationary - phase method .
the condition for finding a stationary point @xmath156 of the phase is given by @xmath157 which gives @xmath158 from the expressions in eq .
( [ alphabetafinal ] ) of the parameters @xmath159 , @xmath160 , and @xmath153 , it can easily be shown that the expression under the root is given by @xmath161\equiv-\kappa^2,\ ] ] which is always negative and scales as @xmath162 .
this fact allows in some circumstances to neglect the imaginary part of the stationary point @xmath156 and to treat it as a real number .
note that the above equations hold for an arbitrary shape function @xmath90 . the choice of the sign in eq .
( [ statpoint ] ) depends on the specific shape function of the four - potential , and it has to provide that the resulting expression of the integral is not diverging . for example
, we will show later that for @xmath163 , we have to choose the minus sign in front of the square root .
for the four - potential in eq .
( [ fourpotential1 ] ) , on the other hand , the choice of the sign depends on the specific stationary point we are considering .
since , in the limit @xmath164 , it holds that @xmath159 and @xmath160 are of the same order , then @xmath165 and @xmath156 is real only for @xmath166 where @xmath167 are the minimum and maximum value which the function @xmath168 takes for real @xmath169 , respectively .
the condition ( [ statptcond ] ) is important because if the stationary point has an imaginary part , the corresponding integral in eq .
( [ f_i ] ) contains an exponentially damping factor , which in turn implies that the emission is suppressed ( note that the possibility that the integral shows an exponentially amplifying term is excluded from physical considerations ) .
consequently , we consider only such situations in which the condition ( [ statptcond ] ) is fulfilled .
once the process parameters such as the laser intensity and the incoming electron energy are fixed according to a specific experimental setup , the condition ( [ statptcond ] ) turns into a boundary condition for the observation angles @xmath106 and @xmath107 . in the case under consideration here
, the emission will be detectable only in a narrow cone around the azimuthal angles @xmath170 and @xmath171 .
thus , fixing @xmath107 to one of these values , the condition in eq .
( [ statptcond ] ) provides two values @xmath172 and @xmath173 , confining the polar angle range within which significant radiation is expected .
considering that for the choice ( [ fourpotential ] ) , we have @xmath174 and @xmath175 , one , for instance , finds from eq .
( [ statptcond ] ) that at @xmath171 , there is no emission expected .
furthermore , it is found that to observe emission at @xmath176 , it must be @xmath177 . for the choice ( [ fourpotential1 ] ) ,
on the other hand , we have @xmath178 and @xmath179 , and consequently , emission into the half space @xmath176 is observed only for @xmath180 . on this respect , we observe that , based on the relation ( [ statptcond ] ) , we have proposed in @xcite a method for determining the cep of laser pulses with @xmath38 . now , following the method of stationary phase
, we expand the exponential function as well as the preexponentials in the integrals ( [ f_i ] ) in a perturbation series in @xmath181 since for values of @xmath169 far away from the stationary point , the rapid oscillations of the integrand will suppress any contribution to the integral s value : @xmath182\right|_{\eta=\eta_0 } \exp\left[{\mathbbm{i}}\sum_{m=0}^{m}\left.\frac{(\eta-\eta_0)^m}{m!}\left(\frac{\partial}{\partial \eta}\right)^m g(\eta)\right|_{\eta=\eta_0}\right ] , \label{expandedfj}\ ] ] with @xmath183 being given in eq .
( [ expfct ] ) and @xmath184 and @xmath185 being the orders up to which the preexponential and the exponential functions are expanded , respectively . due to cancellations in the squared matrix element , the preexponential perturbation series needs to be taken into account up to second order .
the exponential series , on the other hand , needs to be considered up to third order .
in fact , from the above expression of the function @xmath183 , we have @xmath186 from eq .
( [ statpoint ] ) we obtain @xmath187 and @xmath188
. therefore the values of @xmath169 around @xmath156 contributing to the integrals are such that @xmath189 and the second- and the third - order terms in the exponential give contributions of the same order of magnitude . from the above considerations , eq .
( [ expandedfj ] ) takes the form @xmath190 where @xmath191 is an index running over all stationary points @xmath192 found from solving eq .
( [ statpointeq ] ) . for reasons of convenience
, we defined @xmath193 , @xmath194 and @xmath195 and the integrals @xmath196= \nonumber \\ & = \exp\left[{\mathbbm{i}}\left(g_{0,l } + \frac 13 \frac{g^{(2)^3}_{0,l}}{g^{(3)^2}_{0,l}}\right)\right]\int_{-\infty}^{\infty } { \textnormal{d}}y_l \ y_l^i\exp\left[{\mathbbm{i}}\left(\frac{g^{(3)}_{0,l}}{6}y_l^3 - \frac 12 \frac{g^{(2)^2}_{0,l}}{g^{(3)}_{0,l}}y_l\right)\right ] , \label{expandedfi}\end{aligned}\ ] ] where we have introduced the quantities @xmath197 , @xmath198 and @xmath199 and where @xmath200 with @xmath201 . in order to work out the exponential factor outside of the integral in eq .
( [ expandedfi ] ) , we note that @xmath202 where the freedom to choose the sign of @xmath203 is written explicitly .
thus this term gives a purely real contribution to the exponential in eq .
( [ expandedfi ] ) .
if we next expand @xmath204 as a function of @xmath205 , around zero up to third order , we can easily show that the imaginary part of @xmath204 exactly cancels the quantity @xmath206 in such a way that @xmath207 finally , if @xmath208 in eq .
( [ gnrlasympexp ] ) , then the integration in @xmath209 in eq .
( [ expandedfi ] ) can be performed analytically , and the result is @xmath210 where ai@xmath211 is the airy function of first kind @xcite and its argument @xmath212 is defined as @xmath213 now , from the first line in eq .
( [ expandedfi ] ) we obtain that @xmath214 finally , in order to compute the last integral @xmath215 , we employ the same technique as in eq .
( [ f0decomposition ] ) and obtain @xmath216 in conclusion , eqs .
( [ gnrlasympexp ] ) , ( [ j0final ] ) , ( [ j2final ] ) and ( [ j1final ] ) provide us with the asymptotic expansion of the @xmath135 in the ultrarelativistic regime . to illustrate the above techniques , we apply them to the shape functions ( [ fourpotential ] ) and ( [ fourpotential1 ] ) . in the former case we note that by virtue of symmetry considerations , the parameter functions @xmath135 need to be evaluated only in the interval @xmath217 $ ] .
thus eq .
( [ statpointeq ] ) provides as a unique stationary point the positive solution of the equation @xmath218 the exponential function in this case is given by @xmath219 - \beta\ , { \text{tanh}}(\eta)+ \gamma\,\eta , \label{expfct1}\ ] ] where an unimportant constant phase was dropped .
according to eqs .
( [ derivatives ] ) and the discussion following it , the second and third derivatives of ( [ expfct1 ] ) at the stationary point are @xmath220 since it must be @xmath221 to obtain convergent integrals @xmath222 , and noting that @xmath223 , we find that for the choice ( [ fourpotential ] ) in eq .
( [ statpoint ] ) , one has to choose the negative sign of @xmath203 .
finally , by plugging now the solutions ( [ expderivatives ] ) into the above equations , we immediately find the asymptotic transition amplitude .
next , for the cosine - shaped pulse arising from the choice ( [ fourpotential1 ] ) , we meet a different situation .
since in this case the integrals @xmath135 do not feature symmetry properties , we have to integrate the whole range @xmath224 $ ] .
then , however , eq . ( [ statpointeq ] ) gives two real solutions @xmath225 @xmath226\nonumber \\ \eta_{0,2 } & = & \text{sgn}\left(\frac{\alpha}{2\beta}\right)\ \text{arcsech } \left[\sqrt{\frac{2\left(\nicefrac{\alpha}{2\beta}\right)^2}{1+\sqrt{1-(\nicefrac{\alpha}{\beta})^2}}}\right].\end{aligned}\ ] ] it is easy to obtain @xmath227 and @xmath228 by plugging these latter expressions into eqs .
( [ derivatives ] ) , but the resulting expressions are cumbersome and thus not reported .
the asymptotic expansion of the @xmath135 is then simply obtained by summing up the contributions from the two stationary points . in electron - laser scattering , there are essentially two types of quantum effects .
first the fact that the motion of the electron is quantum .
however , due to the fact that at @xmath229 , the electron is highly relativistic and its typical de broglie wavelength is very small , this effect may be neglected . on the other hand ,
if an electron emits a photon whose energy is comparable to the electron s energy , it will feel a recoil . since , for the incident laser pulse , we are considering optical frequencies , the only energy which can possibly be comparable to the electron s initial energy is the outgoing photon s energy .
the classical limit is therefore defined by the condition @xcite @xmath230 this allows for some major simplifications in the emission probability [ eq .
[ ergspec ] ] : @xmath231-|f_0|^2\right\ } , \label{classlim}\end{aligned}\ ] ] with the functions @xmath135 still given by eq .
( [ f_i ] ) but with the parameters @xmath159 , @xmath160 , and @xmath153 given by eq .
( [ alphabetafinal ] ) with the substitution @xmath232 . in this way ,
one exactly recovers the classical expression of the energy spectrum calculated from the trajectory of the electron in a plane wave and from the linard - wiechert potentials @xcite .
the analytical results obtained in the previous section give the differential photon energy spectrum of nonlinear compton scattering in ultrashort laser pulses .
we are going to show low - intensity emission spectra for the choice ( [ fourpotential ] ) only . for high intensities
@xmath43 , however , we aim to show the applicability of the stationary phase analysis to arbitrary shape functions and show emission spectra for the choice ( [ fourpotential1 ] ) as well .
all spectra are plotted at the observation angle @xmath233 , unless otherwise mentioned .
this choice is justified by the fact that an ultrarelativistic electron mainly emits in the plane determined by the laser polarization and propagation direction .
we characterize the low - intensity regime by either @xmath36 or @xmath234 . according to the arguments given in the introduction , the regime in which @xmath235 corresponds to scattering dominated by the single - photon processes as depicted in fig .
[ singlecomptonpic ] . in this linear regime of compton scattering ,
the parameter @xmath27 is not appropriate to distinguish the importance of quantum effects since it does not give an estimate for the importance of linear quantum effects .
as we have mentioned in the introduction , an appropriate parameter to quantify their importance is @xcite @xmath236 with the last equality holding in the special coordinate frame introduced in the last paragraph . from eq .
( [ linearquantumeffects ] ) it follows that in the regime @xmath36 , we have @xmath237 , and quantum effects can affect the spectra even for @xmath238 as soon as @xmath239 ; however , in this case they are linear . in this section
we will plot emission spectra only for @xmath240 because at small values of @xmath6 an initially ultrarelativistic electron experiences only a little deviation from its initial propagation direction upon scattering from the laser pulse .
following the procedure outlined in the theory section , we computed perturbative energy spectra . to this purpose ,
we chose a small nonlinearity parameter @xmath241 corresponding to an optical intensity of @xmath242 w@xmath243@xmath244 ( here and in the following , the laser photon energy is assumed to be @xmath245 ) . in fig .
[ lowintspec](a ) we show the classical ( crosses ) and the quantum ( solid red line ) emission spectra for an electron with a moderate initial lorentz factor of @xmath246 .
the above parameters correspond to a quantum parameter of @xmath247 , and in fact , we observe that the two spectra are identical .
the classical spectrum was obtained by first solving the lorentz equation and then plugging the resulting trajectory into the linard - wiechert potentials [ see eq .
( 66.9 ) in @xcite ] . in fig .
[ lowintspec](b ) , however , we consider an increased lorentz factor for the incoming electron of @xmath248 leading to @xmath249 . thus quantum effects are important in this case .
in fact , the spectrum shows two clear differences with respect to the classical spectrum of fig .
[ lowintspec](a ) , only the first of which can be explained classically . and
( a ) @xmath246 ( @xmath250 ) and ( b ) @xmath251 ( @xmath252 ) .
the blue crosses in both spectra have been obtained by a classical calculation . ]
the first difference is that in fig .
[ lowintspec](b ) , considerably higher photon frequencies are emitted .
this can be explained through a blueshift of the incoming laser s photons .
in fact , since @xmath235 , the observed peak corresponds to the absorption of one laser photon . in the electron s average rest frame ,
the central laser photon frequency @xmath253 is blueshifted to @xmath254 in this expression , @xmath255 is the electron s drift velocity , that is , the velocity of the reference frame in which the electron would be on average at rest during the scattering process if the scattering laser field were periodic @xcite . in this frame
the electron emits a photon with the same frequency as the incident ones .
then the energy of the emitted photon at an angle @xmath106 is obtained by going back to the laboratory frame .
after some algebra , one finds as the theoretically predicted frequency of the spectrum s maximum @xmath256 the maximum of the spectrum in fig .
[ lowintspec](a ) agrees very well with the frequency @xmath257 ev computed via eq .
( [ omegatheo ] ) . however , for the parameters of fig .
[ lowintspec](b ) , this equation predicts a peak at @xmath258 ev , what is indeed reproduced by the crosses but strongly disagrees with the quantum result .
in fact , at very large @xmath259 and at @xmath240 , we have , for eq .
( [ omegatheo ] ) , @xmath260 that is , @xmath261 grows quadratically with @xmath150 .
now , looking back at eq .
( [ omgmx ] ) , we see that there exists a maximally allowed emission frequency , which for large @xmath150 and @xmath240 reads as @xmath262 which grows linearly with @xmath150 .
thus , by comparing the latter two equalities , we already infer that at some value of @xmath150 , the blueshifted center frequency of the incident laser pulse will exceed the maximally allowed emission frequency . at these values of @xmath150 ,
the spectra will change their shapes , and the radiation will pile up toward the cutoff energy @xmath263 @xcite .
these considerations can be put in a more mathematical form by observing that classically , the spectra at small values of @xmath6 are essentially given by the fourier transform of the laser field ( in this regime , the parameters @xmath159 , @xmath160 , and @xmath153 are proportional to @xmath130 ) .
on the other hand , in the quantum regime , the parameters @xmath159 , @xmath160 , and @xmath153 have a nonlinear dependence on @xmath130 , and they diverge in the limit @xmath264 . in this way
, the resulting strong oscillations damp the spectrum in the same limit . by increasing the value of the nonlinearity parameter to @xmath19 ,
new features arise in the energy emission spectra , and if @xmath265 , they can also be interpreted classically . in fig .
[ destintspec ] we show the emission spectrum for @xmath266 and an initial @xmath150 factor of the electron of @xmath267 . and @xmath267 ( @xmath268 ) .
the crosses have been obtained by a classical calculation . ]
this choice corresponds to a quantum parameter of @xmath268 , and the crosses again give the results of a classical computation .
although there are slight discrepancies between classical and quantum results toward higher emission frequencies , which actually hint at upcoming quantum effects , up to around the first minimum , they still match very well .
the appearance of several minima and maxima in the spectrum is the main difference to the previous case in which @xmath235 .
the position of the minima can be interpreted both classically and in terms of a quantized photon field . the quantum picture is easily established , reminding us that the parameter @xmath6 is interpretable as the average number of photons absorbed by the electron over one compton wavelength . then the additional peaks in the spectra can be viewed as multiphoton peaks .
the positions of these maxima are the integer multiples of the blueshifted ground frequency predicted by eq .
( [ omegatheo ] ) . by computing the blueshifted ground frequency @xmath261 for @xmath269 and @xmath267 , as in fig .
[ destintspec ] , we find @xmath270 which indeed agrees well with the first maximum of fig .
[ destintspec ] .
classically the pattern is interpreted as interferences between the field emitted by the electron in different parts of its trajectory .
in fact , it can be seen that the electron emits twice into every angle @xmath271 . to clarify this
, we show in fig .
[ electrontraj ] the trajectory of an electron ( solid curve ) moving in a field corresponding to eq .
( [ elfield ] ) ( dashed blue curve ) .
the two red parts of the trajectory correspond to the emission angle @xmath271 .
furthermore , we note that as visualized in fig .
[ electrontraj ] , a classical electron covers only a certain angle range deviating from its initial direction of propagation when moving in the given electric field . since its velocity vector will only point into this confined angular region ,
the electron will also only emit into this region , provided it is ultrarelativistic . )
( dashed blue line ) . ]
this is the classical analog to the boundary angles found from eq . ( [ statptcond ] ) . in order to quantitatively
investigate the interferences from these two trajectory segments , we employ a classical calculation .
now , classically the differential energy spectrum @xmath129 can be calculated via , for example , eq . ( 14.65 ) in @xcite . in this formula
we can replace the integral over all times by a sum over the two time instants when the electron emits into direction @xmath272 .
these instants can be distinguished by the phase values @xmath273 and @xmath274 .
we then end up with a formula for the differential energy spectrum @xmath275}{\left(1- \bm{\beta}_1 \cdot \mathbf{n } \right)^2 } { \text{e}}^{{\mathbbm{i}}\,\omega ' \left[t_1 - \mathbf{n}\cdot \mathbf{r}(t_1)\right]}+ \frac{\mathbf{n } \times \left[\left ( \mathbf{n}-\bm{\beta}_2\right)\times \dot{\bm{\beta}_2 } \right]}{\left(1- \bm{\beta}_2 \cdot \mathbf{n } \right)^2 } { \text{e}}^{{\mathbbm{i}}\,\omega ' \left[t_2 - \mathbf{n}\cdot \mathbf{r}(t_2)\right]}\right\}\delta t\right|^2 .
\label{linwiechsum}\ ] ] with @xmath276 being the electron s velocity , @xmath277 its acceleration at the two time instants , and @xmath278 the discretized time interval .
the approximation of only two space - time points contributing to the radiation detected under @xmath271 is better fulfilled the larger the electron s lorentz factor at the instant of emission is , that is , the narrower its emission cone becomes . in order to evaluate eq .
( [ linwiechsum ] ) , we note that the points @xmath279 and @xmath280 by construction satisfy that @xmath281 , that the direction of observation is @xmath282 , and that , since the electric field from eq .
( [ elfield ] ) is antisymmetric around its zero point , the forces acting on the electron at the two points are exactly opposite : @xmath283 . following these considerations , we have @xmath284}{\left(1- \bm{\beta } \cdot \mathbf{n } \right)^2}\left\ { { \text{e}}^{{\mathbbm{i}}\,\omega ' \left[t_1 - \mathbf{n}\cdot \mathbf{r}(t_1)\right ] } - { \text{e}}^{{\mathbbm{i}}\,\omega ' \left[t_2 - \mathbf{n}\cdot \mathbf{r}(t_2)\right]}\right\}\right|^2\!\!\!. \label{linwiechsum2}\end{aligned}\ ] ] so we note that we will have @xmath285 , that is , destructive interference at the frequencies @xmath286 , where it holds that @xmath287-\left[t_2 - \mathbf{n}\cdot \mathbf{r}(t_2)\right]}. \label{omegan}\end{aligned}\ ] ] the method described above of course is not rigorously applicable at the polar angle @xmath240 because many points of the electron s trajectory contribute to the emission in that direction .
however , the frequencies at which destructive interference occur can be found by computing the @xmath286 at a polar angle @xmath288 and then considering the limit @xmath289 .
now , by solving eq .
( [ statptcond ] ) for the minimal polar angle where radiation is expected for the choice ( [ fourpotential ] ) of @xmath90 , we find @xmath290 accordingly , by numerically analyzing the electron trajectory , one obtains from eq .
( [ omegan ] ) that @xmath291 matching the observed first frequency of destructive interference from fig .
[ destintspec ] well .
it is noteworthy that in the above derivation the ultrashort duration of the scattering laser pulse did not enter .
thus , in the realm of classical electrodynamics , the position of the multiphoton peaks could equally have been recovered in an analysis of a long pulse . in this case , however , the peaks would have been much narrower going in the limit of a monochromatic wave to delta spikes . on the other hand , in fig .
[ electrontraj ] and in all other emission spectra we show , we observe significant broadening of the multiphoton peaks .
this is the second essential difference in the scattering of an electron from a long and an ultrashort laser pulse apart from the absence of a mass dressing in the latter case .
it is interesting to note in eq .
( [ omegan ] ) that the value of @xmath292 is inversely proportional to the difference @xmath293 $ ] .
now , at @xmath294 , it is @xmath295 . from the analytical solution of the lorentz equation in a plane wave
, it can be seen that at a given @xmath150 , the difference @xmath293 $ ] increases at increasing @xmath6 .
consequently , the first frequency of destructive interference will be smaller .
this assertion is fully consistent with the quantum multiphoton picture . in order to confirm it ,
we show in fig .
[ medintspec ] the energy emission spectra again at @xmath267 , as in fig . 8 , but with a nonlinearity parameter of @xmath296 [ fig .
[ medintspec](a ) ] . and ( a ) @xmath267 ( @xmath297 ) and ( b ) @xmath251 ( @xmath252 ) . the crosses in ( a ) have been obtained by a classical calculation . ]
+ in fig .
[ medintspec](a ) , @xmath298 and the spectrum is classical ( the crosses again give the results of a classical calculation ) .
comparison with fig .
8 clearly shows the appearance of many peaks .
computing for the parameters of fig .
[ medintspec](a ) , the frequencies of the first maximum and minimum according to eq .
( [ omegatheo ] ) ( multiphoton picture ) and eq .
( [ omegan ] ) ( interference picture ) , respectively , we find @xmath299 this is in good agreement with the numerical result . on the other hand , in fig .
10(b ) , it is @xmath300 , and quantum effects dominate the spectrum s structure .
this is manifest from the fact that the distance between two next peaks decreases as the emission frequency tends to @xmath263 .
we finally observe that deviations from eq .
( [ omegan ] ) may also arise for not very large @xmath150 and @xmath6 . in this case
the electron is not ultrarelativistic , and its emission cone will be relatively wide .
the wider the emission cone is , however , the poorer is the assumption that the emission detectable at a certain @xmath271 originates from only two distinct space points , which was an essential assumption to arrive at eq .
( [ omegan ] ) . in order to calculate the spectra in the high - intensity regime ( @xmath38 )
, we substitute the asymptotic expansions ( [ gnrlasympexp ] ) into the general formula ( [ ergspec ] ) .
since the spectra would show numerous maxima and minima , it is clearer to plot only the envelope of the spectra .
furthermore , we show that in fact , in this regime , the parameter @xmath27 is fit to quantify the onset of quantum parameters . finally , apart from the last two examples ( figs .
14 and 15 ) , we will employ the sine - shaped pulse in eq .
( [ fourpotential ] ) .
we begin by presenting the energy emission spectrum for the parameters @xmath301 . according to eq .
( [ chieq ] ) , this corresponds to a quantum nonlinearity parameter of @xmath302 . from eq .
( [ statptcond ] ) , we find that with this choice , the minimal angle where radiation is expected is @xmath303 , and in fig .
[ asymp_spec_chi=0.02 ] , we plot this as the smallest scattering angle . . for the emission angles @xmath304 ( solid red line ) , @xmath305 ( dashed blue line ) , and @xmath306 ( dotted green line ) .
] at @xmath172 , the energy spectrum covers only substantially smaller emission frequencies than at larger angles . the maximal polar observation angle the emission spectrum is plotted for is @xmath307 .
the reason for this is that , as discussed [ eq . ( [ abkasymp1 ] ) ] , for larger @xmath106 , the approximation @xmath308 does not hold any more . at @xmath309 , for example , we find @xmath310 and for growing angles , this ratio quickly exceeds 10 . since for the above parameter choice the quantum nonlinearity parameter @xmath27 is rather small , a classical calculation gives emission spectra comparable to those in fig .
11 . if , according to eq .
( [ omgmx ] ) , we calculate the maximum frequency that may be emitted from an electron under the given process parameters , we find @xmath311 and the maximum frequencies at all other angles in the range @xmath312 $ ] lie between these two frequencies .
the frequencies actually emitted in the spectra in fig .
[ asymp_spec_chi=0.02 ] are approximately 1 order of magnitude smaller than the maximally allowed emission frequency , implying that we are in fact in the classical regime . in order to observe backscattering , that is , @xmath313
, we have to consider @xmath177 .
we thus consider a lower value of @xmath150 and plot the emission spectra in the case of @xmath314 and @xmath315 , such that we expect emission into a large polar angle regime .
the quantum parameter in this scenario is @xmath316 , and the resulting spectra are shown in fig .
[ asymp_spec_chi=0.01 ] . and
@xmath317 for the emission angles @xmath318 ( solid red line ) , @xmath319 ( dashed blue line ) , and @xmath320 ( dotted green line ) . ] the general shape of these spectra is similar to those previously shown in fig .
[ asymp_spec_chi=0.02 ] , albeit with different axis scales .
the emission range indeed is found to extend down to emission angles close to @xmath321 .
due to its lower initial energy as compared to the laser intensity , the deviation of the electron s direction of propagation from its initial orientation is increased .
we have ensured that the emission is strongly suppressed at @xmath322 . in order to study a case where quantum effects become important ,
we choose @xmath314 and @xmath323 , which gives a quantum parameter of approximately @xmath324 .
the resulting spectra are presented in fig .
[ asymp_spec_chi=2 ] . and
@xmath325 for the emission angles @xmath326 ( solid red line ) , @xmath327 ( dashed blue line ) , and @xmath328 ( dotted green line ) . ] in the spectra plotted we find clear differences with respect to those in figs .
[ asymp_spec_chi=0.02 ] and [ asymp_spec_chi=0.01 ] .
not only do the scales of emitted frequencies and of the spectra s amplitudes differ by several orders of magnitude , but also the shapes of the spectra look distinctly different .
so in fig .
[ asymp_spec_chi=2 ] , we find a fast decay of the emitted energy when the emitted frequencies approach the maximally emitted photon energy , while in fig .
[ asymp_spec_chi=0.02 ] , the spectra exhibited a much more moderate decrease .
this again is evidence for the effect of kinematic pile up as described in section [ classresultssec ] .
computing again the maximally allowed emitted photon frequency according to eq .
( [ omgmx ] ) , we find that @xmath329 which is close to the frequency where the fast drop off in fig .
[ asymp_spec_chi=2 ] occurs .
so we conclude that in the regime @xmath38 , as soon as the quantum parameter @xmath27 becomes of order unity , the emitted photons will approach their maximally allowed frequencies .
this interpretation of the spectra s distortions found in fig .
[ asymp_spec_chi=2 ] clearly hints at quantum effects which , due to energy - momentum conservation , prevent the emission of higher energetic photons . up to now
, all emission spectra were obtained for the choice of @xmath90 corresponding to a sine - shaped electric field [ eq . ( [ fourpotential ] ) ] . to investigate the scattering off a cosine - shaped pulse
, we computed emission spectra for the choice ( [ fourpotential1 ] ) . for this choice from eq .
( [ statptcond ] ) we do not only expect emission into the azimuthal angle regime around @xmath233 but also to @xmath330 .
we begin in fig .
[ asymp_spec_cep_chi=0.04](a ) by showing the spectra for @xmath331 , which makes it easy to compare the results to those shown in fig .
[ asymp_spec_chi=0.02 ] .
we have ensured that the spectra at @xmath171 are the same as those at @xmath170 , and we show only the case @xmath170 . at the emission angles @xmath332 ( solid red line ) , @xmath333 ( dashed blue line ) , and @xmath306 ( dotted green line ) . + ( b ) envelopes of the emission spectra for @xmath315 , @xmath334 at the emission angles @xmath321 ( solid red line ) , @xmath335 ( dashed blue line ) , and @xmath336 ( dotted green line ) . ] we note that in eq .
( [ statptcond ] ) , the values @xmath167 are changed .
thus the angular range where emission is predicted changes with respect to the spectra in fig .
[ asymp_spec_chi=0.02 ] for an unchanged ratio @xmath337 . in fig .
[ asymp_spec_cep_chi=0.04 ] , we accordingly plotted the spectrum at @xmath332 as the smallest polar angle because it is found from eq .
( [ statptcond ] ) that the emission extends only to larger angles . in fig .
[ asymp_spec_cep_chi=0.04](b ) , we show the emission spectra for @xmath315 and @xmath338 . in these spectra
we find the minimal emission angle to be @xmath339 .
so it becomes obvious that for the choice ( [ fourpotential1 ] ) , in order to observe radiation in the space region @xmath340 , a nonlinearity parameter @xmath341 must be employed .
this is a clear difference to the sine - shaped pulse , where for backscattering to occur , it had to be satisfied that @xmath177 . finally we show a spectrum for the cosine - shaped pulse where quantum effects become important . for this purpose
we choose the parameter set @xmath314 and @xmath342 , corresponding to a quantum parameter @xmath343 . the resulting spectrum is shown in fig . [ asymp_spec_chi=4 ] . and @xmath325 for the emission angle @xmath328 . ] as for the case of the sine - shaped pulse , we observe in this spectrum the typical quantum pile - up of the radiation toward @xmath263 .
as we mentioned in the introduction there has already been a lot of work on thomson and compton scattering from monochromatic @xcite or long - pulsed @xcite lasers fields . in the present article
we have considered the opposite situation in which the pulse contains only one or a few cycles .
comparing our results to previous work , we find some agreement as well as some differences . we obtain analogous qualitative behavior in our emission spectra concerning the rise of nonlinear quantum effects in the scattering process .
specifically we could show that in the regime @xmath38 , the parameter @xmath27 also in the case of ultrashort laser pulses is suitable to characterize the onset of quantum effects .
however , we could unveil two major differences between these earlier treatments and our analysis .
first of all , we found the momentum conserving @xmath114 functions contained in the transition matrix element to differ from those obtained for scattering off monochromatic laser waves or long pulses .
we interpreted this as the absence of a dressed mass effect in particular for ultrashort pulses containing only one laser cycle .
additionally we pointed out that this prediction could be tested by not only detecting the photons emitted in the scattering process but also the scattered electrons .
then , by measuring the momenta of the outgoing particles , one could judge if the electron propagated with a dressed mass inside the pulse or not .
second , we found a different structure of the energy emission spectra . even though the multiphoton peaks are still present , they are significantly broadened with respect to the case of a monochromatic laser beam .
this can be understood such that for ultrashort durations a laser pulse can no longer be viewed to be composed of many photons with the same energy .
these two features are the effects we find to be expected when considering ultrashort pulses in an electron - laser scattering event .
moreover we have shown analytically that , when classical electrodynamics apply , a classical interference picture and a quantum mechanical multiphoton description both consistently describe the position of the minima and maxima in the emission spectra .
finally , by means of a careful application of the saddle - point method , we have been able to calculate analytically the spectra at large values of @xmath6 also for ultrashort laser pulses .
t. vachaspati , phys . rev . * 128 * , 2 ( 1962 ) .
eberly and a. sleeper , phys .
* 176 * , 5 ( 1968 ) .
sarachik and g.t .
schappert , phys .
d * 1 * , 10 ( 1970 ) .
y.i . salamin and f.h.m .
faisal , phys .
a * 54 * , 4383 ( 1996 ) , y.i .
salamin and f.h.m .
faisal , ibid . * 55 * , 3678 ( 1997 ) .
narozhny and m.s .
fofanov , jetp * 90 * , 753 ( 2000 ) .
zeldovich , soviet physics uspekhi * 18 * 79 ( 1975 ) .
a.i . nikishov and v.i .
ritus , zh .
* 46 * , 776 ( 1964 ) [ sov .
jetp * 19 * , 529 ( 1964 ) ] .
baier et al .
, zh . eksp .
. fiz . * 69 * , 783 ( 1975 ) [ sov .
jetp * 42 * , 400 ( 1975 ) ] .
brown and t.w.b .
kibble , phys . rev . * 133 * , 3a ( 1964 ) .
ritus , journal of russian laser research * 6 * , 5 ( 1985 ) .
ivanov , g.l .
kotkin and v.g .
serbo , eur .
j. c * 36 * , 127 ( 2004 ) .
c. harvey , t. heinzl and a. ilderton , phys .
a * 79 * , 063407 ( 2009 ) .
n.b . narozhnyi and m.s .
fofanov , zh .
. fiz . * 110 * , 26 ( 1996 ) [ jetp * 83 * , 14 ( 1996 ) ] . l.d .
landau and e.m .
lifschitz , _ classical theory of fields _ , akademie verlag , berlin ( 1997 ) .
t. tajima and g. mourou , phys . rev .
beams * 5 * , 3 ( 2002 ) .
englert and e.a .
rinehart , phys .
a * 28 * , 1539 ( 1983 ) .
moore , j.p .
knauer and d.d .
meyerhofer , phys .
lett . * 74 * , 2439 ( 1995 ) .
meyerhofer , ieee journal of quantum electronics * 33 * , 11 ( 1997 ) .
s.y . chen , a. maksimchuk and d. umstadter , nature * 396 * , 653 ( 1998 ) .
m. babzien _
et al . _ , phys .
* 96 * , 054802 ( 2006 ) .
t. kumita _
et al . _ , laser physics * 16 * , 2 ( 2006 )
. c. bula _ et al .
* 76 * , 3116 ( 1996 )
. v. yanovsky et al . , opt .
express * 16 * , 2109 ( 2008 ) .
w. heisenberg and h. euler , zeitschrift fr physik * 98 * , 714 ( 1936 ) .
m. boca and v. florescu , phys .
a * 80 * , 053403 ( 2009 ) .
d. seipt and b. kmpfer , phys .
a * 83 * , 022101 ( 2011 ) .
t. heinzl , d. seipt , and b. kmpfer , phys .
a * 81 * , 022125 ( 2010 ) .
a. bonvalet _
et al . _ ,
* * , ( ) .
g. krauss _
_ , nature photonics * 4 * , 33 ( 2010 ) .
a.l . cavalieri _
_ , new j. phys * 9 * , 242 ( 2007 ) .
g. sansone _ et al .
_ , science * 314 * , 443 ( 2006 ) .
e. goulielmakis _
et al . _ ,
science * 320 * , 1614 ( 2008 ) .
/ research / pfs.html/. http://www.extreme - light - infrastructure.eu/. http://www.hiper - laser.org / index.asp/. f. mackenroth , a. di piazza and c.h .
keitel , phys .
lett * 105 * , 063903 ( 2010 ) .
d. m. volkov , z. phys .
* 94 * , 250 ( 1935 ) .
landau and e.m .
lifschitz , _ quantum electrodynamics _ , akademie verlag , berlin ( 1991 ) .
f. krausz and m. ivanov , rev .
mod . phys . *
81 * , 163 ( 2009 ) .
t. heinzl , a. ilderton and m. marklund , phys .
b * 692 * , 4 ( 2010 ) .
v. n. baier , v. m. katkov , and v. m. strakhovenko , _
electromagnetic processes at high energies in oriented single crystals _ , ( world scientific , singapore , 1994 ) .
m. abramowitz and i.a .
stegun , _ handbook of mathematical functions with formulas , graphs , and mathematical tables _ , dover - new york ( 1964 ) .
f. v. hartemann and c. w. siders and c. p. j. barty , phys .
lett . , * 100 * , 125001 ( 2008 ) .
j. d. jackson , _ classical electrodynamics _
john wiley & sons , new york ( 1962 ) . | a detailed analysis of the photon emission spectra of an electron scattered by a laser pulse containing only very few cycles of the carrying electromagnetic field is presented .
the analysis is performed in the framework of strong - field quantum electrodynamics , with the laser field taken into account exactly in the calculations .
we consider different emission regimes depending on the laser intensity , placing special emphasis on the regime of one - cycle beams and of high laser intensities , where the emission spectra depend nonperturbatively on the laser intensity . in this regime
, we , in particular , present an accurate stationary phase analysis of the integrals that are shown to determine the computed emission spectra .
the emission spectra show significant differences with respect to those in a long pulsed or monochromatic laser field : the emission lines obtained here are much broader , and more important , no dressing of the electron mass is observed . |
the power set of a set @xmath31 is denoted by @xmath44 .
the disjoint union of two sets @xmath45 is denoted by @xmath46 .
we assume that trees are rooted and denote the root of a tree @xmath47 by @xmath48 .
for every @xmath49 , @xmath50 is a @xmath51-times iterated exponential , i.e. , @xmath52 and @xmath53 . for a set @xmath31 and object @xmath37 ,
we let @xmath54 be defined as @xmath55 to avoid cluttered notation , we may , for elements @xmath56 and @xmath57 , abbreviate @xmath58 , @xmath59 , and @xmath60 .
we fix a countably infinite set of _
symbols_. each symbol @xmath61 has an _ arity _ @xmath62 .
we distinguish between _ nullary _ symbols with arity zero and _ relation symbols _ that have arity greater than zero .
relation symbols with arity one are called _
unary_. for convenience , we shall denote relation symbols by capital letters and nullary symbols by lower case letters . a _ vocabulary _
@xmath63 is a finite set of symbols .
we denote by @xmath64 the set of nullary symbols in @xmath63 , by @xmath65 the set of relation symbols in @xmath63 , and by @xmath66 the set of unary relation symbols in @xmath63 .
let @xmath67 be the maximum arity over all relation symbols in @xmath63 . if @xmath68 , we call @xmath63 _ relational_. let @xmath63 be a vocabulary . a _ structure @xmath34 over @xmath63 _ ( or @xmath63-structure ) is a tuple @xmath69 , where @xmath2 is a finite set called the _ universe _ of @xmath34 , and @xmath70 and @xmath71 are _ interpretations _ of the @xmath63-symbols in @xmath34 . here , @xmath72 for each relation symbol @xmath73 .
for a nullary symbol @xmath74 we either have @xmath75 and say that @xmath76 is _ interpreted _ in @xmath34 , or we write @xmath77 and say that @xmath76 is _ uninterpreted_. the set of nullary symbols interpreted in @xmath34 is denoted by @xmath78 .
if all symbols are interpreted , we say the structure is _ fully interpreted _ , and _ partially interpreted _ otherwise .
we note that a related concept of _ partially equipped signatures _ has been used in , e.g. , @xcite .
the set of all @xmath63-structures is denoted by @xmath79 .
we shall always denote structures in script letters @xmath80 and in roman letters @xmath81 their corresponding universes .
if the universe is empty , then we say that the structure is _
empty_. structures over a relational vocabulary @xmath63 are called _
relational structures_. for a structure @xmath34 , we denote by @xmath82 the vocabulary of @xmath34 . for sets
@xmath83 and @xmath84 , we let @xmath85 , and @xmath86 be their corresponding interpretations .
[ ex : taugraph ] a graph @xmath87 can in a natural way be identified with a structure @xmath88 over the vocabulary @xmath89 , where @xmath90 represents the binary adjacency relation .
the universe of @xmath88 is @xmath91 , and we interpret @xmath90 as @xmath92 in @xmath88 .
let @xmath63 be a vocabulary and @xmath93 be a set of symbols , each of which is not contained in @xmath63 .
the vocabulary @xmath94 is called an _ expansion _ of @xmath63 . similarly ,
if @xmath34 is a @xmath63-structure and @xmath35 is a @xmath95-structure that agrees with @xmath34 on @xmath63 , i.e. , @xmath96 for each @xmath73 and @xmath97 for each @xmath98 , then we call @xmath35 a @xmath95-expansion of @xmath34
. if @xmath34 is a @xmath63-structure , and @xmath99 are relations over @xmath2 , such that @xmath100 , @xmath101 , and @xmath102 , we write @xmath103 to indicate that @xmath35 is a @xmath95-expansion of @xmath34 , such that @xmath104 , @xmath101 , and @xmath105 , @xmath106 .
let @xmath34 be a @xmath63-structure and @xmath107 .
then @xmath108 $ ] is the _ substructure of @xmath34 induced by @xmath109 _
, where @xmath108 $ ] has universe @xmath109 , for each relation symbol @xmath110 we have @xmath111 } = r^{{\ensuremath{\mathscr{a } } } } \cap \bar a^{{\ensuremath{{\it arity}}}(r)}$ ] , and nullary symbols @xmath76 are interpreted as @xmath112 } = c^{{\ensuremath{\mathscr{a}}}}$ ] if @xmath113 and become uninterpreted otherwise .
two @xmath63-structures @xmath34 and @xmath114 over the same vocabulary @xmath63 are _ isomorphic _ , denoted by @xmath115 , if there is an _ isomorphism _ @xmath116 , where @xmath117 is a bijection between @xmath2 and @xmath118 and * @xmath119 if and only if @xmath120 for all @xmath98 , * @xmath121 for every nullary symbol @xmath122 , and * for every relation symbol @xmath110 and @xmath123 , where @xmath124 , @xmath125 we call two @xmath63-structures @xmath126 and @xmath127 _ compatible _ , if for all nullary symbols @xmath128 we have @xmath129 and the identity @xmath130 is an isomorphism between @xmath131 $ ] and @xmath132 $ ] . in this case , we define the _ union _ of @xmath126 and @xmath127 , denoted by @xmath133 , as the @xmath63-structure with universe @xmath134 and interpretations @xmath135 for every relation symbol @xmath110 .
nullary symbols @xmath98 with @xmath136 remain uninterpreted in @xmath133 ; otherwise @xmath137 if @xmath138 for some @xmath139 .
tree decompositions and treewidth were introduced by robertson and seymour @xcite in their works on the graph minors project , cf .
@xcite .
a _ tree decomposition _ of a relational @xmath63-structure @xmath34 is a tuple @xmath140 , where @xmath141 is a rooted tree and @xmath142 is a collection of subsets @xmath143 , such that * @xmath144 , * for all @xmath145-ary relation symbols @xmath110 and all @xmath146 , there is an @xmath147 such that @xmath148 , and * for all @xmath149 , if @xmath150 is on the path between @xmath151 and @xmath152 in @xmath47 , then @xmath153 . the sets @xmath154 are called _
bags_. the _ width _ of a tree decomposition is the size of its largest bag minus one , and the _ treewidth _ of a structure @xmath34 is the minimum width of all tree decompositions of @xmath34 . without loss of generality , we assume that each tree decomposition we consider is _ nice_. nice tree decompositions are directed , where each edge in @xmath155 has a direction away from the root , and have the following properties : each node @xmath147 has at most two children . for leafs @xmath147
, we have @xmath156 .
if @xmath150 has exactly one child @xmath157 , then there is @xmath158 such that either @xmath159 or @xmath160 . in the former case ,
we say @xmath150 is an _ introduce _ node , in the latter case we call @xmath150 a _ forget _ node of the tree decomposition . finally , if a node @xmath150 has two children @xmath151 and @xmath152 , then we require @xmath161 and call such nodes _ join _ nodes .
if @xmath162 is a directed path in @xmath47 pointing away from the root , we say @xmath157 appears _ below _
@xmath150 in @xmath47 . with every node @xmath147 of a ( nice ) tree decomposition of a @xmath63-structure @xmath34
we associate a substructure @xmath163 defined as follows : let @xmath164 be the set of objects in @xmath154 or in bags @xmath165 for nodes @xmath157 below @xmath150 in the tree decomposition
. then we let @xmath166 $ ] be the substructure of @xmath34 induced by @xmath167 . computing the treewidth of a graph is np - complete @xcite .
however , the algorithms in this paper rely on a given tree decomposition of the input structure . for graphs
@xmath15 , there is a fixed - parameter tractable algorithm @xcite with a running time of @xmath168 , whose dependence on the treewidth might become a problem in practical applications . in a practical setting , heuristics seem to work well and often nearly optimal tree decompositions can be computed @xcite . using _ gaifman graphs _ , one can also compute tree decompositions of arbitrary structures , cf .
, ( * ? ? ?
* section 11.3 ) . in the following ,
we therefore just assume a tree decomposition is given as part of the input .
for more information on treewidth , we refer the reader to surveys such as @xcite .
mso logic over a vocabulary @xmath63 , denoted by @xmath169 , is simultaneously defined over all vocabularies @xmath63 by induction .
firstly , for every @xmath145-ary relation symbol @xmath110 and any nullary symbols @xmath170 , @xmath169 contains the _ atomic _ formula @xmath171 . if @xmath29 is unary , we may abbreviate @xmath172 as @xmath173 . secondly : * if @xmath174 are in @xmath169 , then @xmath175 , @xmath176 , and @xmath177 are in @xmath169 , * if @xmath178 for some nullary symbol @xmath76 , then both , @xmath179 and @xmath180 are in @xmath169 .
this is called _ first order _ or _
object quantification_. * if @xmath181 for a unary relation symbol @xmath29 , then both , @xmath182 and @xmath183 are in @xmath169 .
the corresponding case is called _ second order _ or _ set quantification_. note that we do not distinguish between `` basic '' symbols ( contained in a certain `` base '' vocabulary such as @xmath184 ) , and symbols that are used as _ variables _ subject to quantification .
let @xmath63 be a vocabulary and @xmath185 be a formula .
let @xmath186 be the smallest vocabulary with @xmath187 .
then we call the symbols in @xmath188 the _ free _ symbols of @xmath8 .
let @xmath9 be the size of a suitable encoding of @xmath8 .
if @xmath189 for some @xmath76 , @xmath29 , @xmath190 , @xmath191 , and @xmath192 , we call @xmath8 _ universal_. similarly , we call @xmath8 _ existential _ if @xmath193 .
if @xmath8 does not contain set quantifiers , then we say @xmath8 is _ first order _ and contained in @xmath194 .
note that in particular all atomic formulas of @xmath169 are first order .
the _ quantifier rank _
@xmath195 of a formula @xmath196 denotes the maximum number of nested quantifiers in @xmath8 , counting both first order and second order quantifiers , and is defined by induction over the structure of @xmath8 as * @xmath197 if @xmath8 is an atomic formula , * @xmath198 , * @xmath199 if @xmath200 , and * @xmath201 if @xmath202 . without loss of generality ,
we assume throughout the paper that every formula is in negation normal form , i.e. , the negation symbol @xmath203 only occurs in front of atomic formulas
. this can be achieved by a simple rewriting of the formula .
for a fully interpreted @xmath63-structure @xmath34 and a formula @xmath204 , we write @xmath38 if and only if @xmath8 _ holds _ in @xmath34 or is _ true _ in @xmath34 in the classical sense , cf .
. we shall do not specify this further , since we will switch to a game - theoretic characterization in the remainder of this paper , cf .
, section [ sec : mcg ] . in @xcite , extended mso was introduced . here
, an mso - formula over a relational vocabulary is given together with an evaluation or optimization goal over the unary relation symbols ( set variables ) .
this principle was furthermore generalized to semiring homomorphisms in @xcite , where satisfying interpretations of the free relation symbols are to be translated into an appropriate semiring . in this paper
, we shall consider mso - definable linear optimization problems , also called @xmath205-definable optimization problems .
it is not hard to see that the methods in this paper extend to other classes of mso - definable problems , such as counting and enumeration problems .
see , e.g. , ( * ? ? ?
* chapter 6 ) for an overview of mso - definable problems and their algorithmic applications .
let @xmath63 be a relational vocabulary , @xmath206 be a set of unary relation symbols , @xmath204 , and @xmath207 .
let @xmath208 and @xmath209 .
then we call the problem of , given a @xmath95-structure @xmath34 , computing @xmath210 a _
@xmath205-definable optimization problem_. [ ex : mso - examples ] consider the following formulas : @xmath211 then , given a @xmath184-structure @xmath88 , @xmath212 encode the well known graph problems minimum vertex cover , minimum dominating set , and 3-colorability , respectively .
the semantics of mso in the classical sense ( cf .
@xcite ) can be characterized using a two player pebble game , called the _ hintikka game _ or _ model checking game _ , cf .
@xcite .
a pebble _ game _ @xmath213 between two players , say player 0 and player 1 , consists of a finite set @xmath214 of _ positions _ , two disjoint sets @xmath215 assigning positions to the two players , an _ initial position _
@xmath216 , and an acyclic binary relation @xmath217 , which specifies the valid _ moves _ in the game .
we only allow moves from positions assigned to one of the two players , i.e. , we require @xmath218 for all @xmath219 . on the other hand , we do allow that positions without outgoing moves are assigned to players .
let @xmath220 be the _ size _ of @xmath6 . for @xmath221 , we let @xmath222 be the set of positions reachable from @xmath145 via a move in @xmath223 . for any position @xmath224 we let @xmath225 be a _
subgame _ of @xmath6 , which is issued from the new initial position @xmath145 .
the set of all subgames of @xmath6 is denoted by @xmath226 .
if @xmath6 is clear from the context , we usually omit the subscript and write @xmath227 and @xmath228 .
a _ play _ of @xmath6 is a maximal sequence @xmath229 of positions @xmath230 , such that between any subsequent positions @xmath231 and @xmath232 there is a valid move , i.e. , @xmath233 for @xmath234 .
such a play is said to have @xmath235 _ rounds _ and to _ end _ in position @xmath236 .
the rules of the game are that in the @xmath150th round of the play , where @xmath237 , the player assigned to position @xmath231 has to place a valid _ move _
, i.e. , has to choose the next position @xmath238 .
if no such position @xmath232 exists , or the position @xmath231 is not assigned to either of the players , the play ends . if the play ends in a position @xmath236 with @xmath239 , where @xmath240 , then the other player , player @xmath241 , _ wins _ the play .
if , however , the play ends in a position @xmath236 with @xmath242 , then there is a draw and none of the players wins the play . the goal of game is to force the other player into a position where they can not move .
we say that a player has a _ winning strategy on @xmath6 _ , if and only if they can win every play of the game irrespective of the choices of the other player .
for instance , player 0 has a winning strategy on @xmath6 if and only if either * @xmath243 and there is a move @xmath244 such that player 0 has a winning strategy on @xmath245 ; or * @xmath246 and player 0 has a winning strategy on @xmath245 for all moves @xmath244 .
note that this includes the case that player 1 can not move at all . a game @xmath6 is said to be _ determined _ or _ well - founded _
if either one of the players has a winning strategy on @xmath6 , otherwise @xmath6 is _
undetermined_. we fix two special games @xmath247 and @xmath248 on which the first player and the second player , respectively , have winning strategies .
one can efficiently test whether one of the player has a winning strategy on a game @xmath6 , cf .
, @xcite .
algorithm [ alg : eval ] determines whether one of the players has a winning strategy on a game @xmath6 and returns either @xmath247 or @xmath248 if this is the case .
if none of the players has a winning strategy , the algorithm returns a corresponding `` proof '' , a list of all the plays of @xmath6 that ended with a draw .
algorithm @xmath249 @xmath250 [ then ] [ return ] @xmath6 let @xmath251 , @xmath252 , @xmath253 , and @xmath254 .
[ for ] @xmath255 [ do ] let @xmath256 .
update @xmath257 and @xmath258 , @xmath259 .
update @xmath260 .
let @xmath261 and compute @xmath262 . [ if ] @xmath263 [ then ] [ if ] @xmath264 [ or ] @xmath265 [ then ] [ return ] @xmath248 [ if ] @xmath266 [ then ] [ return ] @xmath247 [ if ] @xmath267 [ then ] [ if ] @xmath268 [ or ] @xmath265 [ then ] [ return ] @xmath247 [ if ] @xmath269 [ then ] [ return ] @xmath248 [ return ] @xmath270 in the case of the model checking game , we call the two players the _ falsifier _ and the _ verifier_. the verifier wants to prove that a formula is _ true _ on a structure ( or , the structure _ satisfies _ the formula ) , while the falsifier tries to show that it is _ false _ ( or , the structure does not satisfy the formula )
. the reader may therefore call @xmath248 `` true '' and @xmath247 `` false '' . [
def : model - checking - game ] the ( classical ) _ model checking game _ @xmath271 over a fully interpreted @xmath63-structure @xmath34 and a formula @xmath204 is defined by induction over the structure of @xmath8 as follows .
let @xmath272 , \varphi)$ ] , where @xmath273 . if @xmath8 is an atomic or negated formula , then @xmath274 , where * @xmath243 if and only if * * @xmath275 and @xmath276 , or * * @xmath277 and @xmath278 . * @xmath246 if and only if * * @xmath275 and @xmath278 , or * * @xmath277 and @xmath276 .
if @xmath279 for some relation symbol @xmath29 , let @xmath280 for @xmath281 be the @xmath282-expansion of @xmath34 with @xmath283 , and let @xmath284 be the corresponding model checking game over @xmath285 and @xmath190 .
then @xmath271 , where * @xmath286 , * @xmath287 , * @xmath288 , where @xmath289 iff @xmath290 and @xmath291 otherwise , * @xmath292 , where @xmath293 iff @xmath294 and @xmath295 otherwise .
if @xmath296 for some nullary symbol @xmath76 , let @xmath297 be the @xmath298-expansion of @xmath34 with @xmath299 , and let @xmath300 be the corresponding model checking game over @xmath301 and @xmath190 .
then @xmath271 , where * @xmath302 , * @xmath303 , * @xmath304 , where @xmath289 iff @xmath305 and @xmath291 otherwise , *
@xmath306 , where @xmath293 iff @xmath307 and @xmath295 otherwise .
if @xmath200 , let @xmath308 be the model checking game over @xmath34 and @xmath309 . then @xmath271 , where * @xmath310 , * @xmath311 , * @xmath312 , where @xmath289 iff @xmath313 and @xmath291 otherwise , * @xmath314 , where @xmath293 iff @xmath315 and @xmath295 otherwise .
note that the falsifier is the _
universal _ player and moves on universal formulas , while the verifier is the _ existential _ player and moves on existential formulas .
furthermore , if the structure @xmath34 is empty , then , by definition , @xmath316 and @xmath317 for all @xmath190 . in the model checking game , this corresponds to the case that there are no moves from the current position .
consequently , the play ends and the player assigned to this position looses . on non - empty structures
, each play ends in an atomic or negated atomic formula .
the goal of the verifier is to make the play end in a position @xmath318 with @xmath319 , and conversely the goal of the falsifier is to force the play into an ending position @xmath318 with @xmath320 .
it is well - known that the classical model checking game is well - founded @xcite and that the verifier has a winning strategy on @xmath321 if and only if @xmath38 , see , e.g. , @xcite .
we shall now consider an extension of the model checking game that has the following two central properties : * it is defined for partially interpreted structures ; and * it is `` well - defined '' under taking the union of structures in the sense that if one of the players has a winning strategy on the game on @xmath34 and @xmath8 , then the same player has a winning strategy in the game on @xmath322 and @xmath8 for all structures @xmath114 compatible with @xmath34 .
before we give the formal definition of the new game , let us briefly mention why we require these properties : recall that we want to use the model checking game @xmath321 to decide algorithmically whether a @xmath63-structure @xmath34 holds on a formula @xmath323 . if @xmath8 contains set quantifiers ,
then there is a number of positions in @xmath321 that grows exponentially with the size of @xmath2 . in order to avoid exponential running time on structures of bounded treewidth , a tree decomposition @xmath140 of @xmath34 , where @xmath324 , is traversed bottom - up by a dynamic programming algorithm . at a node
@xmath147 , we only consider the substructure @xmath163 of @xmath34 .
let @xmath35 be some expansion of @xmath34 .
then @xmath325 $ ] is in general not fully interpreted , which explains the first requirement . for
the second requirement , note that for each @xmath147 there is a @xmath63-structure @xmath326 , such that @xmath34 can be written as @xmath327 .
the structure @xmath326 is sometimes called the `` future '' of @xmath163 in the literature . therefore ,
if one of the players has a winning strategy in the game on @xmath163 and @xmath8 , we require that the same player has a winning strategy on @xmath327 and @xmath8 . in order to make the inductive construction work
, we additionally need to distinguish the nodes in the `` current '' bag @xmath154 of the tree decomposition .
the game therefore additionally depends on a given set @xmath328 .
[ def : extended - model - checking - game ] [ def : outcome ] the _ extended model checking game _ @xmath329 over a @xmath63-structure @xmath34 , a set @xmath330 , and a formula @xmath204 is defined by induction over the structure of @xmath8 as follows .
let @xmath331 , x , \varphi)$ ] , where @xmath273 . if @xmath8 is an atomic or negated formula , then @xmath332 , where * @xmath243 if and only if either * * @xmath275 , such that @xmath333 , and @xmath276 , or * * @xmath334 , such that @xmath333 , and @xmath278 . * @xmath246 if and only if either * * @xmath275 , such that @xmath333 , and @xmath278 , or * * @xmath334 , such that @xmath333 , and @xmath276 .
if @xmath279 for some relation symbol @xmath29 , or @xmath200 , then @xmath335 is defined analogously to @xmath321 .
if @xmath296 for some nullary symbol @xmath76 , let @xmath336 be the @xmath298-expansion of @xmath34 with @xmath337 , and let @xmath338 be the corresponding extended model checking game over @xmath339 and @xmath190
. then @xmath329 , where * @xmath340 , * @xmath341 , * @xmath342 , where @xmath289 iff @xmath305 and @xmath291 otherwise , * @xmath343 , where @xmath293 iff @xmath307 and @xmath295 otherwise . for the games we consider throughout this paper
, one can derive from a position @xmath221 whether @xmath344 or @xmath345 ( cf .
, the definitions of @xmath346 and @xmath347 ) . to avoid cluttered notation
, we shall therefore usually omit the sets @xmath348 and @xmath349 from the tuple @xmath350 and identify games with the triple @xmath351 .
figure [ fig : unfolding ] shows a simplified schematic of an extended model checking game and the result after an application of the evaluation algorithm @xmath352 . for the structure @xmath34 with universe @xmath353 and @xmath354 .
bottom : @xmath355 .
the lower branch witnesses a play that ends with a draw.[fig : unfolding ] [ fig : eval ] ] for the structure @xmath34 with universe @xmath353 and @xmath354 . bottom : @xmath355 .
the lower branch witnesses a play that ends with a draw.[fig : unfolding ] [ fig : eval ] ] if @xmath34 is a fully interpreted structure , @xmath321 can be embedded into @xmath335 such that for each play of @xmath321 there is a corresponding , equivalent play of @xmath335 .
algorithm [ alg : convert ] effectively computes this embedding ( lemma [ lem : convert - me ] ) .
furthermore , if @xmath335 is determined , then so is @xmath321 ( lemma [ lem : emcg - determines - mcg ] ) . [
lem : convert - me ] let @xmath34 be a fully interpreted @xmath63-structure , @xmath330 , and @xmath204 .
then , using algorithm [ alg : convert ] , we have @xmath356 the proof is an induction over the structure of @xmath8 . for atomic or negated atomic formulas ,
the statement trivially holds by definition of @xmath321 , since @xmath357 .
let @xmath358 .
let @xmath279 or @xmath200 and @xmath309 and consider @xmath359 .
we have @xmath360 , where either @xmath361 is an @xmath282-expansion of @xmath34 for some @xmath281 , or @xmath362 , respectively .
since @xmath34 is fully interpreted , @xmath35 is fully interpreted , and we obtain @xmath363 by the induction hypothesis .
if otherwise @xmath296 , consider @xmath359 . by definition , @xmath360 , where @xmath35 is a @xmath298-expansion of @xmath34 with @xmath364 .
if all constant symbols are interpreted in @xmath365 , then @xmath366 , i.e. , @xmath35 is fully interpreted . by the induction hypothesis
we get @xmath363 . together
, the statement follows .
algorithm @xmath367 @xmath250 [ then ] [ return ] @xmath6 .
let @xmath368 and @xmath369 .
let @xmath370 , \varphi)$ ] , @xmath371 , and @xmath252 .
[ for ] @xmath372 s.t . @xmath373 is fully interpreted [ do ] let @xmath374 .
update @xmath375 and @xmath376 .
[ return ] @xmath377 we now prove that if an extended model game is determined , then the corresponding player can win the game without using any further `` @xmath378-moves '' .
this will be useful in the proof of lemma [ lem : emcg - determines - mcg ] .
[ lem : partially - replaced ] let @xmath126 and @xmath127 be @xmath63-structures with @xmath379 and @xmath98 , such that @xmath380 , @xmath381 for all @xmath73 and @xmath382 for all @xmath383 .
let @xmath204 .
if @xmath384 , then @xmath385 and @xmath386 before we give the formal proof , consider the following high - level argument : suppose that @xmath387 . then there is at least one play of the game @xmath388 that is won by the verifier .
consider an arbitrary play @xmath229 won by the verifier and let @xmath389 .
since @xmath236 is assigned to the falsifier , all constant symbols occurring in @xmath190 are interpreted and hence different from @xmath76 .
the verifier can therefore win the game without depending on formulas where @xmath76 occurs .
the proof is an induction over the structure of @xmath8 .
let @xmath384 .
if @xmath8 is an atomic or negated formula , say @xmath275 , then @xmath390 .
therefore , @xmath385 and for all @xmath391 , we have @xmath392 and @xmath393 , which implies @xmath394 .
if @xmath279 for a relation symbol @xmath29 , let @xmath281 and @xmath395 , @xmath396 be the @xmath282-expansions of @xmath126 and @xmath127 , respectively , with @xmath397 .
then by the induction hypothesis @xmath398 if @xmath399 . similarly , if @xmath400 for a nullary symbol @xmath401 , let @xmath395 and @xmath396 be @xmath402-expansions of @xmath126 and @xmath127 , respectively , such that @xmath403
. then by the induction hypothesis @xmath404 , if @xmath399 . finally , if @xmath200 , then from @xmath405 , where @xmath309 , we get @xmath398 . together , the statement of the lemma follows .
we can now prove that if some player has a winning strategy in the extended model checking game , then the same player has a winning strategy in the classical model checking game .
[ lem : emcg - determines - mcg ] let @xmath34 be a fully interpreted @xmath63-structure , @xmath330 , and @xmath204 .
if @xmath406 , then @xmath407 the proof is an induction over the structure of @xmath8 .
suppose @xmath408 ( the case @xmath247 is shown analogously ) .
if @xmath8 is an atomic or negated atomic formula , then the statement clearly holds .
if @xmath313 , then for each @xmath309 we have @xmath409 .
this implies @xmath410 by the induction hypothesis , and therefore @xmath411 .
similarly , if @xmath290 for a relation symbol @xmath29 , then @xmath412 for each @xmath282-expansion @xmath35 of @xmath34 , each of which is fully interpreted .
we get @xmath413 by the induction hypothesis .
hence , @xmath411 .
if @xmath305 for a nullary symbol @xmath76 , then @xmath412 for each fully interpreted @xmath298-expansion @xmath35 of @xmath34 .
this implies @xmath413 by the induction hypothesis , and therefore @xmath411 . if @xmath315 , then there is @xmath309 with @xmath409 .
we get @xmath410 by the induction hypothesis , and therefore @xmath411 . similarly ,
if @xmath294 for a relation symbol @xmath29 , then there is a @xmath282-expansion @xmath35 of @xmath34 with @xmath412 .
since @xmath34 is fully interpreted , @xmath35 is fully interpreted . using the induction hypothesis
, we have @xmath413 and therefore @xmath411 . finally ,
if @xmath307 for a nullary symbol @xmath76 , then there is a @xmath298-expansion @xmath35 of @xmath34 with @xmath412 . by lemma [ lem :
partially - replaced ] , we can assume @xmath414 .
then @xmath35 is fully interpreted and we get @xmath413 by the induction hypothesis . therefore @xmath411 .
we can significantly strengthen this statement further : if @xmath335 is determined , then @xmath415 is also determined for _ all _ @xmath114 compatible with @xmath34 . note that the union @xmath322 arises on _ join _ or _
introduce _ nodes @xmath150 of the tree decomposition , where @xmath416 is the current bag , cf .
, figure [ fig : join ] .
recall , for instance , the example 3-colorability from the introduction : if a subgraph @xmath35 of a graph @xmath34 is not three - colorable , then clearly @xmath34 is not three - colorable either .
the following lemma formalizes this observation .
let us give a brief high - level explanation before we state the lemma and give its proof .
roughly speaking , if @xmath417 is determined , then moves to objects @xmath418 in @xmath419 are either `` irrelevant '' for a player s strategy or already `` sufficiently '' captured by moves to @xmath378 ( cf .
, lemma [ lem : partially - replaced ] ) .
if therefore one of the players , say the falsifier , has a winning strategy in @xmath6 , then in some sense this winning strategy carries over to @xmath270 . in the case of 3-colorability , if @xmath34 is not three - colorable , then the falsifier has a winning strategy on @xmath420 : no matter which three sets the verifier chooses , either these sets are not a partition or not independent sets . in either case
there are witnessing vertices that the falsifier can choose .
thus , no matter which subsets the verifier chooses in @xmath421 , the falsifier can then choose the same witnessing vertices to win each play of @xmath270 .
[ lem : emcg - eval - introduce ] let @xmath34 and @xmath114 be compatible @xmath63-structures with @xmath422 .
let @xmath330 and @xmath204 .
let @xmath417 and @xmath423 . 1 . if @xmath424 , then @xmath425 .
2 . if @xmath426 , then @xmath427 .
we prove the lemma by induction over the structure of @xmath8 .
let @xmath273 .
let @xmath428 and @xmath429 with @xmath430 and @xmath431 , where @xmath432 $ ] and @xmath433 $ ] .
suppose @xmath424 ( the second case @xmath434 is proven analogously ) .
let @xmath275 or @xmath435 for a relation symbol @xmath110 .
we have @xmath424 , and hence , by definition @xmath436 for all @xmath391 . here , @xmath437 for all @xmath391 , since @xmath34 and @xmath114 are compatible , and therefore @xmath438 , since @xmath439 . hence , @xmath440 if and only if @xmath441 , and thus @xmath425 .
assume now that @xmath313 or @xmath442 . by definition , for each @xmath309
there is a subgame @xmath443 and a subgame @xmath444 . by the induction hypothesis , @xmath445 if @xmath446 , and hence @xmath425 if @xmath424 .
if @xmath290 or @xmath447 , then for each @xmath281 there is a subgame @xmath448 , and for each @xmath449 there is a subgame @xmath450 . if @xmath290 , consider an arbitrary @xmath449 and let @xmath451 . we know , by definition of @xmath249 , that @xmath452 . furthermore , @xmath453 and @xmath454 are compatible , and therefore , by the induction hypothesis , also @xmath455 .
therefore , @xmath456 for all @xmath449 , and hence @xmath425 .
if otherwise @xmath294 , then there is some @xmath281 such that @xmath452 . since @xmath453 and @xmath457 are compatible , @xmath458 by the induction hypothesis .
therefore , @xmath425 .
if @xmath305 , consider an arbitrary @xmath298-expansion @xmath459 of @xmath114 and let @xmath460 $ ] .
note that if @xmath461 , then @xmath462 , and if @xmath463 or @xmath464 , then @xmath465 . in either case ,
@xmath35 and @xmath459 are compatible .
we know , by definition of @xmath249 , that @xmath412 .
hence , by the induction hypothesis , @xmath466 .
all in all , @xmath425 .
assume now that @xmath307 . since @xmath424 , we know that there is a @xmath298-expansion @xmath35 of @xmath34 , such that @xmath412 .
let @xmath459 the @xmath298-expansion of @xmath114 with @xmath467 .
then @xmath35 and @xmath459 are compatible , and using the induction hypothesis as above , we obtain @xmath425 . and @xmath114 are such that @xmath468 $ ] , then winning strategies for @xmath335 carry over to @xmath469 .
join / union ( right ) : if @xmath34 and @xmath114 are compatible , then winning strategies for @xmath335 carry over to @xmath470 .
[ fig : join ] , title="fig : " ] and @xmath114 are such that @xmath468 $ ] , then winning strategies for @xmath335 carry over to @xmath469 .
join / union ( right ) : if @xmath34 and @xmath114 are compatible , then winning strategies for @xmath335 carry over to @xmath470 .
[ fig : join ] , title="fig : " ] [ cor : introduce - join ] let @xmath34 and @xmath114 be compatible @xmath63-structures with @xmath471 .
let @xmath330 and @xmath204 .
let @xmath417 and @xmath472 . 1 . if @xmath424 , then @xmath425 .
2 . if @xmath426 , then @xmath427 .
we use lemma [ lem : emcg - eval - introduce ] and induction over @xmath473 .
let @xmath474 . if @xmath475 and therefore @xmath476 , the statement clearly holds . otherwise , consider @xmath418 and let @xmath477 $ ] .
from @xmath474 we get @xmath478 by lemma [ lem : emcg - eval - introduce ] .
we can now use the induction hypothesis on @xmath35 , @xmath114 and @xmath479 , since @xmath35 and @xmath114 a compatible and @xmath480 , and obtain @xmath481 .
the _ forget _ operation at a node @xmath150 of a tree decomposition does not change the underlying structure @xmath163 .
it is therefore not surprising that any winning strategies carry over .
[ lem : emcg - eval - forget ] let @xmath34 be a @xmath63-structure , @xmath482 and @xmath204 .
let @xmath417 and @xmath483 . 1 . if @xmath424 , then @xmath425 .
2 . if @xmath426 , then @xmath427 .
let @xmath484 and @xmath485 .
it is not hard to see that @xmath6 and @xmath270 are almost identical , the only difference being slightly differently labeled positions : by definition , @xmath368 and @xmath486 , where @xmath487 $ ] and @xmath488 $ ] . in particular , @xmath489 if and only if @xmath490 , where @xmath139 . by induction over the structure of @xmath8 ,
the claim then easily follows .
finally we show that the same holds for _ join _ nodes of a tree decomposition .
note that the corresponding operation on structures is the union .
[ lem : emcg - eval - union ] let @xmath491 be compatible @xmath63-structures , @xmath492 , and @xmath204 .
let @xmath417 and @xmath493 1 . if @xmath424 , then @xmath425 .
2 . if @xmath426 , then @xmath427 .
let @xmath406 . by corollary [ cor :
introduce - join ] , @xmath494 . the claim then immediately follows by lemma [ lem : emcg - eval - forget ] .
in this section we show that for every game @xmath495 one can construct a game @xmath496 such that @xmath497 if @xmath474 , but @xmath498 and @xmath499 are typically much smaller than @xmath214 and @xmath223 . this will be crucial for obtaining the desired running times of our algorithm .
we first define a suitable notion of _ equivalence _ between games .
[ def : equivalent - games ] we say that two positions @xmath500 are _ equivalent _ , denoted by @xmath501 iff * @xmath502 and @xmath503 for some formula @xmath8 and set @xmath504 , * there is an isomorphism @xmath505 between @xmath373 and @xmath506 , such that @xmath507 for all @xmath508 .
we say that two games @xmath509 and @xmath510 are _ equivalent _ , denoted by @xmath511 , if @xmath501 and there is a bijection @xmath512 , such that @xmath513 for all @xmath514 . algorithm @xmath515 @xmath250 [ then ] [ return ] @xmath6 [ if ] @xmath8 is an atomic or negated atomic formula [ then ] [ return ] @xmath249 let @xmath516 and @xmath517 . [ for ] @xmath518 [ do ] let @xmath519 . [ if ] @xmath8 is universal [ and ] @xmath520 [ then ] [ return ] @xmath247 [ if ] @xmath8 is existential [ and ] @xmath521 [ then ] [ return ] @xmath248 [ if ] @xmath522 [ and ] @xmath523 for all @xmath524 [ then ] update @xmath375 and @xmath525 .
[ if ] @xmath251 [ then ] [ return ] @xmath526 .
[ return ] @xmath527 we now define a _ reduce operation _ that significantly shrinks the size of a game @xmath6 ( see algorithm [ alg : reduce ] ) .
firstly , subgames won by the opponent player are removed .
if , for instance , the formula is universal , then the falsifier can safely ignore subgames that evaluate as @xmath248 , i.e. , for which the verifier has a winning strategy .
for example , it is easy to see that we can remove the two subgames @xmath248 and @xmath247 in figure [ fig : eval ] .
secondly , we only need to keep one representation per equivalence class under @xmath528 for all undetermined games . here
, we use the fact that @xmath529 for any @xmath530 with @xmath511 .
we will not explicitly prove this claim . if , however , @xmath531 and @xmath532 for some @xmath63-structures @xmath126 and @xmath127 , for @xmath533 and @xmath204 , then the bijection @xmath534 induced by the definition of @xmath528 yields a bisimulation between @xmath388 and @xmath535 . in particular ,
if both @xmath536 and @xmath537 are subgames of the same game @xmath88 , then it suffices to keep either subgame as `` witness '' for possible winning positions for the respective player in the model checking game .
thus , removing equivalent subgames from a game @xmath6 can be seen as a variant of taking the _ bisimulation quotient _ ( cf .
, ( * ? ? ?
* chapter 7 ) ) of @xmath6 .
see figures [ fig : vc - unfolding ] and [ fig : ds - unfolding ] in section [ sec : concrete - problems ] for two examples .
[ lem : eval - equals - reduce ] let @xmath34 be a @xmath63-structure , @xmath330 , and @xmath204 .
let @xmath417 .
then * @xmath424 , if and only if @xmath538 , and * @xmath426 , if and
only if @xmath539 . let @xmath540 , where @xmath368 .
without loss of generality , we assume that @xmath541 .
we only show the first case ( ) , the second statement is proven analogously .
the proof is an induction over the structure of @xmath8 .
if @xmath8 is an atomic or negated atomic formula or @xmath542 , then the statement holds by definition of @xmath515 . for the induction step ,
assume @xmath8 is not an atomic or negated formula , and @xmath543 .
let @xmath544 for all @xmath518 and let @xmath424 . if @xmath8 is existential , then there is @xmath518 with @xmath545 . by the induction hypothesis , @xmath546 , and therefore @xmath538
similarly , if @xmath8 is universal , then @xmath545 for all @xmath518 . by the induction hypothesis , @xmath547 for each @xmath518 .
hence , we have @xmath251 after the for - loop . since @xmath8 is universal , the call to @xmath526 returns @xmath248 by definition , and therefore @xmath538
. conversely , let @xmath538 . if @xmath8 is existential , then there must be some @xmath518 with @xmath547 . assume for a contradiction that @xmath548 for all @xmath518 .
then @xmath251 after the for - loop , which implies @xmath549 , a contradiction .
let therefore @xmath145 be such a position with @xmath547 .
then , by the induction hypothesis , @xmath545 for this @xmath145 , and therefore also @xmath424 .
if @xmath8 is universal , then we know @xmath251 after the for - loop , as this is the only possibility how @xmath515 can return @xmath248 . therefore , @xmath547 for all @xmath518 , and hence @xmath424 by the induction hypothesis and definition of @xmath249 .
now we prove an upper bound for the size of a reduced game .
since this is a general upper bound for arbitrary formulas and structures , we can not expect better bounds than the known lower bounds ( unless @xmath550 ) @xcite .
let @xmath63 be a vocabulary and @xmath204 .
let @xmath551 be two @xmath63-structures and @xmath533 .
we call @xmath126 and @xmath127 _ equivalent with respect to @xmath8 and @xmath21 _ , denoted by @xmath552 , if @xmath553 . for an arbitrary set @xmath21 of objects ,
we let @xmath554 be the set of all @xmath63-structures that contain @xmath21 , and @xmath555 the set of equivalence classes of @xmath556 under @xmath557 .
we let @xmath558 [ lem : number ] let @xmath63 be a vocabulary , @xmath204 , and @xmath21 be a set of objects .
then @xmath559 where @xmath9 is the length of an encoding of @xmath8 .
without loss of generality , we assume @xmath63 is minimal such that @xmath204 and therefore @xmath560 .
we prove the claim by induction over the structure of @xmath8 .
if @xmath8 is an atomic or negated atomic formula , let @xmath273 , and @xmath561 .
let @xmath562 .
then either @xmath563 , or @xmath564 , where @xmath368 and @xmath487 $ ] .
hence , @xmath565 depends on the number of non - isomorphic structures on at most latexmath:[$n : = |x| + objects .
for a fixed relation symbol @xmath110 , there are @xmath567 ways to choose the interpretation @xmath568 .
the total number of non - isomorphic @xmath63-structures over at most @xmath32 objects is therefore bounded by @xmath569 . if @xmath313 or @xmath570 , then @xmath571 and @xmath572 .
furthermore , by the induction hypothesis we get @xmath573 .
we conclude that @xmath574 . if @xmath575 , then @xmath576 , @xmath577 , and , by the induction hypothesis , @xmath578 .
since @xmath579 ignores equivalent subgames , the total number @xmath565 is upper - bounded by @xmath580 .
[ lem : size ] let @xmath34 be a @xmath63-structure , @xmath330 and @xmath204 .
then @xmath581 where @xmath9 is the length of an encoding of @xmath8 .
we use induction over the structure of @xmath8 . if @xmath8 is an atomic or negated atomic formula , then @xmath417 contains only a single position and @xmath582 . if @xmath313 or @xmath315 , let ,
for @xmath139 , be @xmath583 . by the induction hypothesis , @xmath584 where @xmath585 and @xmath572 , and therefore , @xmath586 if otherwise @xmath575 , then @xmath576 and @xmath577 .
since equivalent subgames are ignored , @xmath587
in this section , we show how model checking games on structures can be computed inductively .
we will introduce two algorithms : algorithm [ alg : combine ] will be used when structures are _ combined _ , i.e. , taking the union of two compatible structures .
this happens at _ join _ and _ introduce _ nodes of the tree decomposition .
algorithm [ alg : forget ] will be used when objects are removed from the set @xmath21 , which happens at _ forget _ nodes of the tree decomposition .
we first will study the case of combining games .
the next lemma is required for technical reasons .
algorithm @xmath588 let @xmath589 , @xmath590 and @xmath591 . [ for ]
each @xmath592 [ do ] let @xmath593 and @xmath594 .
[ if ] @xmath595 [ and ] @xmath596 and @xmath597 are compatible [ then ] let @xmath598 .
update @xmath599 and @xmath600 .
[ return ] @xmath601 [ lem : matches - compatible ] let @xmath126 and @xmath127 be compatible @xmath63-structures , @xmath204 and let @xmath602 and @xmath603 with @xmath604 .
let , for @xmath139 , @xmath605 and @xmath606 , where @xmath607
. then @xmath373 and @xmath506 are compatible .
let @xmath273 .
since @xmath608 , we have , by definition [ def : equivalent - games ] , @xmath609 $ ] for an isomorphism @xmath610 with @xmath611 for all @xmath612 . by definition , @xmath613 , and therefore @xmath138 if and only if @xmath614 .
if @xmath615 , then in particular @xmath616 . hence , @xmath617 . since @xmath126 and @xmath127 are compatible , @xmath129 for all @xmath618 , and therefore @xmath619 for all @xmath615 .
accordingly , @xmath373 and @xmath506 are compatible .
we now prove that for a structure @xmath34 with @xmath620 the reduced model checking game @xmath621 can , up to equivalence , be computed from @xmath622 and @xmath623 . here , @xmath624 essentially computes the cartesian product of plays in the games over @xmath126 and @xmath127 , respectively .
this is possible because each set @xmath281 can be split into @xmath625 and @xmath626 , such that @xmath627 .
similarly , each interpretation of a nullary symbol is either @xmath378 , or contained in @xmath628 , in @xmath629 , or in @xmath630 ( cf .
, figure [ fig : join ] ) .
these cases can be reconstructed from the respective subgames on @xmath126 and @xmath127 .
[ lem : combine ] let @xmath126 and @xmath127 be compatible @xmath63-structures , @xmath204 and let @xmath602 and @xmath603 with @xmath604 .
let , for @xmath139 , @xmath605 and @xmath608 .
then @xmath631 the proof is an induction over the structure of @xmath8 .
let @xmath632 , @xmath633 , and @xmath273 .
let @xmath634 and @xmath635 .
let , for @xmath139 , @xmath636 and @xmath637 . by lemma [ lem : matches - compatible ] , @xmath373 and @xmath506 are compatible .
furthermore , @xmath638 \cong { \ensuremath{\mathscr{h}}}_i$ ] , and thus @xmath639 = { \ensuremath{\mathscr{a}}}_1[x_1 \cup \bar c^{{\ensuremath{\mathscr{a}}}_1 } ] \cup { \ensuremath{\mathscr{a}}}_2[x_2 \cup \bar c^{{\ensuremath{\mathscr{a}}}_2 } ] \cong { \ensuremath{\mathscr{h}}}_1 \cup { \ensuremath{\mathscr{h}}}_2 $ ] .
if @xmath640 , then @xmath641 , x , \varphi)$ ] . therefore , @xmath642 , x , \varphi ) \cong ( { \ensuremath{\mathscr{h}}}_1 \cup { \ensuremath{\mathscr{h}}}_2 , x_1 \cup x_2 , \varphi ) = p_{\mathcal g}.\ ] ] let @xmath8 be an atomic or negated atomic formula . if @xmath643 the lemma already holds with above considerations
. therefore consider the case @xmath644 , say @xmath645 .
then @xmath646 by lemma [ lem : eval - equals - reduce ] .
therefore , @xmath645 if and only if the verifier wins the play @xmath647 , where @xmath648 is the initial position of @xmath649 , x , \varphi)$ ] .
the claim then follows , since @xmath331 , x , \varphi ) \cong p_{\mathcal g}$ ] , where in particular @xmath639 \cong { \ensuremath{\mathscr{h}}}_1 \cup { \ensuremath{\mathscr{h}}}_2 $ ] and @xmath633 . for the induction step ,
we distinguish the following cases . [ [ case - varphi - psi_1-wedge - psi_2-or - varphi - psi_1-vee - psi_2 ] ] case @xmath313 or @xmath650 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let , for @xmath309 , @xmath651 and , for each @xmath139 , be @xmath652 .
consider @xmath309 with @xmath653 and suppose there was @xmath139 , say @xmath654 , with @xmath655 .
let @xmath656 and @xmath657 .
by lemma [ lem : eval - equals - reduce ] , @xmath658 , and therefore by corollary [ cor : introduce - join ] , @xmath659 . since @xmath660 , also @xmath661 .
this contradicts @xmath662 via lemma [ lem : eval - equals - reduce ] .
therefore , we have @xmath663 for each @xmath139 , which implies @xmath664 . since @xmath608 for @xmath139 , there is @xmath665 with @xmath666 .
the algorithm @xmath667 will eventually call @xmath668 . then
, by the induction hypothesis , @xmath669 contains the required subgame @xmath670 .
conversely , let @xmath309 and @xmath671 such that @xmath588 recursively calls @xmath668 . from @xmath608
we get @xmath666 .
then @xmath672 by the induction hypothesis . together
, the statement of the lemma follows .
[ [ case - varphi - forall - r - psi - or - varphi - exists - r - psi ] ] case @xmath290 or @xmath673 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + consider an arbitrary @xmath281 and let @xmath674 , where @xmath361 with @xmath675 .
for @xmath139 , let @xmath676 and @xmath677 , where @xmath678 .
if @xmath679 , then @xmath680 for each @xmath139 by using a combination of lemma [ lem : eval - equals - reduce ] and corollary [ cor : introduce - join ] .
therefore , @xmath681 .
since @xmath608 , there is @xmath682 with @xmath683 .
let @xmath684 .
since @xmath395 and @xmath396 are compatible and @xmath685 , we by lemma [ lem : matches - compatible ] have that @xmath596 and @xmath597 are compatible .
therefore , the algorithm eventually recursively calls @xmath686 . by the induction hypothesis , @xmath687 .
conversely , assume the algorithm recursively calls @xmath686 , where @xmath682 for each @xmath139 .
from @xmath608 we get @xmath688 , where @xmath678 for some @xmath689 .
. since @xmath373 and @xmath506 are compatible and @xmath690 , also @xmath395 and @xmath396 are compatible .
therefore , the induction hypothesis implies @xmath691 , where @xmath692 with @xmath693 .
together , the statement of the lemma follows .
[ [ case - varphi - forall - c - psi - or - varphi - exists - c - psi ] ] case @xmath305 or @xmath307 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + consider a @xmath298-expansion @xmath35 of @xmath34 and let @xmath674 .
let , for @xmath139 , @xmath694 $ ] be the @xmath298-expansion of @xmath163 with @xmath695 if @xmath696 , and @xmath697 otherwise .
let @xmath677 .
if @xmath679 , then @xmath698 by a combination of lemma [ lem : eval - equals - reduce ] and corollary [ cor : introduce - join ] .
therefore , @xmath664 .
since @xmath608 , there is @xmath682 with @xmath685 .
let @xmath684 .
since @xmath395 and @xmath396 are compatible and @xmath683 , lemma [ lem : matches - compatible ] implies that @xmath596 and @xmath597 are compatible .
the algorithm therefore eventually calls @xmath686 . by the induction hypothesis , @xmath687 .
conversely , assume the algorithm recursively calls @xmath686 , where @xmath682 for each @xmath139 .
from @xmath608 we get @xmath699 for some @xmath298-expansion @xmath700 of @xmath163 .
since @xmath373 and @xmath506 are compatible and @xmath690 , also @xmath395 and @xmath396 are compatible . by the induction hypothesis , @xmath691 , where @xmath701 .
together , the statement of the lemma follows .
algorithm @xmath702 there is @xmath703 with @xmath704 [ then ] let @xmath705 [ else ] let @xmath706 , x \setminus \{x\ } , \varphi)$ ] .
let @xmath371 and @xmath252 .
[ for ] each @xmath707 [ do ] let @xmath708 .
set @xmath257 and @xmath709 .
[ return ] @xmath710 [ lem : forget ] let @xmath34 be a @xmath63-structure , @xmath330 and @xmath711 .
let @xmath204 and @xmath712 .
then @xmath713 we use induction over the structure of @xmath8 .
let @xmath273 , @xmath714 , @xmath715 .
let @xmath716 with @xmath717 . here
, @xmath718 $ ] , since @xmath719 . if @xmath8 is an atomic or negated atomic formula and @xmath679 , the statement holds since @xmath720 , x ' , \varphi)$ ] by definition .
if otherwise @xmath8 is an atomic or negated atomic formula and @xmath721 , let @xmath722 $ ] if @xmath723 for all @xmath703 , and @xmath724 otherwise .
if @xmath721 , then @xmath725 by lemma [ lem : eval - equals - reduce ] .
since @xmath726 $ ] , we have @xmath727 and @xmath728 for the induction step , let @xmath707 be an arbitrary subgame of @xmath6 .
since @xmath729 , we know that @xmath730 for some expansion @xmath35 of @xmath34 and subformula @xmath190 of @xmath8 . by the induction hypothesis , @xmath731 conversely , if @xmath732 is a subgame of @xmath733 , then @xmath734 .
this implies @xmath735 by lemmas [ lem : emcg - eval - forget ] and [ lem : eval - equals - reduce ] .
therefore , there is @xmath707 with @xmath730 . by the induction hypothesis , @xmath736 .
together , the statement of the lemma follows .
finally , we come back to algorithm [ alg : convert ] and show that its correctness translates to reduced games .
[ lem : convert ] let @xmath34 be a fully interpreted @xmath63-structure , @xmath330 , and @xmath204 .
let @xmath729 .
then @xmath737 we prove the statement by induction over the structure of @xmath8 .
recall that @xmath738 is determined and hence @xmath739 .
if @xmath250 , then @xmath740 .
we get @xmath741 from lemma [ lem : eval - equals - reduce ] and therefore , using lemmma [ lem : emcg - determines - mcg ] for the first equality , @xmath742 let therefore @xmath743 with @xmath368 and suppose @xmath744 ( the case @xmath247 is shown analogously ) . for atomic or negated atomic formulas , the statement holds since , by definition , @xmath745 , and hence @xmath746 by lemma [ lem : convert - me ] .
if @xmath279 , say @xmath290 , consider @xmath281 and let @xmath361 with @xmath675 . if there is @xmath707 with @xmath730 , then @xmath747 by the induction hypothesis
if otherwise there is no such @xmath270 in @xmath226 , then @xmath748 by definition of @xmath579 , since @xmath541 . by lemmas [ lem : eval - equals - reduce ] and [ lem : convert - me ] , we then conclude @xmath413 .
together , the lemma follows . similarly , if @xmath749 , then for @xmath309 either there is @xmath707 , such that @xmath750 , or there is no such @xmath270 contained in @xmath226 . in the former case we again obtain @xmath751 by the induction hypothesis , and in the latter case
we can again argue that @xmath752 .
finally , let @xmath296 .
for any @xmath158 and @xmath753 , where @xmath754 , we argue analogously to the previous cases that either there is @xmath755 , such that @xmath756 , or there is no such @xmath270 contained in @xmath757 , which implies @xmath758 .
hence , consider the @xmath298-expansion @xmath35 of @xmath34 with @xmath465 . if there is @xmath759 with @xmath730 , then @xmath522 .
in particular , @xmath760 , where @xmath761 is not fully interpreted . therefore , @xmath367 removes the subgame @xmath270 from @xmath6 . in either case ,
@xmath367 does only contain subgames where @xmath76 has been interpreted as an object in @xmath2 , as considered above .
together , the statement of the lemma then follows .
we can now reprove courcelle s theorem for @xmath205-definable optimization problems . throughout this section , we shall abbreviate @xmath762 .
[ thm : main ] fix a relational vocabulary @xmath63 , a set @xmath763 of unary relation symbols , and @xmath207 .
let @xmath204 , and @xmath764 be constants . given a @xmath95-structure @xmath34 together with a tree decomposition @xmath140 of @xmath34 having width at most @xmath1 , where @xmath141 and @xmath142
, one can compute @xmath210 in time @xmath765 .
the remainder of this section is devoted to the proof of this theorem .
we give an algorithm that essentially works as follows : in a first phase , the algorithm uses dynamic programming on the tree decomposition ( based on lemmas [ lem : alg - leaf][lem : alg - join ] ) to compute the reduced extended model checking games @xmath766 and the values @xmath767 for _ all _ structures @xmath768 where @xmath769 for @xmath101 .
note that by the previous sections the algorithm does not need to distinguish between equivalent games . in a second phase
, the algorithm tests whether the verifier has a winning strategy on @xmath770 , or , in other words ( lemma [ lem : convert ] ) , whether @xmath771 .
the algorithm then collects the values @xmath767 for all @xmath772 with @xmath773 and outputs the optimal one . since most of the games considered are equivalent ( lemma [ lem : number ] )
, we can obtain the desired run time bounds . without loss of generality
, we assume @xmath774 . recall that for each @xmath147
, @xmath163 is the substructure of @xmath34 induced by those objects that appear at or below @xmath150 in the tree decomposition .
let , for @xmath147 , @xmath775 be the set of possible interpretations of the free relation symbols @xmath776 in @xmath163 , @xmath777 be the set of their corresponding @xmath63-expansions of @xmath163 , where for each @xmath778 the symbol @xmath779 is interpreted as @xmath780 , and @xmath781 be the corresponding extended model checking games in their reduced form .
we let @xmath782 and @xmath783 be the restriction of @xmath784 to @xmath154 , and let , for @xmath785 , @xmath786 be the set of @xmath63-expansions of @xmath34 that `` match '' @xmath787 on @xmath154 .
let @xmath788 be the corresponding games , and , for arbitrary games @xmath789 , @xmath790 and @xmath791 finally , we let , for @xmath792 , @xmath793,\ ] ] where @xmath794 for each @xmath101 , and @xmath795 we use dynamic programming on the tree decomposition as follows . as usual , we associate with each node @xmath147 of the tree decomposition a _ table _ @xmath796 that contains feasible , _ partial _ solutions and their corresponding _ value _
@xmath797 under the optimization function .
formally , we let @xmath798 map tuples @xmath799 to sets of _ feasible _ games over @xmath163 , i.e. , games @xmath789 with @xmath800 , and let @xmath801 be the corresponding values , where @xmath802 . initially , we let @xmath803 for all @xmath799 and @xmath804 for all @xmath805 .
[ [ phase-1 ] ] phase 1 + + + + + + + the algorithm traverses the tree decomposition bottom - up .
recall that each node @xmath147 is either a leaf , or of one of the three types _ introduce _ , _ forget _ , or _
join_. the algorithm distinguishes these four cases as follows . leaf : : let @xmath806 . for all @xmath807
the algorithm considers @xmath808 .
if @xmath809 , then the algorithm sets @xmath810 introduce : : let @xmath157 be the unique child of @xmath150 and @xmath811 for @xmath812 .
+ for each @xmath813 , and each @xmath814 such that @xmath815 , the algorithm considers each @xmath816
. + let @xmath817 if there is @xmath818 with @xmath819 , then let @xmath820 instead . + if @xmath821 , the algorithm sets @xmath822 forget : : let @xmath157 be the unique child of @xmath150 and @xmath823 for @xmath824 .
+ for each @xmath813 the algorithm considers each @xmath816 .
let @xmath825 and @xmath826 if there is @xmath818 with @xmath819 , then let @xmath820 instead .
if now @xmath821 , the algorithm sets @xmath827 and @xmath828 where @xmath829 as defined in section [ sec : prel ] .
join : : let @xmath830 be the children of @xmath150 . then @xmath831 .
+ for each @xmath785 the algorithm considers each pair @xmath832 .
let @xmath833 if there is @xmath818 with @xmath819 , then let @xmath820 instead .
if now @xmath821 , the algorithm sets @xmath827 and @xmath834 [ [ phase2 ] ] phase 2 + + + + + + + let @xmath835 and @xmath836 the algorithm starts with @xmath837 and considers each @xmath838 .
if @xmath839 , then the algorithm updates @xmath840 finally , the algorithm outputs @xmath841 . in order to show that the algorithm is correct and computes the optimal solution , we use induction over the structure of the tree decomposition to show the following invariant . [ inv : alg ] after the algorithm has processed a node @xmath147 in phase 1 , for each @xmath785 we have that * for each @xmath842 with @xmath843 there is exactly one @xmath844 with @xmath845 , * for each game @xmath846 we have @xmath800 and @xmath847 , and * for each @xmath846 we have @xmath848 here , ( i ) guarantees that @xmath796 is _ complete _ , i.e. , @xmath849 contains games for all feasible partial solutions , ( ii ) guarantees that all games in @xmath849 do , in fact , correspond to a reduced game over some @xmath63-expansion of @xmath163 , and ( iii ) guarantees that we also compute the correct solution , i.e. , @xmath850 is optimal for @xmath851 .
note that the `` exactly one '' in ( i ) is required for the claimed running time , but not for the correctness of the solution .
[ lem : alg - leaf ] invariant [ inv : alg ] holds for leafs of the tree decomposition .
let @xmath147 be a leaf and @xmath852 .
since @xmath150 is a leaf , we have @xmath853 such that ( i ) and ( ii ) clearly hold .
furthermore , @xmath854 for all @xmath778 , since @xmath855 , and therefore @xmath856 for all @xmath805 .
[ lem : alg - introduce ] let @xmath147 be an introduce node of the tree decomposition and @xmath857 be the unique child of @xmath150 .
if invariant [ inv : alg ] holds for @xmath157 before the algorithm processes @xmath150 , then it also holds for @xmath150 .
let @xmath811 , where @xmath812 .
let @xmath814 and @xmath813 with @xmath815 .
consider @xmath858 and let @xmath859 $ ] .
if @xmath860 , then also @xmath861 by lemma [ lem : emcg - eval - introduce ] and lemma [ lem : eval - equals - reduce ] . by invariant [ inv : alg ] , @xmath862 therefore contains exactly one game @xmath863 with @xmath864 .
if @xmath865 , then @xmath866 by lemma [ lem : emcg - eval - introduce ] and lemma [ lem : eval - equals - reduce ] .
otherwise , the algorithm computes @xmath867 . by lemma [ lem :
combine ] , @xmath819 , which implies part ( i ) of the invariant .
conversely , consider @xmath868 .
then either @xmath866 and there is @xmath816 with @xmath869 , or there is @xmath816 with @xmath870 . by the invariant for @xmath157 , @xmath871 . from this
we get there is @xmath872 such that @xmath864 and @xmath873 for some @xmath874 .
let @xmath858 , chosen in a way such that @xmath875 .
if @xmath869 , then , by lemma [ lem : emcg - eval - introduce ] and lemma [ lem : eval - equals - reduce ] , @xmath876 . otherwise , @xmath877 by lemma [ lem : combine ] .
either case implies ( ii ) .
finally , let @xmath868 and @xmath878 with @xmath879 and @xmath880 let @xmath881 $ ] . by lemmas [ lem : emcg - eval - introduce ] and [ lem : eval - equals - reduce ] , @xmath882 .
therefore , either @xmath883 , or otherwise @xmath884 by lemma [ lem : combine ] .
we need that @xmath885 is optimal for @xmath886 . to this end , assume there was @xmath887 with @xmath873 , such that either @xmath865 or @xmath888 , and furthermore @xmath889 since , @xmath864 , we have , by lemma [ lem : combine ] , @xmath890 where @xmath891 and @xmath892
. therefore , @xmath893 a contradiction to the minimality of @xmath894 .
we conclude that @xmath885 is optimal for @xmath886 . from this
we get that @xmath895 by the invariant for @xmath157 , which implies ( iii ) .
[ lem : alg - forget ] let @xmath147 be a forget node of the tree decomposition and @xmath857 be the unique child of @xmath150 .
if invariant [ inv : alg ] holds for @xmath157 before the algorithm processes @xmath150 , then it also holds for @xmath150 .
let @xmath157 be the unique child of @xmath150 and @xmath823 for @xmath896 .
note that @xmath897 .
let @xmath814 .
consider @xmath858 with @xmath860 and let @xmath898 and @xmath899 .
then , by lemma [ lem : emcg - eval - introduce ] and lemma [ lem : eval - equals - reduce ] , @xmath900 . therefore , by the invariant for @xmath157 , there is @xmath901 with @xmath864 .
if @xmath902 , then , by lemma [ lem : emcg - eval - forget ] , also @xmath866 . otherwise , the algorithm computes @xmath903 .
either case implies ( i ) .
conversely , consider @xmath868 .
then either @xmath866 and there is @xmath904 and @xmath816 with @xmath869 and @xmath905 , or there is @xmath904 and @xmath816 , such that @xmath905 and @xmath906 . by the invariant for @xmath157 , in either case @xmath871 . therefore , there is @xmath907 , where @xmath864 and @xmath873 , for some @xmath874
. let @xmath908 . if @xmath869 , then , by lemmas [ lem : emcg - eval - forget ] and [ lem : eval - equals - reduce ] , @xmath876 . otherwise , @xmath909 according to lemma [ lem : forget ] .
either case implies ( ii ) . finally , consider @xmath868 and let @xmath878 such that @xmath910 and @xmath879 .
let @xmath911 .
then , by lemmas [ lem : emcg - eval - forget ] and [ lem : eval - equals - reduce ] , @xmath882 . by ( ii ) , there is @xmath901 with @xmath864 , where @xmath912 .
analogue to the previous case , we obtain that @xmath885 is optimal in @xmath913 .
therefore , by the induction hypothesis , @xmath914 which implies ( iii ) .
[ lem : alg - join ] let @xmath147 be a join node of the tree decomposition with children @xmath915 . if invariant [ inv : alg ] holds for @xmath151 and @xmath152 before the algorithm processes @xmath150 , then it also holds for @xmath150 .
note that @xmath831 .
let @xmath916 .
consider @xmath842 and let , for @xmath917 , be @xmath859 $ ] .
if @xmath860 , then , for @xmath917 , also @xmath861 by lemma [ lem : emcg - eval - union ] and lemma [ lem : eval - equals - reduce ] . by the invariant for @xmath917
, @xmath918 therefore contains exactly one @xmath863 with @xmath864 .
if @xmath865 , then @xmath866 by lemma [ lem : emcg - eval - union ] and lemma [ lem : eval - equals - reduce ] .
otherwise , the algorithm computes @xmath919 .
by lemma [ lem : combine ] , @xmath819 , which implies ( i ) .
conversely , consider @xmath920 .
then either @xmath866 and there is @xmath917 and @xmath921 with @xmath869 , or there is @xmath832 , such that @xmath922 . by the invariant for @xmath917 , we have @xmath871 , and therefore there is @xmath923 with @xmath864 and @xmath873 , where @xmath924 .
let @xmath925 , such that @xmath926 for all @xmath927 .
if @xmath928 , then , by lemmas [ lem : emcg - eval - union ] and [ lem : eval - equals - reduce ] , @xmath929
. otherwise , @xmath930 by lemma [ lem : combine ] .
either case implies ( ii ) .
now consider @xmath920 and @xmath931 with @xmath879 and @xmath932 let , for @xmath917 , @xmath881 $ ] .
then , by lemma [ lem : emcg - eval - union ] and lemma [ lem : eval - equals - reduce ] , @xmath882 .
therefore , either @xmath883 for some @xmath917 , or @xmath933 by lemma [ lem : combine ] .
assume there were @xmath917 , say @xmath934 , and @xmath935 with @xmath873 , such that @xmath936 and either @xmath865 or @xmath937 for some @xmath938 . since @xmath939 , structures @xmath940 are compatible .
by the invariant , part ( ii ) , we have @xmath941 for some @xmath942 . without loss of generality ,
we assume @xmath943 , since each @xmath944 with @xmath945 yields the same contradiction . therefore , since @xmath864 , we have @xmath946 by lemma [ lem : combine ] , where @xmath947 .
therefore , @xmath948 a contradiction to the minimality of @xmath894 . therefore , for @xmath917
, @xmath885 is optimal in @xmath949 , and @xmath895 by the invariant for @xmath157 . by ( ii ) , there is @xmath950 with @xmath864 , which then implies ( iii ) .
[ lem : alg - root ] let @xmath835 be the root of the tree decomposition , where @xmath951 , and let invariant [ inv : alg ] hold for @xmath952 .
let @xmath953 and @xmath954 be an optimal solution for the @xmath205-problem .
then @xmath955 note that @xmath956 .
let @xmath35 be optimal , i.e. , let @xmath35 be a @xmath63-expansion of @xmath34 , such that @xmath957 and @xmath958 let @xmath959 .
we have @xmath960 since @xmath957 , and therefore @xmath961 by lemma [ lem : convert ] .
note that @xmath951 and therefore @xmath962 . by invariant [ inv : alg ] ,
part ( i ) , there is @xmath963 , such that @xmath845 , which implies @xmath964 by part ( iii ) and the optimality of @xmath35 for @xmath965 .
since @xmath966 , we also have @xmath967 conversely , let @xmath968 , such that @xmath969 and @xmath970 by part ( ii ) of the invariant , there is a @xmath63-expansion @xmath971 of @xmath34 , such that @xmath972 . since @xmath969
, we have @xmath973 by lemma [ lem : convert ] . without loss of generality
, we can assume by part ( iii ) , that @xmath971 is optimal for @xmath965 , i.e. , @xmath974 .
we then directly conclude @xmath975 we can now prove theorem [ thm : main ] .
using induction over the structure of the tree decomposition and lemmas [ lem : alg - leaf][lem : alg - join ] for the respective nodes , we know that invariant [ inv : alg ] holds for the root node of the tree decomposition after the algorithm has finished phase 1 . by lemma [ lem : alg - root ] , the algorithm outputs the correct solution in phase 2 . for the running time , consider @xmath147 .
we have @xmath976 , which for constant @xmath977 and @xmath978 is a constant . for @xmath799 , consider the set @xmath849 . since the algorithm only inserts games into @xmath849 , if @xmath849 does not already contain an equivalent game , @xmath979 by lemma [ lem : number ] , which for bounded @xmath980 is constant .
furthermore , by lemma [ lem : size ] , for each @xmath846 , @xmath981 again a constant . finally , each position of each game is of the form @xmath982 , where @xmath983 and @xmath984 , where @xmath985 denotes the size of a suitable encoding of @xmath365 .
all operations on games , i.e. , @xmath579 , @xmath986 , @xmath987 , @xmath988 , and @xmath989 , therefore take constant time . in total , at a node @xmath147 , a constant number of entries or pairs , respectively , is considered , and each operation takes constant time .
the running time is therefore @xmath765 .
[ [ semiring - homomorphisms ] ] semiring homomorphisms + + + + + + + + + + + + + + + + + + + + + + note that the algorithm implicitly used a homomorphism @xmath990 from the semiring @xmath991 into the semiring @xmath992 . here
, @xmath993 is the set of all possible interpretations of the free relation symbols ( i.e. , a set of tuples of sets ) , @xmath994 is a component - wise , disjoint union with neutral element @xmath995 , and @xmath996 is the regular union of sets .
the extension to other semiring homomorphisms , e.g. , to count the number of interpretations satisfying the mso property @xmath8 , is rather straightforward .
see @xcite for a list of many interesting semirings .
[ [ many - sorted - structures ] ] many - sorted structures + + + + + + + + + + + + + + + + + + + + + + in this article , we considered one - sorted structures , i.e. , structures whose universe contains objects of a single sort only .
the corresponding theory is also called @xmath997-theory in the literature and is strictly less powerful than corresponding logics for multi - sorted structures .
for instance , recall from example [ ex : taugraph ] that a graph @xmath998 can in a natural way be identified with a structure over the vocabulary @xmath89 , where @xmath91 is identified with the one - sorted universe of _ vertices _ , and @xmath90 is interpreted as @xmath999 .
the hamiltonian path problem for graphs can not be expressed in @xmath1000 , since this requires the use of edge - set quantification ( see @xcite , for instance ) .
fortunately , this poses no restriction in algorithmic applications .
firstly , it is not hard to extend the techniques in this paper to many - sorted structures .
original works @xcite were already proven for many - sorted structures .
secondly , one can easily simulate many - sorted structures by introducing relation symbols that distinguish the respective objects in a common universe accordingly .
for example , one can consider the incidence graph of a graph and introduce unary relation symbols @xmath91 and @xmath999 , which allow to distinguish objects of sort `` vertex '' or `` edge '' , and a new binary relation symbol @xmath1001 for the incidence relation .
transforming a structure and a corresponding tree decomposition accordingly can be done efficiently and does not increase the width of the decomposition .
graphs with multi - edges can be represented similarly .
in the analysis of the running time of the algorithm , we were rather pessimistic w.r.t . the constants hidden in the @xmath765 .
recall that unless @xmath1002 , these can not be bounded by an elementary function , i.e. , the running time of the algorithm can not be @xmath1003 for a fixed function @xmath1004 that is a nesting of exponentials of bounded depth @xcite .
the picture changes dramatically once we assume the _ problem _ is fixed , i.e. , the problem description consisting of the vocabulary @xmath63 , a formula @xmath204 and the integers @xmath1005 are constants .
specialized and comparably efficient algorithms exist for many problems , e.g. , of running time @xmath1006 for the minimum vertex cover problem , or of @xmath1007 for minimum dominating set and 3-colorability , cf .
@xcite , where @xmath1008 is a fixed polynomial in @xmath1 .
recent results furthermore indicate that better running times are improbable @xcite . assuming small treewidth , such algorithms might still turn out to be feasible in many practical applications , cf .
@xcite . in this section
, we estimate the running times of our generic approach for the three aforementioned problems .
let @xmath1009 be a tree decomposition of the input graph structure @xmath34 over @xmath184 , where @xmath141 and @xmath142 with @xmath1010 for all @xmath147 , i.e. , @xmath34 has treewidth at most @xmath1011 . , where @xmath34 has universe @xmath353 , @xmath1012 and @xmath1013 .
if any of the symbols @xmath37 or @xmath1014 remains uninterpreted ( cases @xmath36 and @xmath1015 in the figure ) , then some of the plays in @xmath1016 end with a draw and still persist in the reduced game .
if @xmath1017 and @xmath1018 for a node @xmath150 of a tree decomposition , then this essentially means that it is still open whether nodes in the `` future '' of @xmath150 will be adjacent or whether they will be contained in @xmath29.[fig : vc - unfolding ] ] recall from example [ ex : mso - examples ] that the formula @xmath1019 is true on a @xmath1020-structure @xmath1021 if and only if @xmath1022 is a vertex cover for the graph @xmath88 . using the notation from the previous section , we claim that for each @xmath147 and for all @xmath1023 , the set @xmath849 contains at most one entry @xmath1024 , and if @xmath846 for some @xmath787 , then @xmath1025 . to this end , consider arbitrary @xmath1026 and let @xmath1027 . for any @xmath1028 , such that @xmath1029 , the verifier has a winning strategy on @xmath1030 , where @xmath1031 with @xmath1032 , since the atomic formula @xmath1033 is always satisfied for all @xmath1014 .
therefore , @xmath1034 and @xmath579 removes the subgame @xmath6 from @xmath1035 .
consider now a subgame @xmath1036 , where @xmath1031 with @xmath1037 .
if there is @xmath1038 , such that @xmath1039 and @xmath1040 , then the falsifier has a winning strategy on @xmath1035 and consequently @xmath1041 .
if otherwise for all @xmath1038 either @xmath1042 or @xmath1043 , then we get @xmath1044 , and the corresponding subgame will be removed by @xmath579 .
therefore only the subgame on @xmath1045 with @xmath1046 remains undetermined .
we conclude @xmath1047 for all @xmath1048 . due to the symmetry of @xmath37 and @xmath1014 in the vertex cover formula
, we can argue analogously for the cases where the roles of @xmath37 and @xmath1014 have been interchanged .
therefore , @xmath1049 for all @xmath1050 , from which we conclude @xmath1051 .
each game is of size @xmath1052 , since by above considerations @xmath1053 and @xmath1054 : in both cases , we have @xmath980 subgames for the vertices in @xmath154 , one subgame for all vertices in @xmath1055 ( since all of them are equivalent ) , and one subgame for the case that @xmath37 and @xmath1014 , respectively , remain uninterpreted .
see figure [ fig : vc - unfolding ] for an example .
it is not hard to see that @xmath1056 , @xmath1057 , @xmath1058 , @xmath1059 and @xmath1060 can be implemented in a way such that they run in time polynomial in @xmath1061 and @xmath1062 .
hence , we immediately find that the generic algorithm introduced in this article reaches , up to factors polynomial in @xmath1 , the running time of @xmath1063 of the specialized algorithm , since @xmath1064 for all @xmath147 .
the formula @xmath1065 holds in @xmath1021 if and only if @xmath1022 is a dominating set for the graph @xmath88 .
let for each @xmath147 and @xmath1066 be @xmath1067 .
we claim that @xmath1068 . to this end , let again @xmath842 and @xmath1069 .
let @xmath1070 be such that @xmath1071 . if @xmath31 dominates @xmath1028 , then either @xmath1072 and @xmath1073 , or there is @xmath1074 that is adjacent to @xmath1075 , and @xmath1076 . in both cases
we get @xmath1077 , and therefore @xmath1078 .
, where @xmath34 has universe @xmath1079 , @xmath1012 , and @xmath1080 , such that @xmath1075 and @xmath1081 are not adjacent . then @xmath1075 might still be dominated by a `` future '' vertex ; the corresponding plays ( following the upper @xmath1015 branch in the figure ) end with a draw and therefore persist in the reduced game .
similarly , the branch @xmath36 corresponds to the case that `` future '' vertices are chosen as interpretations for @xmath37 .
such vertices can also be dominated by @xmath1081 , which is represented by the @xmath1082 branch in the figure .
[ fig : ds - unfolding ] ] if @xmath1028 is not dominated by @xmath31 , then @xmath1083 and @xmath1084 for all @xmath1045 with @xmath1032 and @xmath1085 .
these games are therefore removed by @xmath579 .
only the game @xmath1086 with @xmath1032 and @xmath1046 remains undetermined .
thus for all @xmath1087 that are not dominated by @xmath31 we have @xmath1088 . for @xmath1045 with @xmath1089
the game @xmath1090 remains undetermined .
for all @xmath1091 we have @xmath1092 due to the subformula @xmath1093 ; the corresponding subgame is therefore removed from @xmath1094 . for all @xmath1095 we again
have @xmath1096 .
all in all , either two games @xmath1097 only differ w.r.t .
the subset of _ undominated _ nodes in @xmath154 .
since there are @xmath23 nodes in @xmath154 that are not contained in @xmath31 , this bounds @xmath1098 .
for each of them , we have @xmath1099 corresponding to at most @xmath980 undominated nodes in @xmath154 , at most one undominated node in @xmath1100 and the subgame for @xmath1045 with @xmath1101 .
furthermore , @xmath1102 for @xmath1045 with @xmath1103 and @xmath1104 .
we conclude that @xmath1105 .
see figure [ fig : ds - unfolding ] for an example . in total , at a node @xmath147
, there are therefore at most @xmath1106 entries stored , and each entry has size @xmath1107 .
nodes @xmath147 of type _ leaf _ , _ forget _ are therefore processed in time @xmath1108 . for _
join _ nodes @xmath147 with children @xmath830 , every pair in @xmath1109 is considered .
therefore , at most @xmath1110 entries are considered , which yields a running time of @xmath1111 .
this does not yet match the best specialized algorithm for the minimum dominating set problem @xcite with a running time of @xmath1112 , but is still faster than combining all pairs with a running time of @xmath1113 .
we note that both the @xmath1007 bound from @xcite and the @xmath1114 bound from @xcite exploit a certain `` monotonicity '' property of domination like problems , which does not hold for all problems that are expressible in mso ( independent dominating set being an example ) .
the formula @xmath1115 defining the 3-colorability problem has no free symbols .
therefore @xmath1116 , where @xmath1117 is the empty tuple , and the table @xmath1118 contains at most one entry @xmath1119 .
we estimate the size of @xmath789 . for , let @xmath1120 , where @xmath1121 . here
, @xmath1122 expresses that the @xmath1123 are a partition of the universe , and @xmath1124 ensures that each @xmath779 is an independent set .
if @xmath1125 is not a partition of @xmath167 , then the falsifier wins @xmath1126 , and therefore @xmath1127 .
otherwise , @xmath1128 for all @xmath1028 and undetermined when @xmath37 remains uninterpreted . using the same arguments as for the similar vertex cover formula @xmath1129
, we have @xmath1049 for all @xmath1130 with @xmath1131 and @xmath1132 , @xmath1133 .
this implies @xmath1134 .
thus , @xmath1135 contains at most @xmath1136 subgames @xmath1137 , which bounds @xmath1138 .
thus , assuming @xmath1060 requires time @xmath1139 , we only can bound the total running time by @xmath1140 .
this can probably be improved to @xmath1141 using a similar approach as for the tables @xmath918 .
we started to implement the approach presented in this article in c++ .
the current version works for graphs over the vocabulary @xmath89 . at certain places ,
the implementation varies from the algorithms presented in this paper for increased efficiency .
for instance , @xmath579 is usually not called explicitly but computed directly where needed .
we list some running times and memory usage of the implementation when solving the three problems discussed in the previous section .
input graphs are randomly generated subgraphs of @xmath1142 grids and erds rnyi random graphs .
all graphs have about 200 vertices and the probability to include an edge ranges between @xmath1143 and @xmath1144 . for the grid - subgraphs we used path decompositions of width @xmath32 .
tree decompositions for the random graphs were computed by a triangulation heuristics ( cf .
the tests were done under linux 2.6.32 on a intel core 2 quad cpu q6600 ( 2.40ghz ) with 4 gb ram .
motivated by a practical application , we present an alternative proof of courcelle s theorem . our proof is based on model checking games and tries to avoid expensive constructions such as the power set construction for tree automata , which turned out to cause some problems in practice .
let us mention that our approach could be made simpler if we applied it to graphs of bounded clique - width .
the union operation for _ join _ nodes of a tree decomposition involves a `` fusion '' of elements and of interpretations of nullary symbols .
the clique - width parse trees do not use nullary symbols and the union is replaced by a disjoint union , which simplifies many of the operations . on the other hand , the lack of suitable algorithms to compute the mandatory clique - width parse trees favors treewidth based techniques for practical applications .
first experiments with our approach do indeed indicate practical feasibility .
an implementation based on our proof can solve the 3-colorability problem for some graphs where the automata theoretic approach based on the well - known mona tool failed .
the running times of our generic implementation can still not compete with specialized , hand - written algorithms that can easily solve problems such as , say 3-colorability , for graphs of treewidth 15 and beyond .
we are confident that further optimization can improve the feasibility of our generic approach in practical applications even more .
the authors thank an anonymous referee for valuable comments and suggestions that significantly helped to improve the quality of the paper .
the authors thank somnath sikdar for useful discussions on earlier drafts of this paper .
the third author thanks bruno courcelle for pointing out how to use incidence graphs to handle edge set quantifications in a simple way .
# 1proceedings of the # 1 conference on foundations of software technology and theoretical computer science ( fsttcs ) # 1proceedings of the # 1 international colloquium on automata , languages , and programming ( icalp ) # 1proceedings of the # 1 symposium on latin american theoretical informatics ( latin ) b. courcelle , i. a. durand , tractable constructions of finite automata from monadic second - order formulas , presented at _
logical approaches to barriers in computing and complexity _ , greifswald , germany ( 2010 ) .
k. a. abrahamson , m. r. fellows , finite automata , bounded treewidth , and well - quasiordering , in : graph structure theory , contemporary mathematics , vol .
147 , american mathematical society , 1993 , pp . 539564 | courcelle s theorem states that every problem definable in monadic second - order logic can be solved in linear time on structures of bounded treewidth , for example , by constructing a tree automaton that recognizes or rejects a tree decomposition of the structure .
existing , optimized software like the mona tool can be used to build the corresponding tree automata , which for bounded treewidth are of constant size .
unfortunately , the constants involved can become extremely large every quantifier alternation requires a power set construction for the automaton . here
, the required space can become a problem in practical applications . in this paper
, we present a novel , direct approach based on model checking games , which avoids the expensive power set construction .
experiments with an implementation are promising , and we can solve problems on graphs where the automata - theoretic approach fails in practice .
courcelle s celebrated theorem essentially states that every problem definable in monadic second - order logic ( mso ) can be solved in linear time on graphs of bounded treewidth @xcite .
however , the multiplicative constants in the running time , which depend on the treewidth and the mso - formula , can be extremely large @xcite .
[ thm : courcelle ] let @xmath0 be an mso problem and @xmath1 be a positive integer .
there is an algorithm @xmath2 and a function @xmath3 such that for every graph @xmath4 of order @xmath5 and treewidth at most @xmath1 , @xmath2 solves @xmath0 on input @xmath6 in time @xmath7 , where @xmath8 is the mso formula defining @xmath0 and @xmath9 is its length .
furthermore , unless @xmath10 , the function @xmath11 can not be upper bounded by an iterated exponential of bounded height in terms of @xmath8 and @xmath1 .
this result has been generalized by arnborg , lagergren , and seese to extended mso @xcite , and by courcelle and mosbah to monadic second - order evaluations using semiring homomorphisms @xcite . in both cases , an mso - formula with free set variables
is used to describe a property , and satisfying assignments to these set variables are evaluated in an appropriate way .
courcelle s theorem is usually proved as follows : in time only dependent on @xmath8 and the treewidth @xmath1 , a tree automaton @xmath12 is constructed that accepts a tree decomposition of width @xmath1 if and only if the corresponding graph satisfies the formula .
this construction can either be done explicitly , by actually constructing the tree automaton ( see , e.g. , @xcite ) , or implicitly via auxiliary formulas obtained by applying the feferman
vaught theorem @xcite extended to mso @xcite ( see , e.g. , @xcite ) . in a practical
setting , the biggest strength of courcelle s theorem is at the same time its largest weakness : mso logic has extremely large expressive power , and very short formulas can be used to encode np - hard problems .
this is used in @xcite to prove non - elementary worst - case lower bounds for the multiplicative constants in the linear running time .
even worse , these lower bounds already hold for the class of trees , i.e. , graphs of treewidth one .
on the other hand , these are worst - case lower bounds for very special classes of formulas and trees , and thus there is a good chance that in practice problems can be solved much faster .
in fact , existing software like the mona tool @xcite for weak second - order logic on two successors ( ws2s ) is surprisingly successful even though it is subject to the same theoretical lower bounds .
the automata - theoretic approach is therefore a promising starting point for practical applications of courcelle s theorem , particularly since advanced and optimized tools like mona can be used as a black box for the majority of the work , and techniques like minimizing tree automata are very well understood .
there are , however , some cases where the automata - theoretic approach is infeasible in practice , i.e. , when the automata ( or set of auxiliary formulas ) are too large to be practically computable .
this can even happen when the final minimal automata are small , but intermediate automata can not be constructed in reasonable time and space ( note that each quantifier alternation requires an automaton power set construction ) . in his thesis @xcite , soguet has studied the sizes of tree automata corresponding to various problems for small clique - width @xcite .
the automata were generated using mona , and in many cases , the corresponding automata were surprisingly small , thanks to the well - understood minimization of tree automata . on the other hand , even for graphs of clique - width three , mona was unable to construct the corresponding tree automata for the classical 3-colorability problem .
even worse , the same happened for simple problems such as deciding whether the graph is connected or if its maximum degree is two . these negative results are somewhat unsatisfying because the respective algorithm already fails in the first phase , when the automaton is constructed .
the first phase , however , only depends on the treewidth ( or clique - width in above cases ) and the formula ( i.e. , the problem ) , but is independent of the actual input graph . on the other hand ,
when running the tree automaton on most graphs arising from practical problems , only few states are actually visited .
recently , there have been a few approaches to this problem , see , e.g. , @xcite .
for example , the approach of @xcite avoids an explicit construction of the tree automaton . instead
, the state - transition function is computed _ on - the - fly_. experiments indicate practical feasibility .
courcelle @xcite introduces _ special tree - width _ , where the corresponding automata are easier to construct . in this paper , we present a novel , game - theoretic approach , where the input structure is taken into account from the beginning via model checking games ( cf . , @xcite ) .
therefore , only the amount of information is stored that is needed by the algorithm to solve the problem on this explicit input , and , in some sense , transitions between nodes of the tree decompositions are as well computed _ on - the - fly_. we particularly avoid the expensive power set construction .
we hope that the approach can be used in those cases , where the automata are too large to be constructed in practice , but the input graphs itself are simple enough .
in fact , first experiments are promising . using the generic approach ,
we can , for example , solve the 3-colorability problem on grids of size @xmath13 ( treewidth 6 ) in about 21 seconds and with 8 mb memory usage on standard pc hardware , and the minimum vertex cover problem on the same graph in less than a second and only 1 mb of memory usage .
we note that the automata construction using mona in @xcite already failed for @xmath14 grids ( clique - width 3 ) .
we briefly survey other approaches to courcelle s theorem .
we already mentioned that , given the mso formula @xmath8 , one can construct a finite - state bottom - up tree automaton that accepts a tree decomposition of the input graph @xmath15 if and only if @xmath16 .
this is sometimes called the automata theoretic approach .
a direct construction of the tree automata is described in , e.g. , @xcite or ( * ? ? ?
* chapter 6 ) . in @xcite a myhill - nerode type argument is used to show that the treewidth parse tree operators admit a right congruence with finitely many congruence classes .
the method of test sets can then be used to construct the tree automaton .
one can also use a reduction to the classical model checking problem for mso on labeled trees @xcite .
it is well - known @xcite that this problem can be solved by constructing suitable finite - state tree automata .
this approach is favorable if one likes to use existing software such as the mona tool @xcite .
a model theoretic approach is based on variants of the feferman
vaught theorem @xcite : if a graph @xmath15 can be decomposed into components @xmath17 and @xmath18 , then from the input formula @xmath8 one can construct a suitable _ reduction sequence _ consisting of boolean combinations ( _ and _ , _ or _ ,
_ not _ ) of finitely many formulas that hold in @xmath17 and @xmath18 if and only if @xmath8 holds in @xmath15 ( cf . , @xcite )
one can therefore use dynamic programming on the tree decomposition to compute the _
@xmath19-theory _ of @xmath15 , i.e. , set of formulas of quantifier rank at most @xmath19 that hold in @xmath15 ( cf . ,
@xcite ) .
similarly , one can also inductively compute the set of satisfying assignments to the input formula @xcite .
we are not aware of any implementations of courcelle s theorem based on the feferman
vaught approach .
the construction of all possible reduction sequences for mso formulas `` obviously is not practical '' ( * ? ? ?
* section 1.6 ) .
the algorithms presented in @xcite are therefore infeasible in practice .
however , from @xcite we get that computing the particular reduction sequence for the input formula @xmath8 suffices .
some lower bounds are known for the necessary conversions into disjunctions @xcite , but it would still be interesting to see how this approach behaves in practice .
a few authors studied practical aspects of the automata theoretic approach .
it is mentioned in @xcite that a myhill
nerode based program has been implemented as part of an m.sc .
thesis , which unfortunately does not seem to be publicly available .
the mona tool @xcite is a well - known and optimized implementation for the tree automata construction .
the space required to construct the automata with mona still turns out to cause severe problems in practical applications @xcite .
one idea ( * ? ? ?
* chapter 6 ) is to use precomputed automata for commonly used predicates such as @xmath20 expressing that the set @xmath21 is connected .
note however that the @xmath20 automaton requires @xmath22 states for graphs of clique - width @xmath23 ( * ? ? ?
* chapter 6 ) .
an automatic translation into monadic datalog is proposed in @xcite .
some experiments indeed suggest feasibility in practice ; their prototype implementation was , however , obtained by manual construction and not by an automatic transformation from the underlying mso formula . in
@xcite the power set construction is avoided by considering existential formulas only .
the automata thus remain non - deterministic , but of course standard methods to simulate runs of the automata apply .
since the state transition function is given only implicitly , the automaton is essentially computed _ on - the - fly _ while recognizing a clique - decomposition .
experiments have been conducted on graphs of comparably high clique - width and the approach is quite promising .
in fact , the lack of feasible algorithms to compute the necessary clique - width parse trees seems to be the major limitation . to ease the specification of such _ fly - automata _ , courcelle @xcite introduces _ special tree - width_. special tree - width lies between path - width and treewidth , but the automata are significantly smaller and easier to construct than those for treewidth . in this article
, we present a new approach that neither uses automata theoretic methods nor uses a feferman vaught style splitting theorem .
instead , we essentially evaluate the input formula on the graph using a simple recursive model checking algorithm . in what follows
, we shall outline this approach .
our starting point is the _ model checking game _ for mso ( definition [ def : model - checking - game ] ) , a pebble game between two players called the _ verifier _ and the _ falsifier _ also known as the hintikka game @xcite .
the verifier tries to prove that the formula holds on the input structure , while the falsifier tries to prove the opposite . in the game , the verifier moves on existential formulas ( @xmath24 , @xmath25 ) , while the falsifier moves on universal formulas ( @xmath26 , @xmath27 ) .
this game can in a natural way be identified with a simple algorithm that _ evaluates _ the formula on the input structure in a recursive manner .
if , for example , the formula is @xmath28 for a set variable @xmath29 , the algorithm checks whether @xmath30 holds for all sets @xmath31 . in this sense
, the computation tree of this simple algorithm can be interpreted as the _ unfolding _ ( cf . ,
@xcite ) of the model checking game . on a structure with @xmath32 elements ,
this straight - forward recursive model - checking algorithm takes time @xmath33 for a formula of quantifier rank @xmath19 . by dynamic programming on the tree decomposition
, we can improve this to time linear in @xmath32 on structures of bounded treewidth .
this works as follows : we traverse the tree decomposition of the input structure @xmath34 bottom - up . at each node of the tree decomposition
we preliminary try to evaluate the formula @xmath8 on @xmath34 using the model checking game on the `` current '' substructure @xmath35 of @xmath34 . to this end , we allow `` empty '' assignments @xmath36 to first order variables @xmath37 .
such empty assignments correspond to objects in @xmath34 that are not contained in @xmath35 and are to be assigned in later steps .
then , two things may happen : * we can _ already now _ determine whether @xmath38 or @xmath39 .
+ if , for instance , the formula @xmath40 encodes the 3-colorability problem and even @xmath35 is not three - colorable , it locally violates @xmath40 and we can derive @xmath41 . * we can not _ yet _ determine whether @xmath38 or @xmath39 .
+ for example , if the formula encodes dominating set problem , then a vertex @xmath42 in the `` current '' bag might be undominated in the current subgraph , but we do not know whether in the `` future '' another vertex might dominate @xmath42 .
the first case is formalized in lemma [ lem : emcg - eval - introduce ] and lemma [ lem : emcg - eval - union ] . in the second case
, we found a `` witness , '' i.e. , a subgame that we were unable to evaluate .
we then will re - visit those undetermined subgames during the course of the dynamic programming until we finally arrive in the root of the tree decomposition , where all subgames become determined .
the next crucial observation is that mso and fo formulas with bounded quantifier rank have limited capabilities to distinguish structures ( formally captured in the @xmath43-equivalence of structures , cf .
@xcite ) .
we exploit this fact and show that we can delete redundant equivalent subgames ( cf . , algorithm [ alg : reduce ] ) for a suitable definition of equivalence ( cf . , definition [ def : equivalent - games ] ) .
we can then show that , assuming a fixed formula and bounded treewidth , the number of reduced , non - equivalent games is bounded by a constant ( lemma [ lem : number ] ) , which allows us to obtain running times linear in the size of the tree decomposition . while this game - theoretic approach is subject to the same non - elementary lower bounds as the other approaches ,
the actual number of ways to play the model checking game highly depends on the input graph .
for example , if the graph does not contain , say , a triangle , then the players will never move to a set of nodes that induce a triangle , while a tree automaton must work for all graphs .
this observation is reflected in practical experiments , where the actual number of entries considered is typically much smaller than the corresponding worst - case bound . |
decays are ubiquitous in physics : examples are tunnelling processes in quantum mechanics , very slow decays of nuclei ( such as double-@xmath0 decays , lifetime of about @xmath1 y ) , as well as very fast decays of hadrons and the higgs particle ( lifetime of about @xmath2 sec ) .
although the involved decay times are astonishingly different , the basic phenomenon is the same : a irreversible transition ( infinite poincare time ) of an unstable initial state coupled to a continuum of final states .
both in the context of quantum mechanics ( qm ) and quantum field theory ( qft ) , an unstable state @xmath3 is not an eingenstate of the hamiltonian @xmath4 of the system , but is described by the so - called energy distribution @xmath5 : @xmath6 is the probability that the unstable state @xmath3 has an energy between @xmath7 and @xmath8 the survival probability amplitude @xmath9 that the state @xmath10 , prepared at the initial time @xmath11 , has not yet decayed at the instant @xmath12 , is given by@xmath13 the survival probability is given , as usual , by the amplitude squared : @xmath14 the exponential decay @xmath15 is obtained when @xmath5 takes the breit - wigner ( bw ) form : @xmath16 ^{-1}$ ] .
this is however an unphysical limiting case , because a physical distribution @xmath5 should fulfill the following two requirements : ( i ) existence of an energy threshold , @xmath17 for @xmath18 , which suffices to prove that long - time deviations exist , usually implying a power - law of the type @xmath19 @xcite .
( ii ) finiteness of the mean energy , @xmath20 , which implies that @xmath21 and therefore that deviations at short times occur @xcite .
moreover , a non - exponential behavior at short times implies the existence of the quantum zeno effect , which is the inhibition of the decay ( or , more in general , of a quantum transition ) , due to reiterated frequent measurements @xcite . from an experimental standpoint , the quantum zeno effect was observed by inhibition of a rabi oscillation between atomic energy levels @xcite . an improved set - up , in which the zeno effect was verified by using single ions and by means of interaction - free repeated measurements ,
was shown in ref .
@xcite . for what concerns the decay law of a ` genuine ' unstable quantum state ,
the first experimental demonstration of short - time deviations from the exponential law has been observed by studying the tunneling of cold atoms in an optical potential @xcite .
later on , the same group @xcite could verify also the zeno ( as well as the anti - zeno ) effect by applying a series of intermediate measurements on this system .
finally , the experimental proof of long - time deviations from the exponential law has been observed by investigating the decay via fluorescence of organic molecules @xcite .
an experiment performed in the gsi has also found deviations from the exponential decay law in hydrogen - like ions decaying via electron capture @xcite . in ref .
@xcite these deviations have been linked to a modification of the bw distribution due to interactions with the measuring apparatus .
other explanations , based on neutrino oscillations or energy level splitting of the initial state , have also been put forward @xcite . in this proceeding
we review some aspects of the non - exponential decay at short times both in qm and in qft . in sec .
2 we present for the qm case a simple but general hamiltonian to elucidate the main points . then
, we describe two recent developments in qm : the broadening of the emitted spectrum at short times @xcite and the case of a decay into two decay channels @xcite . in sec .
3 we turn to the qft case and we describe the recent results of refs .
@xcite , in which it has been shown that , contrary to previous claims @xcite , deviations from the exponential decay law also take place in a genuine qft framework . finally , in sec .
4 we present our conclusions .
we introduce a general approach to study decays in qm , which is based on the so - called lee hamiltonian @xmath22 @xcite : @xmath23 the quantity @xmath24 is a coupling constant with the dimension of energy@xmath25 .
the dimensionless function @xmath26 describes the interaction of the unstable state @xmath3 with the ` decay - product ' state @xmath27 ; the energy @xmath28 is the energy of @xmath27 in the interaction free case .
similar approaches to eq .
( [ lh ] ) are used in other areas of physics , e.g. in quantum optics with the jaynes - cummings hamiltonian .
the spectral function @xmath5 arises as the imaginary part of the propagator , @xmath29 $ ] , and is correctly normalized , @xmath30 .
the bw limit is obtained for @xmath31 and @xmath32 .
the decay width reads @xmath33 and the spectral function @xmath34 ^{-1}$ ] @xcite .
note , the choice @xmath32 means that the energy is not bounded from below and the choice @xmath31 means that the unstable state @xmath3 couples to each state of the continuum with the same intensity . for generic functions @xmath26 and @xmath28 the decay is not exponential ( see below ) , but is usually very well approximated by an exponential decay where the decay width is given by the fermi golden rule : @xmath35 , where @xmath36 . restricting to the bw limit
, an interesting feature is found when studying the energy of the emitted spectrum . for definiteness
, we consider the case of atomic spontaneous emission : an electron decays from an atomic excited state to the ground state by emitting a photon .
this is indeed the original framework in which the decay was studied @xcite ( see also e.g. refs . @xcite and refs . therein ) . denoting @xmath37 as the energy difference between the excited and the ground state , the energy distribution of the photon at the time @xmath12 is given by @xmath38 .
the energy uncertainty of the photon @xmath39 is given by the width at mid height : @xmath40 .
for @xmath41 one has @xmath42 , which is the natural broadening of the spectral line . for @xmath43 the quantity @xmath44 increases as @xmath45 .
this means that , if measured at short times after the emission , the photon can have an energy which deviates from the mean value @xmath37 by several decay widths @xmath46 , see ref .
@xcite for details .
[ ptb ] fig1.eps we now turn to the a simple example of non - exponential decay . to this end
, we choose @xmath47 which means that the state @xmath3 couples _ only _ to an energy window of final states .
a numerical example is shown in fig . 1 , in which the survival probability @xmath48 and its derivative , @xmath49 , are plotted ( solid lines ) and compared with the exponential counterparts ( dashed lines , obtained by using the fermi rule ) . deviations from the exponential decay are evident .
note , the function @xmath50 has a clear physical meaning : @xmath51 is the probability that the decay occurs in the time - interval @xmath52 .
for this reason , @xmath50 is also denoted as the density of decay probability .
[ ptb ] fig2.eps we now turn to the two - channel case , in which the unstable state @xmath10 couples to two sets of final states @xmath53 and @xmath54 the hamiltonian is a straightforward generalization of eq .
( [ lh ] ) : @xmath55 in the bw limit @xmath56 and @xmath57 where @xmath58 and @xmath59 are constants .
the partial decay widths are @xmath60 and @xmath61 and the survival probability reads @xmath62 with @xmath63 . in the presence of two decay channels , it is useful to introduce the quantity @xmath64 : @xmath65 is the probability that the state @xmath10 decays in the @xmath66-th channel between @xmath67 and @xmath68 in the bw limit @xmath69 and the ratio @xmath70 is a constant equal to @xmath71 . however ,
when deviations from the exponential decay are considered , the ratio @xmath70 is in general not a constant , but shows sizable departures from @xmath72 @xcite .
this is shown for a particular numerical example in fig .
2 , in which the choices @xmath73 and @xmath74 have been used .
qft is the theoretical framework in which particles are created and annihilated .
it is therefore the most fundamental environment to study the nature of decays . in refs .
@xcite it was claimed that in qft the deviations from the exponential decay law do not take place .
these conclusions , however , rely on a perturbative treatment of the decay law .
this issue was ( re)analyzed in refs .
@xcite with the following results : 1 . in ref .
@xcite the superrenormalizable qft interaction lagrangian @xmath75 , leading to the decay @xmath76 , was studied in detail .
the spectral function @xmath5 is calculated as the imaginary part of the propagator of @xmath77 to this end , it is necessary to perform a resummation of ( at least ) the one - loop self - energy diagram of the field @xmath78 .
( namely , a perturbative expression of @xmath5 in a series of @xmath24 is not a meaningful quantity @xcite . )
the survival probability takes the very same formal expression of eq .
( [ at])and short - time deviations from the exponential law do occur @xcite .
quite remarkably , the time interval in which such deviations take place is independent on the cutoff @xcite .
this is due to the fact that the energy distribution @xmath5 behaves as @xmath79 for large energies , thus being insensitive on the high energy scale of the theory . +
the two - channel case can be easily studied by using the lagrangian @xmath80 which implies that the two decay processes @xmath81 and @xmath82 take place .
the ratio @xmath70 shows also in this qft framework qualitatively similar deviations from the constant bw limit as those in fig .
2 , see ref .
@xcite for plots and details .
2 . in ref .
@xcite the renormalizable interaction lagrangian @xmath83 where @xmath84 is a fermionic field , has been analyzed .
( note , the higgs coupling to fermions has the same form . )
the energy distribution @xmath5 scales only as @xmath85 which in turn implies that the presence of a high energy cutoff is necessary for its correct normalization .
thus , the existence of a high energy cutoff is not only a mathematical step , but is a necessary physical requirement . in this framework ,
the duration of the deviations from the exponential decay at short times lasts only @xmath86 and is thus very small if the high - energy cutoff @xmath87 is large .
in this proceeding we have discussed the deviations from the exponential decay law in qm and qft . the decay law in qm has been studied by using the lee hamiltonian approach @xcite and in qft by using relativistic qft lagrangians @xcite . in qm
we have discussed a peculiar property of the energy spectrum of the photon in a spontaneous emission process : even in the exponential limit , a broadening of the emitted spectrum is realized if the photon is measured at early times @xcite . as a next step
we have studied short - time deviations from the exponential law in the simple case in which the unstable state couples to the continuum in a window of energy ( see fig . 1 and refs .
@xcite ) . a particularly interesting feature in both qm and
qft is that of a decay in which two ( or more ) decay channels are present .
the ratio of decay probability densities is a constant in the bw limit and equals the ratio of decay widths , but show sizable fluctuations on top of this constant limit when deviations from the bw limit are included ( see fig . 2 and ref .
@xcite ) . a systematic study of this ratio for a variety of cases , also including the analysis of realistic physical situations , is an important outlook for the future . c. a. nicolaides and t. mercouris , journal of physics b * 29 * , issue 6 , pp .
1151 - 1167 ( 1996 )
. k. urbanowski , central european journal of physics * 7 * , issue 4 , pp.696 - 703 .
e. torrontegui , j. g. muga , j. martorell , j , and d. w. l. sprung , phys . rev . a * 80 * , issue 1 , 012703 ( 2009 ) .
y. a. litvinov , f. bosch , n. winckler , d. boutin , h. g. essel , t. faestermann , h. geissel , s. hess _ et al .
_ , phys .
* b664 * ( 2008 ) 162 - 168 .
[ arxiv:0801.2079 [ nucl - ex ] ] .
p. kienle _ et al .
_ [ two - body - weak - decays collaboration ] , phys .
b * 726 * ( 2013 ) 638 [ arxiv:1309.7294 [ nucl - ex ] ] .. a. n. ivanov and p. kienle , phys . rev . lett .
* 103 * ( 2009 ) 062502 [ arxiv:0908.0877 [ nucl - th ] ] .
p. kienle , nucl .
a * 827 * ( 2009 ) 510c . c. giunti , phys .
b * 665 * ( 2008 ) 92 [ arxiv:0805.0431 [ hep - ph ] ] .
g. lambiase , g. papini and g. scarpetta , phys .
b * 718 * ( 2013 ) 998 [ arxiv:1205.0684 [ nucl - th ] ] .
a. merle , prog . part .
* 64 * ( 2010 ) 445 [ arxiv:1004.2347 [ hep - ph ] ] .
a. gal , nucl .
a * 842 * 102 ( 2010 ) . .
p. facchi and s. pascazio , phys .
a * 62 * , 023804 ( 2000 ) [ arxiv : quant - ph/9909043 ] .
p. r. berman and g. w. ford , phys .
a * 82 * , issue 2 , 023818 .
a. g kofman , g. kurizki , and b. sherman , journal of modern optics , vol .
41 , issue 2 , p.353 - 384 ( 1994 ) .
f. giacosa , g. pagliara , phys .
* c76 * ( 2007 ) 065204 . [ arxiv:0707.3594 [ hep - ph ] ] .
f. giacosa and t. wolkanowski , mod .
a * 27 * ( 2012 ) 1250229 arxiv:1209.2332 [ hep - ph ] ] .
t. wolkanowski and f. giacosa , acta phys .
* 6 * ( 2013 ) 893 [ arxiv:1305.2603 [ hep - ph ] ] . | we describe some salient features as well as some recent developments concerning short - time deviations from the exponential decay law in the context of quantum mechanics by using the lee hamiltonian approach and quantum field theory by using relativistic lagrangians .
in particular , the case in which two decay channels are present is analyzed : the ratio of decay probability densities , which is a constant equal to the ratio of decay widths in the exponential limit , shows in general sizable fluctuations which persist also at long times . |
the study of sn isotopes beyond @xmath0 is currently the subject of great interest , as it allows to explore for possible changes of the shell - model structure when moving towards the neutron drip line . at present , however , some experimental information is only available for @xmath2sn
. beyond this nucleus , which has an n / z ratio of 1.68 , lies a neutron - rich _ terra incognita _ which may become accessible to the next generation of radioactive ion beam facilities . in this perspective
, it is challenging to make predictions which may stimulate , and provide guidance to , future experiments .
in fact , over the past several years various shell - model studies have been performed @xcite predicting spectroscopic properties of hitherto unknown neutron - rich sn isotopes . in these studies
both empirical and realistic effective interactions have been used , the latter being based on the cd - bonn nucleon - nucleon ( @xmath3 ) potential model . in this connection , it should be mentioned that a main feature of the study of ref .
@xcite is that the short - range repulsion of the bare potential is renormalized by use of the @xmath1 approach @xcite .
a peculiar property of @xmath2sn is the position of its first @xmath4 state which , lying at 726 kev excitation energy , is the lowest first - excited @xmath4 level observed in a semi - magic even - even nucleus over the whole chart of nuclides .
this feature as well as the other observed excited states - @xmath5 , @xmath6 , and @xmath7 - have been successfully reproduced by the realistic calculations of @xcite . in that study a very good agreement was also obtained with the experimental @xmath8 value measured via coulomb excitation of a radioactive @xmath2sn beam @xcite .
very recently , two sophisticated experimental studies @xcite have provided new information of great relevance to the study of sn isotopes beyond @xmath0 .
more precisely , in @xcite a high - precision penning trap mass measurement has revealed a 0.5 mev discrepancy with respect to previous @xmath9 measurements .
this finding provides clear evidence of the robustness of the @xmath0 shell closure . in @xcite ,
the magic nature of @xmath10sn has been explored through the study of the single - particle states in @xmath11sn populated by a ( d , p ) reaction in inverse kinematics .
this study has evidenced the purity of the single - neutron excitations in @xmath11sn and has identified a strong candidate for the @xmath12 single - particle state at 1.363 mev excitation energy .
this is about 300 kev lower than the previous accepted value @xcite .
based on the above , we have found it interesting to revisit our previous realistic shell - model calculations for @xmath2sn and extend them to the heavier isotopes up to @xmath13 . in this paper
we first give a brief description of the theoretical framework in which our calculations are performed and then present and discuss our results which are , for the most part , predictions for possible future experiments .
a short summary is given in the last section .
we have performed shell - model calculations for even tin isotopes beyond the @xmath0 shell closure , namely , from @xmath2sn to @xmath14sn .
we have taken doubly magic @xmath10sn as closed core and let the valence neutrons occupy the six levels @xmath15 , @xmath16 , @xmath17 , @xmath18 , @xmath12 , and @xmath19 of the @xmath20 shell .
our adopted single - particle ( sp ) energies relative to the @xmath16 level are ( in mev ) : @xmath21 , @xmath22 , @xmath23 , @xmath24 , and @xmath25 .
these values are taken from the experimental spectrum of @xmath11sn @xcite , with the exception of the @xmath19 level which was determined from the position of the 2.434 mev level in @xmath2sb assumed to be a @xmath26 state of @xmath27 nature .
as for the @xmath12 level , we use the energy reported in the recent experiment of ref .
it is worth emphasizing that in this experiment the spectroscopic factors of the four levels @xmath16 , @xmath18 , @xmath12 , @xmath17 were extracted , which provide a clear confirmation of their sp nature .
the absolute energy of the @xmath16 level relative to @xmath10sn was placed at @xmath28 mev , as resulting from the new mass measurement of @xmath11sn @xcite .
the shell - model effective interaction has been derived within the framework of perturbation theory @xcite starting , as mentioned in the introduction , from the cd - bonn @xmath3 potential @xcite renormalized by way of the so - called @xmath1 approach @xcite .
more precisely , we start by deriving @xmath1 with a cutoff momentum of @xmath29 @xmath30 .
then , using this potential plus the coulomb force for protons , we calculate the two - body matrix elements of the effective interaction by means of the @xmath31-box folded - diagram expansion , with the @xmath31-box including all diagrams up to second order in the interaction .
these diagrams are computed within the harmonic - oscillator basis using intermediate states composed of all possible hole states and particle states restricted to the five proton and neutron shells above the fermi surface .
the oscillator parameter is 7.88 mev for the @xmath32 region , as obtained from the expression @xmath33 .
the calculations have been performed by using the oslo shell - model code @xcite .
experimental and calculated spectrum of @xmath2sn . ]
we start this section by discussing our results for @xmath2sn , the only tin isotope beyond @xmath0 for which some experimental information is available . in figure [ fig1 ]
all the observed levels @xcite are reported and compared with the calculated ones up to about 2.5 mev excitation energy .
the experimental spectrum consists of a group of low - lying levels separated by an energy gap of about 1.5 mev from the highest observed level which has @xmath34 .
the former correspond to the four calculated members of the @xmath35 multiplet while the latter to the maximum - aligned state of the @xmath36 configuration and all energies are very well reproduced by the theory . in the energy interval between 1.1 and 2.5
mev our calculation predicts the existence of other states , namely the members of the @xmath37 and @xmath38 multiplets , and of the @xmath39 state of the @xmath40 configuration .
it is worth noting that in our previous calculation @xcite the @xmath41 and @xmath42 states were predicted to lie above the @xmath7 state , their downshift in energy being directly related to the new position of the @xmath43 sp level .
we have also calculated some electromagnetic properties and our predicted values for the first four excited states are reported in table [ tab1 ] , where the two experimental transition rates presently available @xcite are also shown .
our calculations have been performed using an effective neutron charge of 0.7@xmath44 @xcite , which leads to @xmath45 values for both the @xmath46 and @xmath47 transitions quite close to the experimental ones .
[ h ] ccc & calc & expt + @xmath48 & 1.64 & @xmath49 + @xmath50 & 1.66 & + @xmath51 & 0.82 & @xmath52 + @xmath53 & 0.35 & + @xmath54 & 2.93 & + @xmath55 & 0.23 & + @xmath56 & 0.02 & + @xmath57 & -1.6 & + @xmath58 & -2.8 & + @xmath59 & -0.57 & + @xmath60 & -0.25 & + we hope that the above predictions for @xmath2sn may provide guidance to future experiments aiming at the identification of the missing states above the yrast @xmath6 state as well as at the measurement of other properties , which may certainly give better insight into the structure of this exotic nucleus . in this context
, it should be mentioned that several interesting questions have been posed by the new , although still scarce , experimental data which have become available for some nuclei in the @xmath10sn region . in particular , special attention has been focused on the properties of the yrast @xmath4 states of tin and tellurium isotopes , whose excitation energy shows an asymmetric behavior with respect to @xmath0 . in fact , in @xmath2sn and @xmath61te
the @xmath62 energy undergoes a significant drop as compared to the @xmath63 isotopes , which was traced to a reduction of the neutron pairing above the @xmath0 shell @xcite . with our realistic effective interaction
we are able , as shown above , to reproduce with good accuracy the energy of the yrast @xmath4 state and its @xmath45 transition rate in @xmath2sn .
this was also the case for @xmath61te ( see @xcite ) whose @xmath4 state is described as the corresponding state in @xmath2sn with the two additional protons coupled to zero angular momentum . as a matter of fact ,
the @xmath4 state , as well as the other three lowest - lying states in @xmath2sn , are characterized by a very weak configuration mixing .
therefore the small @xmath64 spacing reflects the small difference between the @xmath65 and @xmath66 matrix elements , which turns out to be about 0.4 mev .
this result is due to the negligible contribution @xcite of the one particle - one hole excitations in the effective interaction of two neutrons above the @xmath0 shell , which are instead responsible for the increase of pairing below this shell .
the weakness of the cross - shell interaction for @xmath10sn as compared with other lighter closed cores was also pointed out in ref .
@xcite , where it was noted that the excitation energy of the @xmath67 state in @xmath10sn is only 0.5 mev smaller than the shell gap for neutrons . we have investigated the evolution of the yrast @xmath4 state and of the other @xmath68 members of the lowest - lying multiplet when adding neutron pairs . in figure [ fig2 ]
we report the excitation energies of the @xmath69 , @xmath70 , and @xmath71 states as a function of a up to @xmath14sn .
we see that the curves are almost flat , with a slight increase at @xmath72 for the @xmath69 and @xmath70 which becomes more significant for the @xmath73 state .
calculated excitation energies of the yrast @xmath4 , @xmath5 , and @xmath6 states in tin isotopes with @xmath74 , 136 , 138 , and 140 . ]
this result is similar to that found in @xcite , the main difference being that our @xmath70 and @xmath71 curves lie about 100 - 300 mev below those of @xcite , while is in contrast to the conclusion of the recent work of @xcite claiming for a shell closure at @xmath72 .
the results of @xcite predict an upshift of the @xmath4 state in @xmath14sn with respect to the lighter isotopes when using an empirical interaction ( smpn ) as well as a realistic interaction with three - body monopole corrections ( cwg3 m ) .
this upshift was related to the filling of the @xmath75 orbit at @xmath72 as a consequence of a significant increase in the difference between the effective single - particle energies of the @xmath75 and @xmath76 orbits which , starting from 0.854 kev , becomes larger than 2 mev at @xmath13 . as can be seen in figure [ fig3 ]
, our calculations do not predict such an increase .
the energy difference between the two orbits remains in fact almost constant and even decreases slightly by about 100 kev when moving from @xmath0 to @xmath72 . as regards the @xmath6 state , we find that its wave function in @xmath14sn is dominated by the @xmath77 configuration , which explains its increasing in energy .
[ h ] effective single - particle energies of the @xmath75 and @xmath76 levels from @xmath0 to @xmath72 . ] in concluding this section , we discuss binding ( relative to @xmath10sn ) and one - neutron separation energies of the even sn isotopes .
the calculated values are reported in table [ tab2 ] , where the experimental available values for @xmath2sn are also shown . as mentioned in the introduction
, a new measurement of the binding energy of @xmath2sn has revealed a 0.5 mev deviation from the previous observed value . with the new mass value
the maximum of the neutron - shell gap for n=82 is at @xmath78 , instead of @xmath79 as established by the old measurement .
we find an excellent agreement between theory and experiment , which shows the reliability of the @xmath80 matrix elements of our effective interaction .
this confirms the weakening of pairing for two neutrons beyond the @xmath0 shell .
as regards the heavier sn isotopes , we see that the binding energy keeps increasing with increasing mass number , while the one - neutron separation energy remains practically constant .
our calculations , in line with those of @xcite and the estimations of mean field calculations , seem to indicate that in this region we are still quite far from the neutron drip line .
lcccc & @xmath2sn & @xmath61sn & @xmath81sn & @xmath14sn + be calc & 5.92 & 11.83 & 17.68 & 23.41 + be expt & @xmath82 & & & + s@xmath83 calc & 3.55 & 3.55 & 3.53 & 3.50 + s@xmath83 expt & @xmath84 & & & +
we have presented here the results of a shell - model study of exotic even mass sn isotopes from a= 134 to a= 140 . in our calculations
we have employed a realistic low - momentum effective interaction derived from the cd - bonn @xmath3 potential without using any adjustable parameter . the heaviest sn isotope for which there is some , albeit scarce , experimental information is @xmath2sn , so most of our results are predictive in nature .
we have shown that the available experimental data are very well reproduced by our calculations .
this gives confidence in their predictive power , which we hope can be verified by experiment in the not too distant future .
99 coraggio l , covello a , gargano a and itaco n 2002 _ phys .
_ c * 65 * 051306(r ) covello a , coraggio l , gargano a and itaco n 2007 _ eur .
j st _ * 150 * 93 kartamyshev m p , engeland t , hjorth - jensen m and osnes e 2007 _ phys . rev .
_ c * 76 * 024313 sarkar s and sarkar m s 2010 _ phys .
_ c * 81 * 064328 and references therein bogner s , kuo t t s , coraggio l , covello a and itaco n 2002 _ phys .
_ c * 65 * 051301(r )
beene j r _ et al .
_ 2004 _ nucl .
phys . _ a * 746 * 471c dworschak m _
_ 2008 _ phys .
lett . _ * 100 * 072501 jones k l _ et al .
_ 2010 _ nature _ * 465 * 454 hoff p
_ et al . _
lett . _ * 77 * 1020 data extracted using the nndc on - line data service from the ensdf database , file revised as of october 4 , 2010 coraggio l , covello a , gargano a , itaco n and kuo t t s 2009 _ prog .
phys . _ * 62 * 135 and references therein machleidt r 2001 _ phys . rev .
_ c * 63 * 024001 engeland t the oslo shell - model code 1991 - 2006 , unpublished terasaki j , engel j , nazarewicz w and stoitsov m , 2002 _ phys .
_ c * 66 * 054313 shimizu n , otsuka t , mizusaki t and honma m 2004 _ phys .
_ c * 70 * 054313 covello a , coraggio l , gargano a and itaco n 2007 _ prog .
phys . _ * 59 * 401 coraggio l , covello a , gargano a and itaco n 2009 _ phys .
_ c * 80 * 021305(r ) | we report on a shell - model study of sn isotopes beyond @xmath0 employing a realistic effective interaction derived from the cd - bonn nucleon - nucleon potential renormalized through use of the @xmath1 approach . at present ,
the most exotic sn isotope for which some experimental information exists is @xmath2sn with an n / z ratio of 1.68 .
it is the aim of our study to compare the results of our calculations with the available experimental data and to make predictions for the neighboring heavier isotopes which may be within reach of the next generation of radioactive ion beam facilities .
the very good agreement between theory and experiment obtained for @xmath2sn gives confidence in the predictive power of our realistic shell - model calculations . |
the non - commutative geometry of a quasi - periodic tiling studies an appropriate @xmath0-algebra of a dynamical system @xmath3 , for a compact metric space @xmath4 , called _ the hull _ , endowed with a continuous lie group @xmath5 action .
this @xmath0-algebra is of relevance to study the space of leaves which is pathological in any topological sense .
the hull owns also a geometrical structure of _ lamination _ or _ foliated space _ ,
the transverse structure being just metric @xcite .
the @xmath0-algebras and the non - commutative tools provide then topological and geometrical invariants for the tiling or the lamination . moreover
, some @xmath1-theoretical invariants of euclidean tilings have a physical interpretation .
in particular , when the tiling represents a quasi - crystal , the image of the @xmath1-theory under the canonical trace labels the gaps in the spectrum of the schrdinger operator associated with the quasi - crystal @xcite . for an euclidean tiling , the group @xmath5 is @xmath6 and @xmath6-invariant ergodic probability measures on the hull are in one - to - one correspondence with ergodic transversal invariant measures and also with extremal traces on the @xmath0-algebra @xcite .
these algebras are well studied and this leads , for instance , to give distinct proofs of the _ gap labelling conjecture _ @xcite , i.e. for minimal @xmath6-action , the image of the @xmath1-theory under a trace is the countable subgroup of @xmath7 generated by the images under the corresponding transversal invariant measure of the compact - open subsets of the ( cantor ) canonical transversal . for a hyperbolic quasi - periodic tiling ,
the situation is quite distinct .
the group of affine transformations acts on the hull and since this group is not unimodular , there is no transversally invariant measure @xcite .
a new phenomena shows up for the @xmath0-algebra of the tiling : it has no trace .
nevertheless , the affine group is amenable , so the hull admits at least one invariant probability measure .
these measures are actually in one - to - one correspondence with harmonic currents @xcite , and they provide @xmath2-cyclic cocycles on the smooth algebra of the tiling . the present paper is devoted to give a complete description of the @xmath0-algebra and the @xmath1-theory of a specific family of hyperbolic tilings derivated from the example given by penrose in @xcite .
the dynamic of the hulls under investigation , have a structure of double suspension ( this make sens in term of groupoids as we shall see in section [ subsec - groupoid - suspension ] ) which enables to make explicit computations .
this suggests that the pairing with the @xmath2-cyclic cocycle is closely related to the one - dimension gap - labelling for a subshift associated with the tiling .
but the right setting to state an analogue of the gap - labelling seems to be frechet algebras and a natural question is whether this bring in new computable invariants .
background on tiling spaces is given in the next section and we construct examples of hyperbolic quasi - periodic tilings in the third section .
a description of the considered hulls is given in section 4 . in section 5 ,
we recall the background on the groupoids and their @xmath0-algebras .
sections 6 and 7 are devoted to the complete description of the @xmath0-algebras of the examples and their @xmath1-theory groups in terms of generetors are given in section @xmath8 . for readers interested in topological invariants of the hull ,
we compute its @xmath1-theory and its cch cohomology and we relate these computations to the former one . in the last section we construct @xmath2-cyclic cocycles associated to these tilings and we discuss an odd version of the gap - labelling .
let @xmath9 be the real hyperbolic @xmath10-space , identified with the upper half complex plane : @xmath11 with the metric @xmath12 .
we denote by @xmath5 the group of _ affine transformations _ of this space : i.e. the isometries of @xmath9 of the kind @xmath13 with @xmath14 , @xmath15 reals and @xmath16 .
a tiling @xmath17 of @xmath9 , is a collection of convex compact polygons @xmath18 with geodesic borders , called _
, such that their union is the whole space @xmath9 , their interiors are pairwise disjoint and they meet full edge to full edge . for instance , when @xmath19 is a fundamental domain of a co - compact lattice @xmath20 of isometries of @xmath9 , then @xmath21 is a tiling of @xmath9 .
however the set of tilings is much richer than the one given by lattices as we should see later on .
similarly to the euclidian case , a tiling is said of @xmath5-_finite type _ or _ finite affine type _
, if there exists a finite number of polygons @xmath22 called _ prototiles _ such that each @xmath18 is the image of one of these polygons by an element of @xmath5 . besides its famous euclidean tiling , penrose in @xcite constructs a finite affine type tiling made with a single prototile which is not stable for any fuchsian group .
the construction goes as follows .
let @xmath23 be the convex polygon with vertices @xmath24 with affix @xmath25 for @xmath26 and @xmath27 and @xmath28 @xmath23 is a polygon with @xmath29 geodesic edges .
consider the two maps : @xmath30 the hyperbolic penrose s tiling is defined by @xmath31 ( see figure [ pentil ] ) .
this is an example of finite affine type tiling of @xmath9 .
this tiling is stable under no co - compact group of hyperbolic isometries .
the proof is homological : we associate with the edge @xmath32 a positive charge and two negative charges with edges @xmath33 , @xmath34 .
if @xmath35 was stable for a fuchsian group , then @xmath23 would tile a compact surface .
since the edge @xmath32 can meet only the edges @xmath33 or @xmath34 , the surface has a neutral charge .
this is in contradiction with the fact @xmath23 is negatively charged .
g. margulis and s. mozes @xcite have generalized this construction to build a family of prototiles which can not be used to tile a compact surface .
notice the group of isometries which preserves @xmath36 is not trivial and is generated by the transformation @xmath37 . in order to break this symmetry ,
it is possible , by a standard way , to decorate prototiles to get a new finite affine type tiling which is stable under no non - trivial isometry ( we say in this case that the tiling is _ aperiodic _ ) .
in this section , we recall some basic definitions and properties on dynamical systems associated with tilings . we refer to @xcite , @xcite and @xcite for the proofs .
we give then a description of the dynamical system associated to the hyperbolic penrose s tiling .
first , note that the group @xmath5 acts transitively , freely ( without a fixed point ) and preserving the orientation of the surface @xmath9 , thus @xmath5 is a lie group homeomorphic to @xmath9 .
the metric on @xmath9 gives a left multiplicative invariant metric on @xmath5 .
we fix the point @xmath38 in @xmath9 with affix @xmath39 that we call _
origin_. for a tiling @xmath40 of @xmath5 finite type and an isometry @xmath41 in @xmath5 , the image of @xmath40 by @xmath41 is again a tiling of @xmath9 of @xmath5 finite type .
we denote by @xmath42 the set of tilings which are image of @xmath40 by isometries in @xmath5 .
the group @xmath5 acts on this set by the left action : @xmath43 we equip @xmath42 with a metrizable topology , so that the action becomes continuous .
a base of neighborhoods is defined as follows : two tilings are close one to the other if they agree , on a big ball of @xmath9 centered at the origin , up to an isometry in @xmath5 close to the identity .
this topology can be generated by the metric @xmath44 on @xmath42 defined by ( see @xcite ) : for @xmath40 and @xmath45 be two tilings of @xmath42 , let @xmath46 \ \vert\ \exists\ g \in b_{\epsilon}(id ) \subset g \ { \rm s.t . } \
g.t\cap b_{1/\epsilon } = t ' \cap b_{1/\epsilon}\}\ ] ] where @xmath47 is the set of points @xmath48 such that @xmath49 .
we define : @xmath50 @xmath51 the _ continuous hull _ of the tiling @xmath40 , is the metric completion of @xmath42 for the metric @xmath44 .
we denote it by @xmath52 .
actually this space is a set of tilings of @xmath9 of @xmath5-finite type .
patch _ of a tiling @xmath40 is a finite set of tiles of @xmath40 .
it is straightforward to show that patches of tilings in are copies of patches of @xmath40 .
the set is then a compact metric set and the action of @xmath5 on @xmath53 can be extended to a continuous left action on this space .
the dynamical system @xmath54 has a dense orbit : the orbit of @xmath40 .
+ some combinatorial properties can be interpreted in a dynamical way like , for instance , the following .
a tiling @xmath40 satisfies the repetitivity condition if for each patch @xmath23 , there exists a real @xmath55 such that every ball of @xmath9 with radius @xmath55 intersected with the tiling @xmath40 contains a translated by an element @xmath5 of the patch @xmath23 .
this definition can be interpreted from a dynamical point of view ( for a proof see for instance @xcite ) .
the dynamical system @xmath56 is minimal ( any orbit is dense ) if and only if the tiling @xmath40 satisfies the repetitivity condition .
we call a tiling _ aperiodic _ if the action of @xmath5 on is free : for all @xmath57 of @xmath5 and all tilings @xmath45 of we have @xmath58 .
as we have seen in the former section the hyperbolic penrose s tiling is not aperiodic , however , using this example , we shall construct in section [ exemple ] uncountably many examples of repetitive and aperiodic affine finite type tilings .
when the tiling @xmath40 is aperiodic and repetitive , the hull has also a geometric structure of a specific lamination called a @xmath5-_solenoid _ ( see @xcite ) .
locally at any point @xmath59 , there exists a _ vertical germ _ which is a cantor set included in , transverse to the local @xmath5-action and which is defined independently of the neighborhood of the point @xmath59 .
this implies that is locally homeomorphic to the cartesian product of a cantor set with an open subset ( called a _ slice _ ) of the lie group @xmath5 .
the connected component of the slices that intersect is called a _ leaf _ and has a manifold structure .
globally , is a disjoint union of uncountably many leaves , and it turns out that each leaf is a @xmath5-orbit . since the action is free , each leaf is homeomorphic to @xmath9 . in the aperiodic case ,
the @xmath5-action is _ expansive _ : there exists a positive real @xmath60 such that for every points @xmath61 and @xmath62 in the same vertical in , if @xmath63 for every @xmath64 , then @xmath65 .
+ furthermore this action has locally constant return times : if an orbit ( or a leaf ) intersects two verticals @xmath66 and @xmath67 at points @xmath68 and @xmath69 where @xmath64 , then for any point @xmath70 of @xmath66 close enough to @xmath68 , @xmath71 belongs to @xmath67 . first recall the notion of suspension action for @xmath4 a compact metric space and @xmath72 a homeomorphim .
the group @xmath73 acts diagonally on the product space @xmath74 by the following homeomorphism denoted @xmath75 @xmath76 the quotient space of @xmath77 , where two points are identified if they belong to the same orbit , is a compact set for the product topology and is called _ the suspension _ of @xmath78 . the group @xmath7 acts also diagonally on @xmath74 : trivially on x and by translation on @xmath7 .
since this action commutes with @xmath75 , this induces a continuous @xmath7 action on the suspension space @xmath77 that we call : the _ suspension action _ of the system @xmath78 and we denote it by @xmath79 .
we recall here , the construction of the dyadic completion of the integers . on the set of integers @xmath73
, we consider the dyadic norm defined by @xmath80 let @xmath81 be the completion of the set @xmath73 for the metric given by @xmath82 .
the set @xmath81 has a commutative group structure where @xmath73 is a dense subgroup , and @xmath83 is a cantor set .
the continuous action given by the map @xmath84 on @xmath81 is called _ adding - machine _ or _
odometer _ and is known to be minimal and equicontinuous .
we denote by @xmath85 the suspension action of this homeomorphism .
+ recall that a conjugacy map between two dynamical systems is a homeomophism which commutes with the actions .
let @xmath86 be the group of transformations @xmath87 isomorphic to @xmath7 .
[ premsuspension ]
let @xmath88 be the closure ( for the tiling topology ) of the orbit @xmath89 .
then the dynamical system @xmath90 is conjugate to the suspension action of the odometer @xmath91 .
let @xmath92 be the map defined by @xmath93 $ ] where @xmath94 $ ] is the @xmath95-class of @xmath96 .
since the tiling is invariant under no translations , the application @xmath97 is well defined .
it is straightforward to check that @xmath97 is continuous for the tiling topology and for the topology on @xmath98 arising from the dyadic topology on @xmath81 .
so the map @xmath97 extends by continuity to @xmath88 .
let us check that @xmath97 is a homeomorphism by constructing its inverse .
let @xmath99 defined by @xmath100 .
this application is continuous for the dyadic topology , so it extends by continuity to @xmath101 .
notice @xmath102 is constant along the orbits of the @xmath95 action on @xmath103 which is dense in @xmath101 .
thus @xmath104 is constant along the @xmath95-orbits in @xmath105 and @xmath104 factorizes through a map @xmath106 from the suspension @xmath107 to @xmath88 .
it is plain to check that @xmath108 on the dense set @xmath109 and that @xmath110 on the dense set @xmath111 where @xmath112 denotes the canonical projection .
hence @xmath113 is an homeomorphism from @xmath88 onto @xmath114 .
it is obvious that @xmath113 commutes with the @xmath7-actions .
we construct in this section a family of tilings of @xmath9 of finite affine type , indexed by sequences on a finite alphabet . for uncountably many of them , the tilings will be aperiodic and repetitive , the action on the associated hull will be free and minimal .
a description of these actions in terms of double - suspension is given . to construct such tilings we will use the hyperbolic penrose s tiling described in section [ hpt ] , so we will keep the notations of this section .
recall that its stabilizer group under the action of @xmath5 , is the group @xmath115 generated by the affine transformations @xmath37 .
the main idea is to `` decorate '' this tiling in order to break its symmetry , the decoration will be coded by a sequence on a finite alphabet . by a decoration , we mean that we will substitute to each tile @xmath116 the same polygon @xmath116 equipped with a color .
we take the convention that two colored polygons are the same if and only if the polygons are the same up to an affine map and they share the same color . by substituting each tile by a colored tile ,
we obtain a new tiling of finite affine type with a bigger number of prototiles .
notice that the coloration is not canonical .
it also possible to do the same by substituting to a tile @xmath116 , an unique finite family of convex tiles @xmath117 , like triangles , such that the union of the @xmath118 is @xmath116 and the tiles @xmath118 overlaps only on their borders .
we choose the coloration only for presentation reasons .
let @xmath119 be an integer bigger than @xmath120 .
we associate to each element of @xmath121 an unique color .
let @xmath23 be the polygon defined in section [ hpt ] to construct the penrose s tiling .
for an element @xmath39 of @xmath121 , we denote by @xmath122 the prototile @xmath23 colored in the color @xmath39 . to a sequence @xmath123\{1 , , r}@xmath124 , we associate the @xmath5-finite - type tiling @xmath125 built with the prototiles @xmath122 for @xmath39 in @xmath121 and defined by : @xmath126 notice that the stabilizer of this tiling is a subgroup of @xmath127 .
the set of sequences on @xmath121 is the product space @xmath128 which is a cantor set for the product topology .
there exists a natural homeomorphism on it called the @xmath129 . to a sequence @xmath130 the shift
@xmath131 associates the sequence @xmath132 .
let @xmath133 denote the closure of the orbit of @xmath70 by the action of the shift @xmath131 : @xmath134 .
the set @xmath133 is a compact metric space stable under the action of @xmath131 .
[ rem - cont - colour ] the map @xmath135 is continuous .
since @xmath136 denotes the tiling image of @xmath137 by @xmath37 , we get the relation @xmath138 thanks to this , we obtain the following property : * the sequence @xmath70 is aperiodic for the shift - action , if and only if @xmath137 is stable under no non - trivial affine map . *
the dynamical system @xmath139 is minimal , if and only if @xmath140 is minimal .
the first point comes from the relation [ relation ] and from the fact that the stabilizer of @xmath137 is a subgroup of @xmath141 .
the last point comes from the characterization of minimal sequences : @xmath139 is minimal if and only if each words in @xmath70 appears infinitely many times with uniformly bounded gap @xcite .
this condition is equivalent to the repetitivity of @xmath137 .
recall that we have defined the group @xmath142 and that @xmath88 stands for the closure ( for the tiling topology ) of the @xmath86-orbit @xmath109 in @xmath143 of the uncolored tiling @xmath36 .
notice that the continuous action of @xmath37 on @xmath143 preserves the orbit @xmath109 so we get an homeomorphism of @xmath144 that we denote also by @xmath37 .
we consider on the space @xmath145 equipped with the product topology , the homeomorphism @xmath146 defined by @xmath147 .
the quotient space @xmath148 , where the points in the same @xmath149 orbit are identified , is a compact space .
the affine group @xmath5 also acts on the left on @xmath145 ; where the action of an element @xmath150 is given by the homeomorphism @xmath151 it is straightforward to check that the application @xmath149 commutes with this action , so this defines a @xmath5-continuous action on the quotient space @xmath152 .
[ caracterisation ] let @xmath70 be an element of @xmath153
. then the map @xmath154\end{aligned}\ ] ] where @xmath155 $ ] denotes the @xmath149-class of @xmath59 , extends to a conjugacy map between @xmath156 and @xmath157 let @xmath158 be the transformation @xmath159 defined by @xmath160 where @xmath161 denotes the map @xmath162 . according to remark [ rem - cont - colour ] , the application @xmath158 is continuous for the tiling topology on @xmath109 , so it extends by continuity to a continuous map from @xmath163 to @xmath164 .
thanks to relation ( [ relation ] ) , we get @xmath165 on the dense subset @xmath166
. therefore the map @xmath158 factorizes throught a continuous map @xmath167 . since the stabilizer of the tiling @xmath168 is a subgroup of the one generated by the transformation @xmath37 , and by relation ( [ relation ] ) , the map @xmath169\ ] ] is well defined .
it is straightforward to check that @xmath170 is continuous , so it extends to a continuous map from @xmath171 to @xmath172 that we denote again @xmath170 .
furthermore we have @xmath173 on @xmath174 and @xmath175 on the dense set @xmath176 where @xmath177 denotes the canonical projection @xmath178 .
hence the map @xmath179 is an homeomorphism .
the homeomorphism @xmath170 obviously commutes with the action .
notice that , @xmath180 is locally the cartesian product of a cantor set by an interval of @xmath7 . for @xmath181 ,
@xmath182 is locally homeomorphic the product of a cantor set by an open subset ( a slice ) of @xmath183 since the cartesian product of two cantor sets is again a cantor set .
the @xmath5-action maps slices onto slices . for a metric space @xmath4 and a continuous action of a group @xmath20 on it , a @xmath20-_invariant measure _ is a measure @xmath184 defined on the borel @xmath131-algebra of @xmath4 which is invariant under the action of @xmath20 i.e. : for any measurable set @xmath185 and any @xmath186 , @xmath187 .
for instance , any group @xmath20 acts on itself by right multiplication , there exists ( up to a scalar ) only one measure invariant for this action : it is called the _ haar _ measure .
any action of an amenable group @xmath20 ( like @xmath73 , @xmath7 and all their extensions ) on a compact metric space @xmath4 admits a finite invariant measure and in particular , any homeomorphism @xmath188 of @xmath4 preserves a probability measure .
an _ ergodic _
invariant measure @xmath184 is such that every measurable functions constant along the orbits are @xmath184 almost surely constant .
every invariant measure is the sum of ergodic invariant measures @xcite .
a conjugacy map sends the invariant measure to invariant measure and the ergodic measures to the ergodic measures . in our case ,
the group of affine transformations @xmath5 , is the extension of two groups isomorphic to @xmath7 , hence is amenable .
it is well known that the only invariant measures for the suspension action @xmath189 are locally the images through the canonical projection @xmath190 of the measures @xmath191 where @xmath184 is a @xmath188-invariant measure on @xmath4 and @xmath192 denotes the lebesgue measure of @xmath7 . the proof is actually contained in property [ carmesureinv ] .
it is well known also that the map @xmath193 on the dyadic set of integers @xmath81 , admits only one invariant probability measure : the haar probability measure on @xmath81 .
hence the suspension of this action @xmath194 admits only one invariant probability measure . by proposition [ premsuspension ] , @xmath195 has only one invariant probability measure @xmath196 .
notice that the map @xmath37 preserves @xmath195 , and since @xmath197 , the probability @xmath198 is @xmath86-invariant and hence @xmath37 preserves @xmath196 .
nevertheless , the local product decomposition of @xmath196 is not invariant by @xmath37 , because @xmath37 divides by 2 the length of the intervals of the @xmath86-orbit .
so @xmath37 has to inflate the haar measure on @xmath81 by a factor @xmath10 .
[ carmesureinv ] if @xmath70 is an element in @xmath128 , then any finite invariant measure of @xmath199 is locally the image through the projection @xmath200 of a measure @xmath201 where * @xmath196 is the only invariant probability measure of @xmath202 ; * @xmath203 is the haar measure of @xmath204 ; * @xmath184 is a finite invariant measure of @xmath139 .
it is enough to prove this for an ergodic finite @xmath5-invariant measure @xmath205 on the suspension @xmath206 .
since @xmath149 acts cocompactly on @xmath207 , @xmath208 defines a finite measure on a fundamental domain of @xmath149 , and the sum of all the images of this measure by iterates of @xmath149 and @xmath209 defines a @xmath131-finite measure @xmath210 on @xmath207 which is @xmath5 and @xmath149-invariant .
let @xmath211 be the projection to the second coordinate , then @xmath212 is a shift invariant measure on @xmath133 .
the measure @xmath210 can be disintegrated over @xmath213 by a family of measures @xmath214 defined for @xmath215-almost every @xmath216 on @xmath217 such that @xmath218 for any borel sets @xmath219 and @xmath220 .
the @xmath5-invariance of @xmath210 implies that the measures @xmath221 are @xmath5-invariant for almost all @xmath222 .
the projection to the first coordinate @xmath223 is @xmath86-equivariant .
the measures @xmath224 are then @xmath86-invariant measures , hence are proportional to @xmath196 .
each measure @xmath221 can be disintegrated over @xmath196 by a family of measures @xmath225 on @xmath226 defined for @xmath196-almost every @xmath227 , so that @xmath228 for any borel sets @xmath229 and @xmath230 .
each measure @xmath221 is invariant under the action of transformations of the kind @xmath231 for @xmath232 .
it is then straightforward to check that the measures @xmath233 are multiplication - invariant for almost every @xmath59 . by unicity of the haar measure
, there exists a measurable positive function @xmath234 defined almost everywhere so that @xmath235 .
the @xmath86-invariance of the measures @xmath221 implies that the map @xmath236 is almost surely constant along the @xmath86-orbits , and the @xmath149-invariance of @xmath210 implies that @xmath236 is almost surely constant along the @xmath149-orbits .
this defines then a measurable map on the quotient space by @xmath149 which is @xmath5-invariant , the ergodicity of @xmath208 implies this map is almost surely constant .
notice that an invariant measure on @xmath237 can be decomposed locally into the product of a measure on a cantor set by a measure along the leaves .
since the map @xmath37 does not preserve the transversal measure on @xmath81 in @xmath180 , the holonomy groupoid of @xmath182 does not preserve the transversal measure on the cantor set . +
the @xmath5-action is locally free and acts transitively on each leaf , so each orbits inherits a hyperbolic 2-manifolds structure .
actually , @xmath182 can be equipped with a continuous metric with a constant curvature @xmath238 in restriction to the leaves .
the invariant measures of @xmath182 have then also a geometric interpretation in terms of _ harmonic _ measures , a notion introduced by l. garnett in @xcite .
a probability measure @xmath184 on @xmath239 is harmonic if @xmath240 for any continuous function @xmath188 with restriction to leaves of class @xmath241 , where @xmath242 denotes the laplace - beltrami operator in restriction on each leaf .
actually , it is shown in @xcite , that on @xmath182 the notions of harmonic and invariant measures are the same and such measures can be described in terms of inverse limit of vectoriel cones .
we gather this section with results on groupoids and their @xmath0-algebras . good material on this topic can be found in @xcite .
let us fix first some notations .
let @xmath243 be a locally compact groupoid , with base space @xmath4 , range and source maps respectively @xmath244 and @xmath245 . recall that @xmath4 can be viewed as a closed subset of @xmath243 ( the set of units ) .
for any element @xmath59 of @xmath4 , we set @xmath246 and @xmath247 let us denote for any @xmath248 in @xmath243 by @xmath249 the left translation by @xmath248 .
thourought this section , all the groupoids will be assumed locally compact and second countable .
recall that a haar system @xmath192 for @xmath243 is a family @xmath250 of borelian measures on @xmath243 such that 1 .
the support of @xmath251 is @xmath252 ; 2 . for any @xmath188 in @xmath253 ,
the map @xmath254 is continuous ; 3 .
@xmath255 for all @xmath256 in @xmath257 .
our prominent examples of groupoid will be _ semi - direct product groupoid _ : let @xmath258 be a locally compact group acting on a locally compact space @xmath4 .
the semi - direct product groupoid @xmath259 of @xmath4 by @xmath258 is defined by * @xmath260 as a topological space ; * the base space is @xmath4 and the structure maps are @xmath261 and @xmath262 ; * the product is @xmath263 for @xmath59 in @xmath4 and @xmath236 and @xmath264 in @xmath258 .
let @xmath184 be a left haar mesure on @xmath258 .
then the groupoid @xmath259 is equipped with a haar system @xmath265 given for any @xmath188 in @xmath266 and any @xmath59 in @xmath4 by @xmath267 .
recall that any automorphism @xmath268 of a groupoid @xmath243 induces a homeomorphism of its base space @xmath4 that we shall denote by @xmath269 .
let @xmath243 be a groupoid with base space @xmath4 equipped with a haar system @xmath270 .
a groupoid automorphism @xmath271 is said to preserve the haar system @xmath192 if there exists a continuous function @xmath272 such that for any @xmath59 in @xmath4 the measures @xmath273 and @xmath274 on @xmath275 are in the same class and @xmath276 restricted to @xmath275 is @xmath277 .
the map @xmath276 is called the density of @xmath268 .
[ rem - density]let @xmath243 be a groupoid with base space @xmath4 and haar system @xmath270 and let @xmath271 be an automorphism of groupoid preserving the haar system @xmath192 . 1 .
since @xmath278 for any @xmath248 in @xmath243 , we get that @xmath279 since @xmath280 is a measure on @xmath281 absolutly continuous with respect to @xmath282 with density @xmath283 we see that @xmath284 and @xmath276 coincide on @xmath285 .
in particular @xmath276 is constant on @xmath286 for any @xmath59 in @xmath4 .
the automorphism of groupoid @xmath287 also preserves the haar system @xmath192 and @xmath288 .
let @xmath243 be a groupoid with base space @xmath4 , range and source map @xmath119 and @xmath289 and let @xmath271 be a groupoid automorphism . using the notations of section [ subsec - hull ] the suspension of the groupoid @xmath243 respectively to @xmath268
is the groupoid @xmath290 with base space @xmath291 . for
any @xmath248 in @xmath243 and @xmath116 in @xmath7 , let us denote by @xmath292 $ ] the class of @xmath293 in @xmath294 .
* the range map @xmath295 and the source map @xmath296 are defined in the following way : * * @xmath297)=[r(\gamma),t]$ ] for every @xmath248 in @xmath243 and @xmath116 in @xmath7 ; * * @xmath298)=[s(\gamma),t]$ ] for every @xmath248 in @xmath243 and @xmath116 in @xmath7 ; * let @xmath248 and @xmath299 be elements of @xmath243 such that @xmath300 and let @xmath116 be in @xmath7 , then @xmath292\circ[\gamma',t]=[\gamma\circ\gamma',t]$ ] ; * @xmath292^{-1}=[\gamma^{-1},t]$ ] .
there is an action of @xmath7 on @xmath294 by automorphisms given for @xmath289 in @xmath7 and @xmath292 $ ] in @xmath294 by @xmath301=[\gamma , t+s]$ ] .
[ rem - pr ] let @xmath243 be a groupoid with base space @xmath4 equipped with a haar system @xmath270 and let @xmath271 be an automorphism preserving the haar system @xmath192 .
let us assume that @xmath302
. then @xmath294 admits a haar system @xmath303}\right)_{[x , t]\in x_{\alpha}}$ ] given for any @xmath304 $ ] in @xmath305 and any continuous fonction @xmath188 in @xmath306}\right)$ ] by @xmath307}_\alpha(f)=\int_{{\mathcal{g}}^x}\rho_\alpha(\gamma)^{-t}f([\gamma , t])d\lambda^x(\gamma).\ ] ] * let us prove first that the definition of @xmath308}_\alpha(f)$ ] for @xmath304 $ ] in @xmath305 and @xmath188 in @xmath306}\right)$ ] makes sense .
@xmath309)d\lambda^x(\gamma)&=&\int_{{\mathcal{g}}^x}\rho_\alpha(\alpha(\gamma))^{-t}f([\alpha^{-1}(\alpha(\gamma),t])d\lambda^x(\gamma)\\ & = & \int_{{\mathcal{g}}^{\alpha(x)}}\rho_\alpha(\gamma)^{-t+1}f([\alpha^{-1}(\gamma),t])d\lambda^{\alpha(x)}(\gamma)\\ & = & \int_{{\mathcal{g}}^{\alpha(x)}}\rho_\alpha(\gamma)^{-t+1}f([\gamma , t-1])d\lambda^{\alpha(x)}(\gamma).\\\end{aligned}\ ] ] * it is clear that the continuity condition is fullfilled .
let us show then that @xmath310})_{[x , t]\in x_{\alpha}}$ ] is a haar system .
let @xmath311 be an element of @xmath243 , let @xmath116 be a real number and let @xmath188 be a function in @xmath312}\right)$ ] .
then @xmath313}(f)&=&\int_{{\mathcal{g}}^{r(\gamma')}}\rho_\alpha(\gamma)^{-t}f([\gamma , t])d\lambda^{r(\gamma')}(\gamma)\\ & = & \int_{{\mathcal{g}}^{s(\gamma')}}\rho_\alpha(\gamma'\cdot\gamma)^{-t}f([\gamma'\cdot\gamma , t])d\lambda^{s(\gamma')}(\gamma)\\ & = & \int_{{\mathcal{g}}^{s(\gamma')}}\rho_\alpha(\gamma)^{-t}f([\gamma'\cdot\gamma , t])d\lambda^{s(\gamma')}(\gamma)\\ & = & \lambda_\alpha^{[s(\gamma'),t]}(f\circ l_{[\gamma',t]}),\end{aligned}\ ] ] where the third equality holds in view of remark [ rem - density ] .
let us recall first the construction of the reduced @xmath0-algebra @xmath314 associated to a groupoid @xmath243 with base @xmath4 and haar system @xmath270 .
let @xmath315 be the @xmath316-hilbert completion of @xmath253 equipped with the @xmath316-valued scalar product @xmath317 for @xmath97 and @xmath318 in @xmath253 and @xmath59 in @xmath4 , i.e the completion of @xmath253 with respect to the norm @xmath319 .
the @xmath316-module structure on @xmath253 extends to @xmath315 and @xmath320 extends to a @xmath316-valued scalar product on @xmath315 .
recall that an operator @xmath321 is called adjointable if there exists an operator @xmath322 , called the adjoint of @xmath40 such that @xmath323 for all @xmath97 and @xmath318 in @xmath315 .
notice that the adjoint , when it exists is unique and that operator that admits an adjoint are automatically @xmath316-linear and continuous .
the set of adjointable operators on @xmath315 is then a @xmath0-algebra with respect to the operator norm .
then any @xmath188 in @xmath253 acts as an adjointable operator on @xmath315 by convolution @xmath324 where @xmath97 is in @xmath253 , the adjoint of this operator being given by the action of @xmath325 .
the convolution product provides a structure of involutive algebra on @xmath253 and using the action defined above , this algebra can be viewed as a subalgebra of the @xmath0-algebra of adjointable operators of @xmath315 .
the reduced @xmath0-algebra @xmath314 is then the closure of @xmath253 in the @xmath0-algebra of adjointable operators of @xmath315 .
namely , if we define for @xmath59 in @xmath4 the measure on @xmath286 by @xmath326 for any @xmath97 in @xmath327 , then @xmath315 is a continuous field of hilbert spaces over @xmath4 with fiber @xmath328 at @xmath59 in @xmath4 .
the corresponding @xmath316-structure on @xmath314 is then given for @xmath236 in @xmath316 by the multiplication by @xmath329 .
let @xmath258 be a locally compact group acting on a locally compact space @xmath4 , and consider the semi - direct product groupoid @xmath259 equipped with a haar system arising from the haar measure on @xmath258 .
then @xmath330 is the usual reduced crossed product @xmath331 .
let us denote for any @xmath59 in @xmath4 by @xmath332 the representation of @xmath314 on the fiber @xmath328 .
then for any @xmath188 in @xmath314 , we get that @xmath333 .
[ lem - mt ] let @xmath243 be a locally compact groupoid with base space @xmath4 equipped with a haar system @xmath270 and let @xmath271 be an automorphism preserving the haar system @xmath192 .
let us define the continuous map @xmath334 .
then there exists a unique automorphism @xmath335 of the @xmath0-algebra @xmath314 such that for every @xmath188 in @xmath253 we have @xmath336 .
the map @xmath337 extends uniquely to a continuous linear and invertible map @xmath338 such that @xmath339 for all @xmath59 in @xmath4 .
its inverse @xmath340 is defined by @xmath341 for all @xmath97 in @xmath253 .
let us define @xmath342 then @xmath343 for all @xmath188 in @xmath253 . recall that if @xmath344 is a @xmath0-algebra and if @xmath345 is an automorphism of @xmath344 then the mapping torus of @xmath344 is the @xmath0-algebra @xmath346,a)\text { such that } \beta(f(1))=f(0)\}.\ ] ] namely
, the mapping torus @xmath347 can be viewed as the algebra of continuous function @xmath348 such that @xmath349 for all @xmath116 in @xmath7 . in this picture , there is an action of @xmath7 on @xmath347 by translations defined for @xmath116 in @xmath7 and @xmath188 in @xmath347 by @xmath350 for any @xmath289 in @xmath7 .
translations then define a strongly continuous action by automorphisms @xmath351 of @xmath7 on @xmath347 . by the mapping torus isomorphism , we have a natural morita equivalence between @xmath352 and @xmath353 .
let @xmath268 be an automorphism of a groupoid @xmath243 preserving a haar system @xmath192 and with density @xmath276 .
for a function @xmath188 in @xmath354 , we define @xmath355 in @xmath356\times{\mathcal{g}})\subset c([0,1],c_r^*({\mathcal{g}},\lambda))$ ] by @xmath357)$ ] .
we can check easily that @xmath355 belongs to the mapping torus @xmath358 .
[ prop - iso ] let @xmath243 be a locally compact groupoid with base space @xmath4 equipped with a haar system @xmath270 and let @xmath271 be an automorphism preserving the haar system @xmath192 such that @xmath359 .
then there is an unique automorphism of @xmath0-algebras @xmath360 such that @xmath361 for any @xmath188 in @xmath354 .
let @xmath188 be a function of @xmath354
. then @xmath362}\|\hat{f}(t,\bullet))\| _ { c_r^*({\mathcal{g}},\lambda)}\\ & = & \sup_{t\in[0,1],\,x\in x}\|\nu_x(\hat{f}(t,\bullet))\|\end{aligned}\ ] ] on the other hand , @xmath363,\,x\in x}\|\nu_{[x , t]}(f)\|,\ ] ] where @xmath364}$ ] is the representation of @xmath365 on the fiber @xmath366},\lambda_{\alpha,[x , t]})$ ] at @xmath304\in ( x\times{\mathbb r})/{\mathcal a}_{\alpha_x}$ ] .
if we define for @xmath116 in @xmath367 $ ] the map @xmath368 $ ] , then @xmath369})\to c_c({\mathcal{g}}_x):\,\phi\mapsto{\rho'}_\alpha^{-t/2}\phi\circ\pi_t\ ] ] extends to an isometry @xmath370},\lambda_{\alpha,[x , t]})\to{\mathcal{l}}^2({\mathcal{g}}_x,\lambda_x)$ ] and @xmath371 conjugate @xmath364}(f)$ ] and @xmath372 .
thus @xmath373 and @xmath374 extends to a monomorphism @xmath375 .
the set @xmath376\times{\mathcal{g}})\text { such that } h(1,\alpha(\gamma))={\rho'}_\alpha^{t/2}\rho_\alpha^{t/2}h(0,\gamma)\text { for all } \gamma\in{\mathcal{g}}\}\ ] ] is dense in @xmath358 .
let us define for an element @xmath236 of @xmath377 the map @xmath378 as the unique map such that * @xmath379 for all @xmath248 in @xmath243 and @xmath116 in @xmath367 $ ] ; * @xmath380 for all @xmath248 in @xmath243 and @xmath116 in @xmath7 .
then @xmath381 defines a continuous map of @xmath354 whose image under @xmath382 is @xmath236 .
hence @xmath382 has dense range in @xmath358 and thus is surjective .
[ rem - action ] with the notations of above proposition , let us define for a real @xmath289 the automorphism of groupoid @xmath383\mapsto[\gamma , s+t]$ ] .
then @xmath384 is preserving the haar system @xmath385})_{[x , t]\in x_{\alpha}}$ ] with density @xmath386\mapsto \rho_\alpha(\gamma)^{s}.\ ] ] we obtain from lemma [ lem - mt ] an automorphism @xmath387 of @xmath365 which gives rise to a strongly continuous action of @xmath7 on @xmath365 by automorphism .
the isomorphism @xmath360 of proposition [ prop - iso ] is then @xmath7-equivariant , where the action of @xmath7 on @xmath358 is the action @xmath388 associated to a mapping torus .
as we have seen before , the closure @xmath389 of @xmath390 for the tiling topology is the suspension @xmath391 of the odometer homeomorphism @xmath392 , where @xmath81 is the dyadic completion of the integers .
the @xmath7-algebra @xmath393 is then the mapping torus algebra of @xmath394 with respect to automorphism induced by @xmath395 . in consequence ,
the crossed product algebras @xmath396 and @xmath397 are morita equivalent .
the purpose of this section is to recall the explicit description of the isomorphism @xmath398 arising from this morita equivalence . for this ,
let us define on @xmath399 the @xmath397-valued inner product @xmath400for @xmath401 and @xmath402 in @xmath399 and @xmath403 in @xmath404 .
this inner product is positive and gives rise to a right @xmath397-hilbert module @xmath405 , the action of @xmath397 being given for @xmath236 in @xmath406 and @xmath401 in @xmath399 by @xmath407 for @xmath403 in @xmath408 .
the right @xmath397-hilbert module @xmath405 is also equipped with a left action of @xmath409 given for @xmath188 in @xmath410 and @xmath401 in @xmath399 by @xmath411,s)\xi(\omega , t - s)ds\]]for @xmath403 in @xmath408 .
we get in this way a @xmath412 imprimitivity bimodule which implements the morita equivalence we are looking for .
actually , there is an isomorphism of right @xmath397-hilbert module @xmath413)\otimes c(\omega)\rtimes{\mathbb z}\ ] ] defined in a unique way by @xmath414 for @xmath415 in @xmath416 supported in @xmath417 , where @xmath418 is the unitary of @xmath397 corresponding to the positive generator of @xmath73 . using the right @xmath409-module structure of the @xmath412 imprimitivity bimodule @xmath405 and the isomorphism @xmath170
, we get an isomorphism @xmath419))\otimes c(\omega)\rtimes{\mathbb z}.\ ] ] this isomorphism can be described as follows .
let us define for @xmath188 and @xmath415 in @xmath420)$ ] the rank one operator @xmath421)\to{l}^2([0,1]);\,h\mapsto f\langle g , h\rangle.\ ] ] we define for @xmath401 and @xmath402 in @xmath399 the continuous function of @xmath410 @xmath422,t)=\sum_{k\in{\mathbb z}}\xi(\omega+k , s - k)\bar{\xi'}(\omega+k , s - t - k)\ ] ] for all @xmath423 in @xmath81 and @xmath289 and @xmath116 in @xmath7 .
it is straightforward to check that @xmath424 is well defined and that @xmath425for all @xmath426 in @xmath399 .
if we set for @xmath188 and @xmath415 in @xmath416 with support in @xmath417 and for @xmath97 in @xmath394 , @xmath427 , @xmath428 and @xmath429 , then the image of @xmath424 under the isomorphism of equation ( [ equ - morita ] ) is @xmath430))\otimes c(\omega)\rtimes{\mathbb z}$ ] and moreover , @xmath431,t)=\sum_{k\in{\mathbb z}}f(s - k)\bar{\phi}(\omega+k)\bar{g}(s - t - k).\ ] ] the image of @xmath432 under the isomorphism of equation ( [ equ - morita ] ) is @xmath433))\otimes c(\omega)\rtimes{\mathbb z}$ ] and moreover , @xmath434,t)=\sum_{k\in{\mathbb z}}f(s - k)\bar{g}(s+1-t - k).\ ] ] let us define the automorphism @xmath268 of the groupoid @xmath435 in the following way * @xmath436,t)=([\omega/2,s/2],t/2)$ ] if @xmath423 is even ; * @xmath436,t)=([(\omega+1)/2,(s+1)/2],t/2)$ ] if @xmath423 is odd .
notice that @xmath437,t)=([2\omega,2s],2t)$ ] for all @xmath423 in @xmath81 and @xmath289 and @xmath116 in @xmath7 .
then @xmath268 preserves the haar system of @xmath435 arising from the haar mesure on @xmath7 and has constant density @xmath438 . hence according to lemma [ lem - mt ]
, the automorphism of groupoid @xmath268 induces an automorphism @xmath335 of @xmath0-algebra @xmath409 such that @xmath439 for all @xmath236 in @xmath440 .
we are now in position to describe how @xmath335 is transported under the isomorphism of equation ( [ equ - morita ] ) to an automorphism @xmath441 of @xmath442))\otimes c(\omega)\rtimes{\mathbb z}$ ] . with @xmath401 , @xmath402 and @xmath443
as defined above , @xmath444,t)&=&2\theta^\omega_{\xi,\xi'}([2\omega,2s],2t)\\ \nonumber & = & 2\sum_{k\in{\mathbb z}}f(2s - k)\bar{\phi}(2\omega+k)\bar{g}(2s-2t - k)\\ & = & 2\sum_{k\in{\mathbb z}}f(2s-2k)\bar{\phi}(2\omega+2k)\bar{g}(2s-2t-2k)+\\ \nonumber & & 2\sum_{k\in{\mathbb z}}f(2s-2k-1)\bar{\phi}(2\omega+2k-1)\bar{g}(2s-2t-2k-1)\end{aligned}\ ] ] and @xmath445,t)&=&2\theta^\omega_{\xi,\xi''}([2\omega,2s],2t)\\ \nonumber & = & 2\sum_{k\in{\mathbb z}}f(2s - k)\bar{g}(2s+1 - 2t - k)\\ & = & 2\sum_{k\in{\mathbb z}}f(2s-2k)\bar{g}(2s+1 - 2t-2k)+\\ \nonumber&&\quad\quad 2\sum_{k\in{\mathbb z}}f(2s-2k-1)\bar{g}(2s-2t-2k).\end{aligned}\ ] ] to complete the description of the automorphism @xmath441 of @xmath442))\otimes c(\omega)\rtimes{\mathbb z}$ ] corresponding to @xmath335 , we need to introduce some further notations .
we define the partial isometries @xmath446 and @xmath66 of @xmath420)$ ] by * @xmath447 if @xmath448 $ ] and @xmath449 otherwise ; * @xmath450 if @xmath451 $ ] and @xmath452 otherwise ; * @xmath453 if @xmath448 $ ] and @xmath454 otherwise , for @xmath188 in @xmath455)$ ] .
let use define also the endomorphisms @xmath456 and @xmath457 of the @xmath0-algebra @xmath394 by @xmath458 and @xmath459 , for @xmath97 in @xmath394 and @xmath423 in @xmath81 .
using this notations , equations ( [ equ - th7 ] ) and ( [ equ - th8 ] ) can be rewriten as @xmath460,t)=\sum_{k\in{\mathbb z}}u_0 f(s - k)w_0\bar{\phi}(\omega+k)u_0\bar{g}(s - t - k)+\sum_{k\in{\mathbb z}}u_1 f(s - k)w_1\bar{\phi}(\omega+k)u_1\bar{g}(s - t - k)\ ] ] and @xmath461,t)=\sum_{k\in{\mathbb z}}u_0 f(s - k)u_1\bar{g}(s - t - k+1)+\sum_{k\in{\mathbb z}}u_1 f(s - k)u_0\bar{g}(s - t - k).\ ] ] thus , in view of equations ( [ equ - th1 ] ) and ( [ equ - th2 ] ) , we get that @xmath462 and @xmath463 from this we deduce @xmath464 and @xmath465 where the second equality holds since @xmath466 and @xmath467 and the third holds since @xmath468 . in consequence , if we extends @xmath441 to the multiplier algebra of @xmath442))\otimes c(\omega)\rtimes{\mathbb z}$ ] , we finally obtain that the automorphism @xmath441 is the unique homomorphism of @xmath0-algebra such that @xmath469 and @xmath470 where @xmath471 is in @xmath442))$ ] , @xmath97 is in @xmath394 and @xmath472 and @xmath473 are viewed as multipliers of @xmath442))\otimes c(\omega)\rtimes{\mathbb z}$ ]
. the following lemma will be helpful to compute the @xmath1-theory of the @xmath0-algebra of the penrose hyperbolic tiling . for short
, we will denote from now on @xmath442))$ ] by @xmath474 .
[ lem - action - k1 ] let @xmath475 be the unitarisation of @xmath476 and let @xmath188 be a norm one function of @xmath420)$ ]
. then the unitaries @xmath477 and @xmath478 of @xmath344 are homotopic .
if we set @xmath479 and complete to a hilbertian base @xmath480 of @xmath420)$ ] , then @xmath481 is a hilbertian basis of @xmath420)$ ] . in this base
the unitary of equation ( [ equ - action - k1 ] ) can be written down as @xmath482 which is homotopic to @xmath483 all unitaries that can be writen down in such way in some hilbertian basis of @xmath420)$ ] are homotopic and since this is the case for @xmath484 , we get the result .
let us consider the semi - direct product groupoid @xmath485 corresponding to the diagonal action of @xmath7 on @xmath486 , by translations on @xmath389 and trivial on @xmath487 .
let us denote by @xmath488 the haar system provided by the left haar mesure on @xmath7 .
let us define the groupoid automorphism @xmath489 .
then @xmath490 preserves the haar system @xmath192 with constant density @xmath491 and thus according to lemma [ rem - pr ] the suspension groupoid @xmath492 admits a haar system @xmath493 .
the semi - direct product groupoid @xmath494 , where @xmath7 acts on @xmath495 by translations , is equipped with an action of @xmath7 by automorphisms @xmath496 for any @xmath116 in @xmath7 . the automorphism @xmath497 preserves the haar system with constant density @xmath498 and thus in view of proposition [ prop - iso ] induced a strongly continuous action of @xmath7 on the crossed product @xmath0-algebra @xmath499 . [ lem - psi ]
let @xmath70 be an element of @xmath500 .
then there is a unique isomorphism of groupoids @xmath501 such that : 1 .
@xmath502)=(\mathcal{p}(w)+x , y)$ ] for all @xmath59 and @xmath503 in @xmath7 ; 2 .
@xmath504 is equivariant with respect to the actions of @xmath7 ; 3 .
@xmath505 is the haar system on @xmath494 provided by the haar measure on @xmath7 . with notations of the proof of proposition [ caracterisation ] ,
let us define @xmath506)$ ] , were @xmath507 is in @xmath389 and @xmath222 is in @xmath508 .
then the map @xmath509 is continuous and since @xmath510 , the continuous map @xmath511 induces a continuous homomorphism of groupoids @xmath512 this map is clearly one - to - one since the equality @xmath513 for @xmath116 in @xmath7 , @xmath507 and @xmath514 in @xmath389 and @xmath515 and @xmath222 in @xmath516 holds if and only if @xmath116 is integer , @xmath517 and @xmath518 . to prove that @xmath504 is onto ,
let us remark that any element of @xmath495 can be written as @xmath519 , with @xmath14 in @xmath7 , @xmath507 in @xmath389 and @xmath222 in @xmath487 .
we get then @xmath520)=(r_{2^{a}}\mathcal{t}(w'),t)\ ] ] for all @xmath116 in @xmath7 .
it is then straightforward to check that condition ( 3 ) of the lemma is satisfied .
the uniqueness of @xmath504 is a consequence on one hand of its equivariance and on the other hand of the density of the @xmath7-orbit of @xmath36 in @xmath389 . as a consequence of lemma [ lem - psi ]
, we get the map @xmath521 induces an @xmath7-equivariant isomorphism @xmath522 [ prop - morita ] using the notations of lemmas [ lem - mt ] and [ lem - psi ] , the @xmath0-algebras @xmath523 and @xmath524 are morita equivalent .
recall that @xmath525 , where the group @xmath526 acts on @xmath527 by multiplication .
iterate crossed products leads to an isomorphism @xmath528if we identify the groups @xmath527 and @xmath526 using the isomorphism @xmath529 this provides the action under consideration in lemma [ lem - psi ] of @xmath7 on @xmath499 and hence , the algebras @xmath523 and @xmath530 are isomorphic . in view of lemma [ prop - iso ] and of remark [ rem - action ] , the @xmath0-algebra @xmath523 is isomorphic to @xmath531 .
but since @xmath532 is the mapping torus algebra with respect to the automorphism @xmath533 , the crossed product @xmath0-algebra @xmath531 is morita equivalent to @xmath524 and hence we get the result .
let us consider the semi - direct groupoid @xmath485 corresponding to the diagonal action of @xmath7 on @xmath486 , by translations on @xmath389 and trivial on @xmath516 .
according to proposition [ prop - morita ] we have an isomorphism @xmath534 induced by the morita equivalence . in order to compute this @xmath1-theory group ,
we need to recall some basic facts concerning the @xmath1-theory group of a crossed product of a @xmath0-algebra @xmath344 by an action of @xmath73 provided by an automorphism @xmath210 of @xmath344 .
this @xmath1-theory can be computed by using the pimsner - voiculescu exact sequence @xcite @xmath535 where @xmath536 is the homomorphism induced in @xmath1-theory by the inclusion @xmath537 and @xmath538 is the homomorphism in @xmath1-theory induced by @xmath210 .
the vertical maps are given by the composition @xmath539 where * @xmath540 is the mapping torus of @xmath344 with respect to the action @xmath210 endowed , with its associated action @xmath541 of @xmath7 ; * the first map is induced by the morita equivalence between @xmath542 and @xmath543 ; * the second map is the thom - connes isomorphism ; * the third map is induced in k - theory by the evaluation map @xmath544 for an automorphism @xmath170 of an abelian group @xmath239 , let us define @xmath545 as the set of invariant elements of @xmath239 and by @xmath546 the set of coinvariant elements .
we then get short exact sequences @xmath547 and @xmath548 moreover the inclusions in these exact sequences are induced by @xmath536 .
the first step in the computation of @xmath549 is provided by next lemma , which is straightforward to prove .
[ lem - cantor ] let @xmath550 be a cantor set and let us denote by @xmath551 the algebra of continuous and integer valued functions on @xmath550 . 1 .
we have an isomorphism @xmath552 $ ] .
2 . @xmath553 , where for a compact - open subset @xmath554 of @xmath550 , then @xmath555 stands for the characteristic function of @xmath554 .
plugging @xmath524 into the short exact sequences ( [ equ - pv0 ] ) and ( [ equ - pv1 ] ) , we get @xmath556 and @xmath557 according to equation ( [ equ - morita ] ) , the @xmath0-algebra @xmath314 is isomorphic to @xmath558 .
the @xmath1-theory of @xmath314 can be the computed by using the knneth formula : in view of lemma [ lem - cantor ] , @xmath559 is torsion free and @xmath560 and by morita equivalence , we get that @xmath561 and @xmath562 this isomorphism , up to the morita equivalence and to the isomorphism of equation ( [ equ - morita ] ) are implemented by the external product in @xmath1-theory and will be precisely described later on .
once again , @xmath563 can be computed from the short exact sequences ( [ equ - pv0 ] ) and ( [ equ - pv1 ] ) , and we get , using lemma [ lem - cantor ] that @xmath564 and @xmath565 the isomorphism of equation ( [ equ - mt1 ] ) is induced by the composition @xmath566 which factorizes through @xmath567 , where the first map is described in lemma [ lem - cantor ] , and the second map is induced on @xmath1-theory by the inclusion @xmath568 . in the first isomorphism of equation ( [ equ - mt2 ] ) the class of @xmath569 $ ] in @xmath570 of the unitary @xmath418 of @xmath397 corresponding to the positive generator of @xmath73 is mapped to the constant function @xmath120 of @xmath571 .
[ lem - coinv ] let @xmath196 be the haar measure on @xmath81
. then 1 .
@xmath572 is in @xmath573 $ ] for all @xmath188 in @xmath571 ; 2 .
@xmath574;f\mapsto\int fd\nu$ ] factorizes through an isomorphism @xmath575.\ ] ] it is enought to check the first point for characteristic function of compact - open subset of @xmath81 . for an integer @xmath576 and @xmath471 in @xmath577 , we set @xmath578
. then @xmath579 is a basis of compact - open neighborhoods for @xmath81 and thereby , every compact - open subset of @xmath81 is a finite disjoint union of some @xmath580 . since @xmath581
, we get the first point . the measure @xmath184 being invariant by translation , the map @xmath582;f\mapsto\int fd\nu\ ] ] factorizes through a group homomorphism @xmath583 $ ] .
this homomorphism admits a cross - section @xmath584\to { \operatorname{coinv}}c(\omega,{\mathbb z});\ , 2^{-n}\mapsto [ \chi_{f_{n,0}}].\ ] ] this map is well defined since @xmath585 and thus @xmath586=[\chi_{f_{n+1,0}}]+[\chi_{2^n+f_{n+1,0}}]=2[\chi_{f_{n+1,0}}]\ ] ] in @xmath587 . since the @xmath588 generates @xmath571 as an abelian group , it is enought to check that the cross - section of equation ( [ equ - cross - sec ] ) is a left inverse on @xmath589 , which is true since @xmath590= [ \chi_{k+f_{n,0}}]=[\chi_{f_{n,0}}]$ ] in @xmath587 .
[ prop - rhull ] let @xmath591)\cong c(z_w,{\mathbb z})\otimes{\mathbb z}[1/2]$ ] be the algebra of continuous function on @xmath516 , valued in @xmath573 $ ] ( equipped with the discrete topology ) . then with the notations of the proof of lemma [ lem - coinv ] , we have isomorphisms 1 .
@xmath592)&\stackrel{\cong}{\longrightarrow}&k_0(c(z_w)\otimes c(\omega)\rtimes{\mathbb z})\\ \frac{\chi_e}{2^{n}}&\mapsto&[\chi_e\otimes\chi_{f_{n,0}}],\end{aligned}\ ] ] + where @xmath554 is a compact - open subset of @xmath516 and @xmath555 is its characteristic function .
2 . @xmath593,\end{aligned}\ ] ] + where @xmath418 is the unitary of @xmath397 corresponding to the positive generator of @xmath73 .
as we have already mentionned , @xmath594 is torsion free and the knneth formula provides isomorphisms @xmath595 where @xmath41 and @xmath596 are some matrix projectors with coefficients respectively in @xmath597 and @xmath397 , and @xmath598 where @xmath41 is a projector in @xmath599 and @xmath68 is a unitary in @xmath600 . the proposition is then a consequence of lemmas [ lem - cantor ] , [ lem - coinv ] and of the discussion related to equations ( [ equ - mt1 ] ) and ( [ equ - mt2 ] ) . in order to compute the invariants and the coinvariants of @xmath601
we will need a carefull description of the action induced in @xmath1-theory by the automorphism @xmath602 of @xmath558 , where @xmath603 is the automorphism of @xmath604 induced by the shift @xmath131 and where @xmath441 was defined in section [ sec - penrose - uncolor ] .
[ lem - mt1 ] if we equip @xmath558 with the @xmath73-action provided by @xmath602 and under the @xmath73-equivariant isomorphism @xmath605 the action induced by @xmath606 on @xmath607)$ ] and on @xmath608 are given by the automorphisms of abelian groups @xmath609)&\to & c(z_w,{\mathbb z}[1/2])\\ f&\mapsto & 2f\circ \sigma^{-1}\end{aligned}\ ] ] and @xmath610 according to proposition [ prop - rhull ] and using the morita equivalence between @xmath611 and @xmath558 , in order to describe @xmath612 , we have to compute the image under @xmath613 of @xmath614\in k_0(c(z_w)\otimes{\mathcal{k}}\otimes c(\omega)\rtimes{\mathbb z})\ ] ] where , * @xmath555 is the characteristic function of a compact - open subset @xmath554 of @xmath516 ; * @xmath615 is the characteristic function of @xmath616 for @xmath617 ; * @xmath618 is the rank one projector associated to a norm @xmath120 function @xmath188 of @xmath619)$ ] .
we have @xmath620where the last equality holds since @xmath621 and @xmath622 . since @xmath623 is again a rank one projector , then up to the morita equivalence between @xmath611 and @xmath558 , the image of @xmath624\in k_0(c(z_w)\otimes c(\omega)\rtimes{\mathbb z})$ ] under @xmath613 is @xmath625\in k_0(c(z_w)\otimes c(\omega)\rtimes{\mathbb z})$ ] . using proposition [ prop - rhull ] , this completes the description of @xmath612 . for @xmath626 ,
notice first that up to the isomorphism @xmath627 provided by the knneth formula , the action of @xmath613 is @xmath628 and then the result is a consequence of lemma [ lem - action - k1 ] and of proposition [ prop - rhull ] .
let us equip @xmath591)$ ] and @xmath629 with the @xmath73-actions respectively provided by @xmath612 and @xmath626 .
then since @xmath630 for any @xmath236 in @xmath591)$ ] , we get that @xmath631)=\{0\}$ ] we are now in position to get a complete description of the @xmath1-theory of @xmath523 .
in view of the short exact sequences of equations ( [ eq - exsq1 ] ) and ( [ eq - exsq2 ] ) , the two following theorems are then consequences of lemma [ lem - mt1 ] and of proposition [ prop - rhull ] .
[ thm - k0-gl ] we have a short exact sequence @xmath632)\stackrel{\iota_0}{\to } k_0(c(x^g_{\mathcal{p}(w)})\rtimes g)\to{\operatorname{inv}}c(z_w,{\mathbb z})\to 0,\ ] ] where up to the morita equivalence @xmath633 , the element @xmath634 $ ] is the image of @xmath635\in k_0(c(z_w)\otimes{\mathcal{k}}\otimes c(\omega)\rtimes{\mathbb z})$ ] under the homomorphism induced in @xmath1-theory by the inclusion @xmath636 where * @xmath555 is the characteristic function of a compact - open subset @xmath554 of @xmath516 ; * @xmath615 is the characteristic function of @xmath616 ; * @xmath618 is the rank one projector associated to a norm @xmath120 function @xmath188 of @xmath619)$ ] .
we have an isomorphism @xmath637 induced on the coinvariants by the composition @xmath638}{\to } k_1(c(z_w)\otimes c(\omega)\rtimes{\mathbb z})\cong k_1(c_r^*({\mathcal{g}},\lambda)\to k_1(c_r^*({\mathcal{g}},\lambda)\rtimes_{\tilde{\alpha_w}}{\mathbb z}),\]]where * @xmath639 $ ] is the external product in @xmath1-theory by the class in @xmath570 of the unitary @xmath418 of @xmath397 corresponding to the positive generator of @xmath73 ; * the last map in the composition is the homomorphism induced in @xmath1-theory by the inclusion @xmath640 . the short exact sequence of theorem [ thm - k0-gl ] , admits an explicit splitting which can be described in the following way : assume first that @xmath641 is minimal . in particular , @xmath642 is generated by @xmath643 .
let us considerer the following diagram , whose left square is commutative @xmath644 where * the horizontal maps of the left square are induced by the inclusion @xmath645 .
* vertical maps are the thom - connes isomorphisms . *
the map @xmath646 is induced by the continuous map @xmath647 $ ] ; up to the morita equivalence between @xmath648 and @xmath649 , the right down staircase is the boundary of the pimsner - voiculescu six - term exact sequence that computes @xmath650 . from this , we see that @xmath651 is generated by the image of the generator @xmath652 of @xmath653 corresponding under the canonical identification @xmath654 to the class of any rank one projector in some @xmath655 .
on the other hand , we have a diagram with commutative squares where , * the horizontal maps of the left square are induced by the inclusion + @xmath656 * the horizontal maps of the middle square are induced by the isomorphism of lemma [ lem - psi ] * the horizontal maps of the right square are induced by the isomorphism of proposition [ prop - iso ] * the first row of vertical maps are thom - connes isomorphisms .
it is then straightforward to check that the down staircase of the diagram is indeed induced by the inclusion @xmath657 .
notice that @xmath658 ( by thom - connes isomorphism ) .
moreover , under the inclusion @xmath659 , any rang one projector @xmath660 of @xmath661 provides a generator for @xmath662 whose image under the left vertical map is the generator @xmath652 for @xmath663 . using the description of the boundary map of pimsner - voiculescu six - term exact sequence , we see that @xmath660 , viewed as an element of @xmath664 whose class in @xmath1-theory provides a lift for @xmath643 in the short exact sequence of theorem [ thm - k0-gl ] . in general
, @xmath665 is generated by characteristic functions of @xmath73-invariant compact - open subsets of @xmath516 . according to proposition [ caracterisation ] , any @xmath73-invariant compact - open subset @xmath554 of @xmath516 provides a @xmath7-invariant compact subset @xmath666 of @xmath667 .
hence , with above notations , if @xmath668 is the characteristic function for @xmath666 , then @xmath669 can be viewed as an element of @xmath664 .
let @xmath670 be the group homomorphism uniquelly defined by @xmath671 for @xmath554 a @xmath73-invariant compact - open subset of @xmath516 .
then @xmath672 is a section for the short exact sequence of theorem [ thm - k0-gl ] .
it is known that for euclidian tilings , topological invariants of the continuous hull are closely related to the @xmath1-theory of the @xmath0-algebra associated to the tiling .
the @xmath1-theory of the latter turn out to be isomorphic to the @xmath1-theory of the hull which is using the chern character rationally isomorphic to the integral cch cohomology .
moreover , in dimension less or equal to @xmath2 , the chern character can be defined valued in integral cohomology and we eventually obtain an isomorphism between the integral cch cohomology of the hull and the @xmath1-theory of the @xmath0-algebra associated to the tiling . in consequence of this fact ,
a lot of interest has been generated in the computation of topological invariants of the hull .
for penrose hyperbolic tilings , since the group of affine isometries of the hyperbolic half - plane is isomorphic to a semi - direct product @xmath673 , we get using the thom - connes isomorphism that @xmath674 moreover , since the cohomological dimension of @xmath495 is @xmath10 , the chern character can also be defined with values in integral cch cohomology and hence we get as in the euclidian case of low dimension an isomorphism @xmath675 these topological invariants can be indeed computed directly using technics very closed to those used in section [ section_k_th ] . indeed for a @xmath0-algebra @xmath344 provided with an automorphism @xmath345 , there is natural isomorphisms @xmath676 called the mapping torus isomorphisms , where @xmath347 is the mapping torus algebra constructed at the end of section [ subsec - suspension ] .
recall from proposition [ caracterisation ] that @xmath495 can be viewed as a double suspension @xmath677 with @xmath678,\omega')\mapsto ( [ 2x,2t],\sigma(\omega'))$ ] . in regard of the mapping torus isomorphism
, the double crossed product by @xmath73 corresponds in @xmath1-theory to the double suspension structure on @xmath495 . for people interested in topological invariants ,
we explain how a straight computation can be carried out . for a @xmath0-algebra @xmath344 provided with an automorphism @xmath345 , the short exact sequence @xmath679 where @xmath680 is the evaluation at @xmath681 of elements of @xmath682,a)$ ] ,
gives rise to short exact sequences @xmath683 and @xmath684 where invariants and coinvariants are taken with respect to the action induced by @xmath345 on @xmath685 ( see section [ section_k_th ] ) .
in particular , if @xmath4 is a compact set and @xmath686 is a homeomorphism , and with the notations of section [ subsec - hull ] the mapping torus of @xmath687 with respect to the automorphism induced by @xmath188 is @xmath688 .
we deduce short exact sequences @xmath689 and @xmath690 similarly , we have short exact sequences in cch cohomology @xmath691 derived from the inclusion @xmath692 since the space @xmath495 has a structure of double suspension , we see following the same route as in section [ section_k_th ] that @xmath495 only has cohomology in degree @xmath693 and @xmath10 and we get isomorphisms @xmath694)\\ \label{equ - k1-hull } k^1(x^g_{\mathcal{p}(w)})&\cong & { \operatorname{coinv}}c(z_w,{\mathbb z})\\ \label{equ - h0-hull } \check{h}^0(x^g_{\mathcal{p}(w)},{\mathbb z})&\cong & { \operatorname{inv}}c(z_w,{\mathbb z})\\ \label{equ - h1-hull } \check{h}^1(x^g_{\mathcal{p}(w)},{\mathbb z})&\cong & { \operatorname{coinv}}c(z_w,{\mathbb z})\\ \label{equ - h2-hull } \check{h}^2(x^g_{\mathcal{p}(w)},{\mathbb z})&\cong & { \operatorname{coinv}}c(z_w,{\mathbb z}[1/2]).\end{aligned}\ ] ] recall that invariants and coinvariants of @xmath629 are taken with respect to the automorphism @xmath695 and that coinvariants of @xmath591)$ ] are taken with respect to the automorphism @xmath696)\to
c(z_w,{\mathbb z}[1/2]);\ , f\mapsto 2f\circ{\sigma}^{-1}.\ ] ] let us describe explicitly these isomorphisms .
the identification of equation ( [ equ - double - suspension ] ) yields to a continuous map @xmath697induced by the equivariant projection @xmath698 .
together with the inclusion @xmath699 , this gives rise to a homomorphism @xmath700 inducing under bott peridodicity the isomorphism of equation ( [ equ - k1-hull ] ) ( recall from lemma [ lem - cantor ] that @xmath701 ) .
the identification of equation ( [ equ - h1-hull ] ) is obtained in the same way by using the isomorphism @xmath702provided by the cup product by the fundamental class of @xmath703 .
recall that @xmath665 is generated by characteristic functions of invariant compact - open subsets of @xmath487 .
if @xmath554 is such a subset , then @xmath704 is a compact - open subset of @xmath705 and is pulled - back under the map of equation ( [ equ_proj ] ) to a compact - open subset @xmath666 of @xmath495 .
the isomorphism of equation ( [ equ - h0-hull ] ) identifies @xmath706 with the class of @xmath707 in @xmath708 and the isomorphism of equation ( [ equ - k0-hull ] ) identifies @xmath555 with the class of @xmath668 in @xmath709 .
using twice the inclusion of equation ( [ equ - inclusion ] ) for the double suspension structure of @xmath495 , we obtain an inclusion @xmath710 and hence by bott periodicity a map @xmath711 .
then , if @xmath554 is a compact - open subset of @xmath487 and @xmath576 is an integer , the image of @xmath712 under this map is up to identification of equation ( [ equ - k0-hull ] ) the class of @xmath713 in @xmath714)$ ] .
the description of the identification of equation ( [ equ - h2-hull ] ) is obtained in the same way by using the isomorphism @xmath715 provided by the cup product by the fundamental class of @xmath716 . moreover , since the chern character is natural and intertwins bott periodicity and the cup product by the fundamental class of @xmath717 , we deduce that up to the identifications of equations ( [ equ - k0-hull ] ) to ( [ equ - h2-hull ] ) , it is given by the identity maps of @xmath718)$ ] and of @xmath719 .
it is easy to guess how the generators of @xmath720 described in equations ( [ equ - k0-hull ] ) and ( [ equ - k1-hull ] ) should be identified with those of @xmath721 described in section [ section_k_th ] under thom - connes isomorphism of equation ( [ equ - mt - iso ] ) .
recall first that for a unital @xmath0-algebra @xmath344 provided with an automorphism @xmath345 , * the mapping torus @xmath347 is provided with an action @xmath351 of @xmath7 by automorphisms ( see section [ subsec - suspension ] ) and moreover @xmath352 and @xmath722 are morita equivalent ; * the mapping torus isomorphisms are the composition of the thom - connes isomorphisms @xmath723 and @xmath724 with the isomorphism @xmath725 induced with the above morita - equivalence ; it is then straightforward to check that viewing @xmath726 as a double crossed product by @xmath73 as we did in section [ sec - c*-tiling ] , the thom - connes isomorphism @xmath727 is obtained by using twice the mapping torus isomorphim ( up to stabilisation for the second one ) . in view of our purpose of idenfying the generators of @xmath728 with those of @xmath721
, we will need the following alternative description of the mapping torus isomorphism using the bivariant kasparov @xmath1-theory groups @xmath729 @xcite .
let @xmath344 be a @xmath0-algebra and let @xmath345 be an automorphism of @xmath344 .
since the action of @xmath73 on @xmath7 by translations is free and proper , we have a morita equivalence between @xmath347 and @xmath730 , where @xmath731 is equipped with the diagonal action of @xmath73 .
recall that the @xmath73-equivariant unbounded operator @xmath732 of @xmath733 gives rise to a unbounded @xmath1-cycle and hence to an element @xmath503 in @xmath734 .
then the mapping torus isomorphism of equation ( [ equ - mt - iso ] ) is the composition @xmath735 where , * the first map comes from the morita equivalence ; * @xmath736 is the kasparov transformation in bivariant @xmath737-theory ; * for any @xmath0-algebra @xmath738 equipped with an action of @xmath20 by automorphism @xmath739 is the tensorisation operation ; * @xmath740 stands for the right kasparov product by @xmath741 .
then the identification between the generators of @xmath720 and of @xmath742 can be achieved using the next two lemmas .
[ lem - inv]let @xmath344 be a unital @xmath0-algebra together with an automorphism @xmath345 .
let @xmath660 be invariant projector in @xmath344 and let @xmath743 be the class in @xmath744 of the projector @xmath367\to a;\ , t\mapsto e$ ] .
then the image of @xmath743 under the mapping torus isomorphism is equal to the class of the unitary @xmath745 of @xmath352 in @xmath746 ( here @xmath418 is the unitary of @xmath352 corresponding to the positive generator of @xmath73 ) ; the invariant projector @xmath660 gives rise to an equivariant map @xmath747 and hence to a homomorphism @xmath748 .
by naturality of the mapping torus , this amounts to prove the result for @xmath749 which is done in ( * ? ? ?
* example 6.1.6 ) .
let @xmath344 be a unital @xmath0-algebra together with an automorphism @xmath345 and let @xmath59 be an element in @xmath685 .
the two following elements then coincide : * the image of @xmath59 under the composition @xmath750 where * * the first map is the bott periodicity isomorphism ; * * the second map is induced by the inclusion @xmath751 ; * * the third map is the mapping torus isomorphism . *
the image of @xmath59 under the map @xmath752 induced by the inclusion @xmath753 .
let us first describe the imprimity bimodule implementing the morita equivalence between @xmath347 and @xmath730 .
indeed , in a more general setting , if * @xmath4 is a locally compact space equipped with a proper action of @xmath73 by homeomorphisms , * @xmath738 is a @xmath0-algebra provided with an action of @xmath73 by automorphisms ; * @xmath754 stands for the algebra of equivariant continuous maps @xmath755 such that @xmath756 belongs to @xmath757 .
then , if we equip @xmath758 with the diagonal action of @xmath73 , there is an imprimitivity @xmath759-bimodule defined in the following way : let us consider on @xmath760 the @xmath761-valued inner product @xmath762 for @xmath401 and @xmath402 in @xmath760 and @xmath576 in @xmath73 .
this inner product is namely positive and gives rise to a right @xmath761-hilbert module @xmath763 , the action of @xmath761 on the right being given for @xmath401 in @xmath760 and @xmath236 in @xmath764 by @xmath765 the action by pointwise multiplication of @xmath766 on @xmath767 extends to a left @xmath754-module structure on @xmath763 .
let us denote by @xmath768 $ ] the class of the @xmath759-bimodule @xmath763 in @xmath769 .
it is straightforward to check that * @xmath768 $ ] is natural in both variable , in particular , if @xmath770 is an open invariant subset of @xmath4 and let us denote by @xmath771 and @xmath772 the homomorphisms induced by the inclusion @xmath773 and respectively by @xmath774 $ ] and @xmath775 $ ] the corresponding classes in @xmath776 and @xmath777 , then @xmath778\otimes_{b_x^{\mathbb z}}[\ce(b , x)]=[\ce(b , y)]\otimes_{c_0(y , b)}j_{\mathbb z}([\iota_{y , x , b}]).\ ] ] * up to the identification @xmath779 , the class @xmath780 $ ] is induced by the composition @xmath781 where the first map is @xmath782 . *
if @xmath66 is any locally compact space , and if we consider @xmath73 acting trivially on it , then up to the identifications @xmath783 and @xmath784 , we have @xmath785=\tau_{c_0(v)}([\ce(b , x)])$ ] . noticing that for a @xmath0-algebra @xmath344 provided with an automorphism @xmath345 , we have a natural identification @xmath786 ,
the mapping torus isomorphism of equation ( [ equ - mt - iso ] ) is obtained by right kasporov product with @xmath787\otimes_{c_0({\mathbb r},a)\rtimes{\mathbb z}}j_{\mathbb z}(\tau_a([y]))$ ] . from this , we see that the composition in the statement of the lemma is given by right kasparov product with @xmath788)\otimes_{c_0((0,1),a)}[\iota^{\mathbb z}_{(0,1)\times{\mathbb z},{\mathbb r},a}]\otimes_{a_\beta}[\ce(a , { \mathbb r})]\otimes_{c_0({\mathbb r},a)\rtimes{\mathbb z}}j_{\mathbb z}(\tau_a(y))\ ] ] where * @xmath789 $ ] in @xmath790 is the boundary of the evaluation at @xmath681 extension @xmath791 * we have used the identification @xmath792 to see @xmath793 as an invariant open subset of @xmath7 , in particular we have @xmath794 .
according to the naturality properties of @xmath795 $ ] listed above , we get that @xmath796\otimes_{a_\beta}[\ce({\mathbb r},a)]&=&[\ce((0,1)\times{\mathbb z},a)]\otimes_{c_0({\mathbb z},a)\rtimes{\mathbb z}\otimes c_0(0,1)}j_{\mathbb z}([\iota_{(0,1)\times{\mathbb z},{\mathbb r},a}])\\ & = & \tau_{c_0(0,1)}([\ce({\mathbb z},a)])\otimes_{c_0({\mathbb z},a)\rtimes{\mathbb z}\otimes c_0(0,1)}j_{\mathbb z}([\iota_{(0,1)\times{\mathbb z},{\mathbb r},a}])\end{aligned}\ ] ] using commutativity of exterior kasparov product , we get from equation ( [ equ - kasp - prod ] ) that @xmath797\otimes_{c_0({\mathbb z},a)\rtimes{\mathbb z}}j_{\mathbb z}(\tau_{c_0({\mathbb z},a)}([\partial])\otimes_{c_0((0,1)\times{\mathbb z},a)}[\iota_{(0,1)\times{\mathbb z},{\mathbb r},a}]\otimes \tau_a(y)).\ ] ] let @xmath798 be the element of @xmath799 corresponding in the unbounded picture to the unbounded operator @xmath732 on @xmath800 and let @xmath801 $ ] be the element of @xmath802 corresponding to the equivariant representation by compact operator of @xmath803 onto @xmath804 ( equipped with the left regular representation ) given by pointwise multiplication .
then it is straightforward to check that @xmath805\otimes_{c_0({\mathbb r},a)}\tau_a(y)= \tau_a(\tau_{c_0({\mathbb z})}(y')\otimes_{c_0({\mathbb z})}[\mathcal{f}]).\]]but it is a standard fact that @xmath789\otimes_{c_0(0,1)}y'=1 $ ] in the ring @xmath806 and hence we eventually get that @xmath807\otimes_{c_0({\mathbb z},a)\rtimes{\mathbb z}}j_{\mathbb z}(\tau_a([\mathcal{f}])).\ ] ] a direct inspection of the right hand side of this equality shows that @xmath808 is indeed the class of @xmath809 induced by the inclusion @xmath810 recall that in section [ section_k_th ] , we have established isomorphisms @xmath811)\oplus { \operatorname{inv}}c(z_w,{\mathbb z})\ ] ] and @xmath812 under this identification , and using lemma [ lem - inv ] and twice lemma [ lem - coinv ] , we are now in position to describe the image of the generators of @xmath720 under the double thom - connes isomorphism . under the identification of equations ( [ equ - k0-hull ] ) , ( [ equ - k1-hull ] ) , ( [ equ - id - k0 ] ) and ( [ equ - id - k1 ] ) , the double thom - connes isomorphism @xmath813 corresponds to the identity maps of @xmath714)\oplus { \operatorname{inv}}c(z_w,{\mathbb z})$ ] and @xmath719 .
the statement concerning the factor @xmath665 is indeed a consequence of the discussion at the end of section [ section_k_th ] . before proving the statements concerning the factors @xmath719 and @xmath714)$ ] ,
we sum up for convenience of the reader the main features , described in sections [ exemple ] and [ sec - c*-tiling ] of the dynamic of the continuous hull for the coloured and the uncoloured penrose hyperbolic tilings . *
the closure @xmath389 of @xmath814 in @xmath667 for the tiling topolology of the penrose hyperbolic tiling @xmath815 is homeomorphic to the suspension of the homeomorphism @xmath816 ; * the continuous hull @xmath495 of the coloured penrose hyperbolic tiling @xmath817 is homeomorphic to the suspension of the homeomorphism @xmath818 ; * if we provide @xmath819 with the diagonal action of @xmath7 , by translations on @xmath389 and trivial on @xmath487 , and equip the groupoid @xmath820 with the haar system arising from the haar measure of @xmath7 , then @xmath821 is the mapping torus of @xmath822 with respect to the automorphim @xmath823 arising from the automorphism of groupoid @xmath824 ( see section [ sec - c*-tiling ] ) ; * @xmath822 is morita - equivalent to the crossed product @xmath825 for the action of @xmath73 on @xmath826 arising from @xmath827 ( see section [ sec - penrose - uncolor ] ) . let us consider the following diagram : @xmath828 where * the bottom and the right middle horizontal arrows are the maps defined for any @xmath0-algebra @xmath344 provided by an automorphism @xmath345 as the composition @xmath829 of the bott peridocity isomorphism with the homomorphism induced in @xmath1-theory by the inclusion @xmath830 ; * the left middle horizontal arrow is up to the morita equivalence between @xmath822 and @xmath825 induced by the inclusion @xmath831 ; * the top horizontal arrow is up to the morita equivalence between @xmath832 and @xmath664 is induced by the inclusion @xmath833 . * the vertical maps @xmath834
stand for the thom - connes isomorphisms .
then the inclusion @xmath714)\hookrightarrow k^0(x^g_{\mathcal{p}(w)})$ ] of equation ( [ equ - k0-hull ] ) is induced by the composition of the bottom arrows and the inclusion @xmath714)\hookrightarrow k_0(c(x^g_{\mathcal{p}(w)})\rtimes g)$ ] of equation ( [ equ - id - k0 ] ) is induced by the upper staircase . according to lemma [ lem - inv ] , the left bottom and the right top squares are commutative . hence ,
the proof of the statements regarding to the @xmath714)$ ] summand amounts to show that the bottom right square is commutative . to see this , let us equip @xmath835 with the action of @xmath7 by homeomorphisms @xmath836 if we restrict this action to @xmath837 , then the inclusion @xmath838 is @xmath7-equivariant and the bott periodicity isomorphism is the boundary of the equivariant short exact sequence @xmath839 provided by evaluation at @xmath681 .
this equivariant short exact sequence gives rise to a short exact sequence for crossed products @xmath840 and since the thom - connes isomorphism is natural , it intertwins the corresponding boundary maps and hence we get a commutative diagram @xmath841 where the left horizontal arrows are induced by the boundary maps corresponding to the exact sequences of equation ( [ equ - bott ] ) and ( [ equ - bott - equ ] ) and the right horizontal arrows are induced by the equivariant inclusion @xmath838 .
let us consider the family groupoids @xmath842 and @xmath843 .
notice that if @xmath494 is viewed as the suspension of the groupoid @xmath243 , then @xmath842 is the restriction of @xmath494 to a fundamental domain .
the reduced @xmath0-algebras of these two groupoids are respectively @xmath844 and @xmath845 and the automorphism of groupoids @xmath846 gives rise to a commuting diagram using naturality of the boundary map , we see that the composition of the top horizontal arrows in diagram ( [ diag - bott ] ) is the composition @xmath847 of the bott peridocity isomorphism with the homomorphism induced in @xmath1-theory by the inclusion @xmath848 . this concludes the proof for the statement concerning the summand @xmath849)$ ] . the statement concerning the summand @xmath850 is a consequence of the commutativity of the top square of diagram [ diag - tc ] and of lemma [ lem - inv ] applied to the middle bottom vertical arrow
recall that according to the discussion ending section [ subsec - ergotic ] , a probability is harmonic if and only if it is @xmath5-invariant . in this section ,
we associate to a harmonic probability a @xmath2-cyclic cocycle on the smooth crossed product algebra of @xmath851 .
this cyclic cocycle is indeed builded from a @xmath120-cyclic cocycle on the algebra of smooth ( along the leaves ) functions on @xmath852 by using the analogue in cyclic cohomology of the thom - connes isomorphism ( see @xcite ) .
we give a description of this cocycle and we discuss an odd version of the gap - labelling .
we collect here results from @xcite concerning smooth crossed products by an action of @xmath7 that we will need later on .
let @xmath853 be a frechet algebra with respect to an increasing family of semi - norms @xmath854 .
a smooth action on @xmath853 is a homomorphism @xmath855 such that 1 . for every @xmath116 in @xmath7 and @xmath14 in @xmath853 ,
the function @xmath856 is smooth .
2 . for every integers @xmath471 and @xmath203 , there exist integers @xmath857 and @xmath576 and a real @xmath858 such that @xmath859 for every @xmath14 in @xmath853 .
if @xmath268 is a smooth action on @xmath853 , then the smooth crossed product @xmath860 is defined as the set of smooth functions @xmath861 such that @xmath862 for all integers @xmath863 and @xmath576 .
the smooth crossed product @xmath860 provided with the family of semi - norm @xmath864 for @xmath863 and @xmath576 integers together with the convolution product @xmath865 is then a frechet algebra .
notice that a smooth action @xmath866 on a frechet algebra @xmath853 gives rise to a bounded derivation @xmath867 of @xmath860 defined by @xmath868 for all @xmath188 in @xmath869 and @xmath116 in @xmath7 .
let @xmath870 be the algebra of continuous and smooth along the leaves functions on @xmath852 , i.e functions whose restrictions to leaves admit at all order differential which are continuous as functions on @xmath852 .
let @xmath871 and @xmath872 be the two actions of @xmath7 on @xmath870 respectively induced by @xmath873 and @xmath874 let @xmath4 and @xmath770 be respectively the vector fields associated to @xmath871 and @xmath872 .
then @xmath870 is a frechet algebra with respect to the family of semi - norms @xmath875 where @xmath471 and @xmath876 run through integers .
it is clear that @xmath871 is a smooth action on @xmath870 .
moreover , @xmath877\ ] ] is an action of @xmath7 on @xmath878 by automorphisms .
this action is not smooth in the previous sense .
nevertheless , the action @xmath872 satisfies conditions ( 1),(2 ) and ( 3 ) of ( * ? ? ?
* section 7.2 ) with respect to the family of functions @xmath879 , where @xmath576 runs through integers .
in this situation , we can define the smooth crossed product @xmath880 of @xmath878 by @xmath872 to be the set of smooth functions + @xmath881 such that @xmath882 for all integers @xmath883 and @xmath203 ( we have reindexed for convenience the family of semi - norms on @xmath878 using integers ) . then @xmath880 provided with the family of semi - norms @xmath884 for @xmath883 and @xmath203 integers together with the convolution product is a frechet algebra .
moreover , this algebra can be viewed as a dense subalgebra of @xmath885 . as for smooth actions ,
the action @xmath872 gives rise to a derivation @xmath886 of @xmath880 ( defined by the same formula ) .
let @xmath426 be a @xmath5-invariant probability on @xmath852 .
define @xmath887using the leibnitz rules and the invariance of @xmath5 , it is straightforward to check that @xmath888 is @xmath120-cyclic cocycle . in @xcite
was constructed for a smooth action @xmath866 on a frechet algebra @xmath853 a homomorphism @xmath889 , where @xmath890 stands for the cyclic cohomology .
this homomorphism is indeed induced by a homomorphism at the level of cyclic cocycles @xmath891 and commutes with the periodisation operator @xmath892 .
hence it gives rise to a homomorphism in periodic cohomology @xmath893 which turns out to be an isomorphism .
this isomorphism is for periodic cohomology the analogue of the thom - connes isomorphism in @xmath1-theory .
we give now the description of @xmath894 .
let us define first @xmath895 and @xmath896 respectively by @xmath897 and @xmath898 , for all @xmath188 in @xmath878 and @xmath116 in @xmath7 . using the relation @xmath899 and applying the definition of @xmath900 ( see @xcite ) , we get : for any elements @xmath188 , @xmath415 and @xmath236 in @xmath901 , we have 1 .
@xmath902 2 .
@xmath903 for all @xmath116 in @xmath7 , i.e the cocycle @xmath894 is @xmath872-invariant .
according to ( * ? ? ?
* section 7.2 ) , the action @xmath872 on @xmath901 also gives rise to a homomorphism @xmath904 which induces an isomorphim @xmath905 .
a direct application of the definition of @xmath906 leads to let @xmath97 be a @xmath907-invariant @xmath2-cyclic cocycle for @xmath878 .
let us define for any @xmath908 and @xmath236 in @xmath880 .
@xmath909 then @xmath910 with above notations , the @xmath2-cyclic cocycle on @xmath880 associated to the penrose hyperbolic tiling coloured by @xmath70 and to a @xmath5-invariant probability @xmath426 on @xmath852 is @xmath911 notice that if we carry out this construction for a tiling @xmath912 of the euclidian space with continuous hull @xmath913 with respect to the @xmath914-action by translations , we get taking twice the crossed product by @xmath7 a @xmath2-cyclic cocycle which is indeed equivalent ( via the periodisation operator ) to the @xmath120-cycle cocycle on @xmath915 arising from the trace on @xmath916 associated to an @xmath7-invariant probability on @xmath913 . the class @xmath917 $ ] of @xmath918 in @xmath919 is the image of the class of @xmath888 under the composition of isomorphism @xmath920 since pairing with periodic cohomology provides linear forms for @xmath1-theory groups , the @xmath2-cyclic cocycle @xmath918 provides a linear map @xmath921,x \rangle.\ ] ] the main issue in computing @xmath922 is that the thom - connes isomorphism a priori may not hold for @xmath923 . if it were the case , the inclusion @xmath924 would induces an isomorphism @xmath925 and from this we could get that @xmath926\stackrel{\text{def}}{=\!=}\{\hat{\eta}(e),\,e\text { compact - open subset of } z_w\},\ ] ] where @xmath927 is the probability on @xmath516 of proposition [ carmesureinv ] in one - to - one correspondance with @xmath426 .
since @xmath928 $ ] is indeed the one dimension gap - labelling for the subshift corresponding to @xmath70 , this would be viewed as an odd version of the gap labelling .
nevertheless , the right setting to state this generalisation of the gap - labelling seems to be the frechet algebra and a natural question is whether we have @xmath929,x\rangle;\,x\in k_1(\afx\rt_{\be^0}{\mathbb r}\rtimes_{\be^1}^\rho{\mathbb r } ) \}={\mathbb z}[\hat{\eta}]\ ] ] or if the pairing bring in new invariants .
this paper was partially written during a visit of the first author at the pacific institut for mathematical sciences in victoria , canada .
he is also indebted to j. renault for some very helpful discussions on haar systems for groupoids . .
_ tilings , @xmath0-algebras and @xmath1-theory _ , directions in mathematical quasicrystals , crm monograph series * 13 * ( 2000 ) , 177 - 206 , m.p .
baake & r.v .
moody eds . , ams providence . .
_ aperiodic tiling of the hyperbolic plane by convex polygons _ , israel journ .
* 107 * ( 1998 ) , 319 - 325 .
_ on invariant measures of finite affine type tilings _ , ergod .
th . & dyn .
( 2006 ) * 26 * , 1159 - 1176 .
_ pentaplexity _ , mathematical intelligencer * 2 * ( 1979 ) , 32 - 37 , ann . of math .
( 2)*102 * ( 1975 ) , 327361 . _ the c*-algebras associated with minimal homeomorphisms of the cantor set _ , pacific j. math .
* 136 * ( 1989 ) , 329 - 352 . ,
_ exact sequences for @xmath1-groups and ext - groups of certain cross - product @xmath931-algebras _ , j. operator theory 4 ( 1980 ) , n^o^. 1 , p. 93118 . .
_ substitution dynamical systems spectral analysis _ , lnm , 1294 .
springer - verlag , berlin , 1987 , _ a groupoid approach to @xmath932-algebras _ , lecture notes in mathematics , 793 , springer , berlin . .
_ symbolic dynamics and tilings of @xmath6 _ , symbolic dynamics and its applications , proc .
, * 60 * , amer . math .
soc . , providence , ri , 2004 , p. 81119 . .
_ cycles for the dynamical study of foliated manifolds and complex manifolds _ , invent . math .
* 36 * ( 1976 ) , 225 - 255 . .
_ introduction to the baum - connes conjecture _ , lectures in mathematics eth zrich , from notes taken by indira chatterji , with an appendix by guido mislin , birkhuser verlag , basel , 2002 . | penrose hyperbolic tilings are tilings of the hyperbolic plane which admit , up to affine transformations a finite number of prototiles . in this paper
, we give a complete description of the @xmath0-algebras and of the @xmath1-theory for such tilings .
since the continuous hull of these tilings have no transversally invariant measure , these @xmath0-algebras are traceless . nevertheless , harmonic currents give rise to @xmath2-cyclic cocycles and we discuss in this setting a higher - order version of the gap - labelling . _
* keywords : * hyperbolic aperiodic tilings , @xmath0-algebras of dynamical systems , k - theory , cyclic cohomology _ _ * 2000 mathematics subject classification : * 37a55 , 37b051 , 46l55 , 46l80 _ _ * subject classification : * noncommutative topology and geometry , dynamical systems _ |
in the development of our understanding of the hadron production mechanism , the inclusive one - particle production by electron - positron annihilation plays the same role as the deeply inelastic scattering does in our learning of the nucleon structures .
experimentally , a number of electron - positron colliders have been operated and will be constructed at various center of mass energies . on the theoretical side ,
a series of parton fragmentation functions have been identified by jaffe and ji @xcite , which characterize the information about the inclusive hadron production .
therefore , it is worthwhile to explore the possibility to measure them , expecially those spin - dependent ones , by the inclusive one - particle production in electron - positron annihilation . to study spin physics related to the inclusive spin - half baryon production , it is usually needed to monitor the polarization of the produced particles .
if we are restricted with the inclusive hyperon production , it is quite feasible to measure the polarization of the produced particles .
in fact , a series of experiments have already been done at fermilab measuring the polarization of the inclusive hyperons in fixed - target experiments @xcite .
we also note that some preliminary results @xcite on the measurement of the @xmath8 polarization at cern lep have been existent as well . because what can be accessed directly by experiments are structure functions , we explore in this paper the relation between @xmath0 and @xmath1 , two spin - dependent structure functions for the inclusive spin - half baryon production by electron - positron annihilation , and the quark fragmentation functions defined by jaffe and ji . without loss of any generalities
, we consider the inclusive @xmath8 hyperon production for among the hyperons the @xmath8 particle is of the largest production cross section and its polarization is the easiest to measure .
the advantages of measuring the polarization of the inclusive @xmath8 hyperon are also realized by several groups of authors in different cases . among others ,
burkardt and jaffe @xcite investigated the possibility to measure the @xmath9 fragmentation functions at lep , and chen _ et . al . _
@xcite discussed the more complicated semi - inclusive process @xmath10 . for simplicity
, we work in the energy region not too high so that only the photon channel needs to be considered . then , all the information about the hyperon production is entailed by the photon fragmentation tensor , which is defined as @xmath11 where @xmath12 represents the summation over all the possible final states that contain the inclusive @xmath8 hyperon . throughout the work ,
we normalize the spin vector in such a way that @xmath13 for a pure state .
the electromagnetic current is defined as @xmath14 , with @xmath15 being the quark flavor index and @xmath16 being the electric charge of the quark in unit of the electron charge . in our presentation
, we will suppress the flavor index whenever possible .
subjected to the gauge invariance , hermiticity and parity conservation , @xmath17 assumes the following general lorentz decomposition @xcite : @xmath18 \nonumber \\ & & + i m \varepsilon_{\mu\nu\lambda\sigma}q^\lambda s^\sigma \frac{\hat g_1(z_b , q^2)}{p\cdot q } + i m \varepsilon_{\mu\nu\lambda\sigma}q^\lambda ( s^\sigma -\frac{s\cdot q}{p\cdot q } p^\sigma ) \frac{\hat g_2(z_b , q^2 ) } { p\cdot q } \nonumber \\ & & + m \left[(p_\mu-\displaystyle\frac{p\cdot q}{q^2 } q_\mu)\varepsilon_{\nu \rho\tau\eta}p^\rho q^\tau s^\eta + ( p_\nu-\displaystyle\frac{p\cdot q}{q^2}q_\nu ) \varepsilon_{\mu
\rho\tau\eta}p^\rho q^\tau s^\eta \right ] \frac{\tilde f ( z_b , q^2)}{(p\cdot q)^2},\end{aligned}\ ] ] where @xmath19 , @xmath20 , @xmath21 is the mass of the @xmath8 hyperon , and @xmath22 , @xmath23 , @xmath0 , @xmath1 , and @xmath24 are the scaling structure functions .
it should be noted that due to the final - state interactions , the time reversal invariance can not exert any constraints on the lorentz decomposition of @xmath25 .
obviously , there exists one - to - one correspondence between @xmath22 , @xmath23 , @xmath0 , and @xmath1 , on the one hand , and four structure functions @xmath26 ,
@xmath27 , @xmath28 and @xmath29 in the deeply inelastic scattering , on the other hand . @xmath22 and
@xmath23 have been vastly discussed in the literature @xcite , so we will simply ignore them in our discussion .
as for the structure function @xmath24 , which arises from the final - state interaction , we will not address it , either , to avoid dispersing the reader s sight
. our attention will be focused on @xmath0 and @xmath1 so that our results can be compared with their counterparts in the deeply inelastic scattering case @xcite .
we establish our coordinate system by letting the @xmath30 axis be along the outgoing direction of the inclusive hyperon and the @xmath31@xmath30 plane in the production plane .
we adopt the light - cone coordinates and parameterize the @xmath8 momentum as @xmath32 where @xmath33 @xmath34 with @xmath35 being the magnitude of the @xmath8 momentum .
we will work in the frame in which @xmath35 has a large value
. obviously , @xmath36 and @xmath37 are light - like and they satisfy @xmath38 .
we feel it desirable to present a naive parton model approach to @xmath0 and @xmath1 , since the corresponding results can supply us with a benchmark for comparison . in the naive parton model
, the fragmented hadrons will be completely collinear to their parent quark parton , with no transverse momentum effects .
correspondingly , we need to assign a mass @xmath39 to the quark of flavor @xmath15 , where @xmath40 is the momentum fraction of the quark carried by the inclusive @xmath8 particle .
@xmath41 can be obtained from the convolution of the underlying tensor @xmath42 for photon fragmentation into a quark - antiquark pair with the @xmath8 multiplicity distribution @xmath43 in the jet initiated by the quark of flavor @xmath15 and spin four - vector @xmath44 : @xmath45 by analogy with the well - known leptonic tensor , we know @xmath46.\ ] ] therefore , we have the following results for @xmath0 and @xmath1 : @xmath47 @xmath48 where @xmath49 is the @xmath8 number density difference with spin parallel and antiparallel to the quark spin in the @xmath15-flavored quark jet . in the naive parton model ,
if one boosts into the quark rest frame , all its fragmenting products will be at rest , too .
hence , @xmath50 does not depend on the spin quantization direction @xcite .
in the qcd field theory approach , the photon fragmentation tensor is composed of the contributions of an infinite series of the so - called cut feynman diagrams .
the cut diagram is formed by piecing together the feynman diagram of one amplitude with the complex conjugate of another one with the same initial and final states . owing to lack of the methods to treat nonperturbative interactions , one usually adopt the strategy to factorize @xcite the hadron - involved process into short- and long - distance parts .
the former can be calculated perturbatively whereas the latter can be measured experimentally .
in such a factorization approach , qcd long - distance interactions are conventionally represented by blobs in the cut diagram . in general , these cut diagrams begin to contribute at the different powers of the hard scattering scale . if a physical quantity goes like @xmath51 or begins to make contributions at @xmath52 , it is referred to as being twist-@xmath53 . according to qiu and sterman @xcite
, the twist at which a cut diagram begins to contribute is controlled by the number of @xmath54 parton lines that attach to the long - distance interaction blobs .
now it is well known that some potentially power - suppressed twist - three spin asymmetries are relatively large , so in this paper we will work up to the twist - three level , i.e. , to the first nonleading power .
in addition , we work in the light - cone gauge specified by @xmath55 .
correspondingly , the cut diagrams shown in figs . 1 , and 2 consist of our subjects .
we begin with the case in which the inclusive @xmath8 is of plus helicity . in this case
, the leading spin dependence of @xmath41 comes about at twist two , while its next - to - leading spin dependence at twist four . to protrude the main physics
, we work at leading twist and accordingly only the lowest - order diagram shown in fig .
1 gets into work . as a result , @xmath56 , \label{t2}\ ] ] where @xmath57 and @xmath58 stands for making traces in the dirac and color spaces , and @xmath59 and @xmath60 for the dirac indices .
the color summation is implicit in our parton fragmentation matrices , so there is a color factor @xmath61 ( @xmath62 ) on the right - hand side of eq .
( [ t2 ] ) .
since we work in the physical gauge , the gauge linkages of the form @xmath63 $ ] are identical to unity and hence suppressed in our parton fragmentation matrices .
since high - energy reactions are light - cone dominant , the most efficient way to extract the leading contributions from each diagram is to make an expansion about the components of the parton momenta that are collinear to the corresponding hadron momentum .
after that , one must further decouple the discrete indices between the long - distance matrices and the associated short - distance parts . for completing these tasks
, one can employ the collinear technique developed by ellis , furmanski and petonzio @xcite .
let us parameterize the quark momentum as @xmath64 then , the collinear expansion for the lowest - order diagram reads @xmath65 since working at twist two , here we need only to consider the the contributions associated with the leading term . as a result , we obtain @xmath66,\ ] ] with @xmath67 since the color index has been decoupled in eq .
( [ t2 ] ) , now the only task for us is to decouple the dirac indices between the expanded hard - interaction part @xmath68 and the simplified quark fragmentation matrix @xmath69 . expanding the latter in the dirac space
, we have @xmath70 where we have suppressed the spin - independent terms as well as those arising from the hadronic final - state interaction because we do not address @xmath26 and @xmath27 as well as @xmath24 .
@xmath71 , @xmath72 and @xmath73 are quark fragmentation functions ( matrix elements ) at twist two , three , and four respectively , and their definitions can be found in ref .
@xcite . from the chiral - odd structure of @xmath72
, one can immediately see that only by a quark mass insertion in the hard part can it make a nonzero contribution , so its contribution is actually at twist four
. keeping the first term in eq .
( [ dc ] ) , we will have @xmath74 where @xmath75 .
notice that we have recovered the quark flavor index here . to confront with the general decomposition of @xmath76
, we can simply substitute @xmath77 for @xmath36 and correspondingly @xmath78 for @xmath79 .
such replacements amount to including some effects at two higher twist so they are allowed . as a consequence
, we have the following twist - two formulas : @xmath80 @xmath81 in fact , our factorization at leading twist reproduced the naive parton model results about @xmath0 and @xmath1 . the easiest way to recognize
this point is to quantize the quark filed in the light - cone quantization formalism @xcite and carry out the integration over the unobserved hadron system @xcite . physically , if one notices that the quark transverse momentum with respect to the @xmath8 momentum has been integrated out , it is not difficult to understand the equivalence of this twist - two factorization result to the naive parton model approach
. now we turn to the case in which the spin of the inclusive @xmath8 hyperon is aligned along the normal of the production plane , namely , the @xmath82 axis . in this case , @xmath83=0 , so the @xmath0 and @xmath1 terms in eq .
( [ dec ] ) become degenerate . therefore , we hope to express the sum of @xmath0 and @xmath1 in terms of some spin - dependent quark fragmentation functions .
related to the transverse spin of the inclusive @xmath8 particle , the twist - two quark fragmentation function @xmath2 is chiral - odd . as a result ,
its contribution comes about via a quark mass insertion in the hard part and correspondingly is at twist three .
hence , we need to work at least at twist three and the lowest - order diagram as well as those two in fig . 2
construct our subject .
to the order at which we work , @xmath84 \nonumber \\ & & + \frac{1}{4\pi n}\int \frac{d^4 k}{(2\pi)^4 } \frac{d^4 k_1}{(2\pi)^4 } { \rm tr_d } { \rm tr_c}\left [ h^{\mu\nu\sigma}_{(2a)}(q , k , k_1 ) x^\prime _
\sigma(k , k_1,p , s_\perp ) \right ] \nonumber \\ & & + \frac{1}{4\pi n}\int \frac{d^4 k}{(2\pi)^4 } \frac{d^4 k_1}{(2\pi)^4 } { \rm tr_d } { \rm tr_c}\left [ h^{\mu\nu\sigma}_{(2b)}(q , k , k_1 ) y^\prime_\sigma(k , k_1,p , s_\perp)\right ] , \label{www}\end{aligned}\ ] ] where @xmath85 @xmath86 here we note that in writing down eq .
( [ www ] ) , we have absorbed the color matrix along with a minus strong coupling @xmath87 into @xmath88 and @xmath89 . obviously , the gauge invariance of eq .
( [ www ] ) is not manifest .
since we work up to twist three , we need to expand the hard part for the lowest - order diagram to the first nonleading term . as for those shown in fig .
2 , it is enough to take their leading contributions . making use of the ward identity ,
the twist - three contributions associated with the first derivative term in the expansion of the lowest - order hard part can be combined with the leading contributions of the two diagrams shown in fig .
the net effect of such a combination is to replace the gluon field tensor by the covariant derivative operator in the two - variable fragmentation matrices coupled with the hard parts of the diagrams in fig .
2 . however , the ellis - furmanski - petronzio scheme is not a satisfactory procedure to extract nonleading twist contributions @xcite .
the basic reason is that the leading term in the collinear expansion contains as well the nonleading contributions . to isolate such nonleading effects implicit in the leading term ,
the efficient way is to pull down a certain number of `` special '' propagators into the hard part . in our twist - three case
, we need to pull down @xmath90 special propagator along with the connected quark - gluon vertex into the hard part , either on the left - hand side or on the right - hand side of the final - state cut . in other words ,
twist - three contributions hidden in the leading term of the collinear expansion of the lowest - order diagram can be taken into account by including the two diagrams in fig .
3 . see what follows .
the notation of the special propagator @xcite can be introduced as follows .
consider the momentum carried by the quark propagator in fig .
2 that links the electromagnetic vertex to the quark - gluon one , one can parameterize it as @xmath91 where @xmath92 is the on - shell part of @xmath93 .
correspondingly , the quark propagator is decomposed into two parts : @xmath94 technically , @xmath95 is termed the special propagator by qiu and labelled by adding a bar on the normal propagator in the graphic representation . to unravel the physical content of the special propagator ,
let us consider the fourier transformation @xmath96\frac{i\rlap / k}{k^2+i\varepsilon } = \delta(\xi_1^+ - \xi^+_2 ) \frac{i\gamma^+}{2 k^+ } + \theta ( \xi_1^+ -\xi^+_2 ) \frac{\rlap/{\hat k}}{2k^+ } \exp[-i\frac{k^2_t}{2 k^+}(\xi_1^+ -\xi^+_2)],\ ] ] from which it can be seen that the special propagator describes a short - distance , or `` contact '' interaction in the light - cone .
hence , as one goes beyond the leading twist for the lowest - order diagram , a special propagator along with the connected quark - gluon vertex should be pulled down into the hard part . considering that the diagrams shown in fig
3 can be obtained from the lowest - order diagram by pulling out a quark propagator and its connected quark - gluon vertex , it can also be stated that the non - contact part of its linking propagator should be discarded , because they have been included in @xmath97 .
it should be stressed that qiu @xcite has demonstrated that such a cure can naturally reserve the gauge invariance for the hard part . from the above discussions , we know that the formula for calculating the hadronic tensor reads @xmath98 \nonumber \\ & & + \frac{1}{4\pi n}\int d(\frac{1}{z_1 } )
dz { \rm tr_d } { \rm tr_c}\left[\left ( ( h^{\mu\nu\sigma}_{(2a)}(q , p / z , p / z_1 ) + h^{\mu\nu\sigma}_{(3a);{\rm spec}}(q , p / z ) \right)x_\sigma(z , z_1,p , s_\perp ) \right ] \nonumber \\ & & + \frac{1}{4\pi
n } \int d(\frac{1}{z_1 } ) dz { \rm tr_d } { \rm tr_c}\left[\left ( h^{\mu\nu\sigma}_{(2b)}(q , p / z , p / z_1 ) + h^{\mu\nu\sigma}_{(3b);{\rm spec}}(q , p / z ) \right)y_\sigma(z , z_1,p , s_\perp)\right ] , \label{www1}\end{aligned}\ ] ] where @xmath99 @xmath100 @xmath101 in our work , the covariant derivative operator is defined as @xmath102 and @xmath103 .
to arrive at factorized expressions , we decompose the above three fragmentation matrices in the dirac and lorentz spaces .
again , we suppress the spin - independent terms as well as those arising from the final - state interactions for the brevity of formulas . as a result ,
@xmath104_{\alpha\beta } , \label{dec1}\end{aligned}\ ] ] @xmath105 @xmath106 where @xmath107 , @xmath108 , @xmath109 , and @xmath110 are jaffe and ji s parton fragmentation matrix elements , whose definitions can be easily projected out from the above decompositions . using qcd equation of motion
, one can prove @xcite that @xmath111 = -\frac{1}{z}\hat g_t ( z)+ \frac{m}{m } \hat h_1 ( z ) , \label{proj}\ ] ] where @xmath112 is the quark mass . inserting eqs .
( [ dec1])([vdec ] ) into ( [ www1 ] ) and completing the algebra , we obtain @xmath113 \varepsilon_{\mu\nu\tau\rho}q^\tau s^\rho_\perp . \label{res}\end{aligned}\ ] ] here we note this result is complete up to twist three .
again , we make substitutions @xmath114 and @xmath115 . by confronting with eq .
( [ dec ] ) , we attain for @xmath0 and @xmath29 the following relation : @xmath116 .
\label{ss2}\ ] ]
we conclude this paper by discussing the phenomenological implications of our factorization results about @xmath0 and @xmath1 . here
we assume that the @xmath8 hyperon is predominantly produced via the strange quark fragmentation .
then , both in eq .
( [ ss1 ] ) and in ( [ ss2 ] ) the flavor summation can be dropped .
futhermore , the terms associated with the quark mass in ( [ ss2 ] ) can be ignored at a first - order approximation because @xmath117 .
although not discussed here , we can believe that by controlling judiciously the polarization of the initial - state electron beam and analyzing the final - state @xmath8 polarization , it is possible to measure @xmath0 and @xmath1 .
therefore , the measurement of these two structure functions will allow for the determination of two @xmath9 fragmentation functions @xmath71 and @xmath118 .
it should be noted that eq .
( [ ss1 ] ) might be not so helpful .
the reason is that @xmath119 can be independently measured at the lep , which burkardt and jaffe have discussed in ref .
@xcite . at the @xmath120 resonance ,
the parity violation can naturally generate longitudinally - polarized strange quarks so one does not need to polarize the initial - state beam . however , eq . ( [ ss2 ] ) is by no means trivial .
so far , the measurement of chiral - odd nonperturbative matrix elements , to which @xmath121 and @xmath3 belong , have been a challenge to the particle physics community .
so far , the general strategy to treat them has been tackling them in pairs [ 1 ] , [ 2 ] , @xcite .
our results , eq .
( [ ss2 ] ) , indicates that once we have known @xmath121 from other sources or it can be neglected at the first approximation , the measurement of @xmath0 and @xmath1 , present us a relatively clean situation , in which information about twist - three fragmentation function @xmath118 can be obtained .
the penalty in such a scheme is to polarize one of the initial - state beams .
as is well known , it is now quite feasible to obtain highly polarized electron beams . in summary , we worked out a qcd field theory study of @xmath0 and @xmath1 , two spin - dependent structure functions for the inclusive spin - half baryon production in electron - positron annihilation . making use of the collinear expansion procedure by ellis , furmanski and petronzio , and the special propagator concept by qiu
, we derived a formula relating the sum of @xmath0 and @xmath1 to the transverse - spin - dependent quark fragmentation functions @xmath121 and @xmath118 .
on the basis of this finding , we pointed out that it is possible to extract data about @xmath118 in the inclusive @xmath8 hyperon production by electron - positron annihilation . considering the common believe that chiral - odd hadron matrix elements and/or parton fragmentation matrix elements should be measured in pairs
, our results can be taken as an interesting counter example .
jaffe and x. ji , phys .
lett . * 71 * , 2547 ( 1993 ) .
x. ji , phys .
d * 49 * , 114 ( 1994 ) .
b. lundberg et al . , phys .
d * 40 * , 3557 ( 1989 ) ; and the papers cited therein .
p. hansen and u. dickow , aleph 95 - 055 . m. burkardt and r.l .
jaffe , phys .
70 * , 2537 ( 1993 ) .
k. chen , g.r .
goldstein , r.l .
jaffe and x. ji , mit - cpt-2365 .
w. lu , phys .
d * 51 * , 5305 ( 1995 ) .
p. nason and b.r .
webber , nucl .
b421 * , 473 ( 1994 ) .
m. anselmino , a. efremov and e. leader , cern - th/7216/94 , to appear in physics reports .
m. anselmino and e. leader , phys . lett .
* 293b * , 216 ( 1992 ) . j. collins , d.e . soper and g. sterman , in _ perturbative quantum chromodynamics
, edited by a.h .
mueller(world scientific , singapore , 1989 ) .
j. qiu and g. sterman , in _ polarized collider workshop _ , edited by j.c .
collins , s.f . heppelman and r.w .
robinett(american institute of physics , new york , 1990 ) ; nucl .
b353 * , 137 ( 1991 ) .
ellis , w. furmanski and r. petronzio , nucl .
* b207 * , 1 ( 1982 ) ; * b212 * , 29 ( 1983 ) . j.b . kugot and d.e .
soper , phys .
d * 1 * , 2901 ( 1970 ) .
j. levelt and p.j .
mulders , phys . rev .
d * 49 * , 96 ( 1994 ) .
j. qiu , phys .
d * 42 * , 30 ( 1990 ) .
j. levelt and p.j .
mulders , phys .
338b * , 357 ( 1994 ) . 1 . the lowest - order cut diagram for the inclusive @xmath8 hyperon production by a time - like photon .
the cut diagram for the inclusive hyperon production by a time - like photon with one gluon correlation at the stage of parton fragmentation .
the cut diagram for the inclusive @xmath8 hyperon production by a time - like photon with one gluon radiation in the quark fragmentation .
one `` special '' propagator is pulled down into the hard partonic interaction part , either ( a ) on the left - hand side or ( b ) on the right - hand side of the final - state cut . | two spin - dependent structure functions @xmath0 and @xmath1 for the inclusive spin - half baryon production in electron - positron annihilation are studied in the context of qcd factorization as well as in the naive quark parton model . as a result , it is found that the sum of @xmath0 and @xmath1 is related to @xmath2 and @xmath3 , two quark fragmentation functions defined by jaffe and ji . in connection with the measurement of quark fragmentation functions ,
the possible phenomenological consequences are discussed .
spin - dependent structure functions @xmath0 and @xmath1 for inclusive spin - half baryon production in electron - positron annihilation @xmath4 .
ccast ( world laboratory ) , p.o .
box 8730 , beijing 100080 , china @xmath5 .
institute of high energy physics , p.o .
box 918(4 ) , beijing 100039 , china @xmath6 .
department of physics , nankai university , tianjin 300071 , china @xmath7 .
department of physics , peking university , beijing 100871 , 300071 , china |
we consider the model @xmath10 where the random vector @xmath1 and the random @xmath2 matrix @xmath3 are observed , the @xmath2 matrix @xmath4 is unknown , @xmath5 is an @xmath2 random noise matrix , @xmath6 is a random noise vector , @xmath11 is a vector of unknown parameters to be estimated , and @xmath12 is a given subset of @xmath13 .
we consider the problem of estimating an @xmath14-sparse vector @xmath7 ( i.e. , a vector @xmath7 having only @xmath14 non zero components ) , with @xmath8 possibly much larger than @xmath9 . if the matrix @xmath4 in is observed without error ( @xmath15 ) , this problem has been recently studied in numerous papers .
the proposed estimators mainly rely on @xmath16 minimization techniques .
in particular , this is the case for the widely used lasso and dantzig selector , see among others cands and tao ( 2007 ) , bunea et al .
( 2007a , b ) , bickel et al .
( 2009 ) , koltchinskii ( 2009 ) , the book by bhlmann and van de geer ( 2011 ) , the lecture notes by koltchinskii ( 2011 ) , belloni and chernozhukov ( 2011 ) and the references cited therein . however , it is shown in rosenbaum and tsybakov ( 2010 ) that dealing with a noisy observation of the regression matrix @xmath4 has severe consequences . in particular
, the lasso and dantzig selector become very unstable in this context .
an alternative procedure , called the matrix uncertainty selector ( mu selector for short ) is proposed in rosenbaum and tsybakov ( 2010 ) in order to account for the presence of noise @xmath5 .
the mu selector @xmath17 is defined as a solution of the minimization problem @xmath18 where @xmath19 denotes the @xmath20-norm , @xmath21 , @xmath22 is a given subset of @xmath23 characterizing the prior knowledge about @xmath7 , and the constants @xmath24 and @xmath25 depend on the level of the noises @xmath5 and @xmath26 respectively .
if the noise terms @xmath26 and @xmath5 are deterministic , it is suggested in rosenbaum and tsybakov ( 2010 ) to choose @xmath25 such that @xmath27 and to take @xmath28 with @xmath29 such that @xmath30 where , for a matrix @xmath31 , we denote by @xmath32 its componentwise @xmath33-norm . in this paper
, we propose a modification of the mu selector for the model where @xmath5 is a random matrix with independent and zero mean entries @xmath34 such that the sums of expectations @xmath35 are finite and admit data - driven estimators .
our main example where such estimators exist is the model with data missing at random ( see below ) .
the idea underlying the new estimator is the following .
in the ideal setting where there is no noise @xmath5 , the estimation strategy for @xmath7 is based on the matrix @xmath4 . when there is noise this is impossible since @xmath4 is not observed and so we have no other choice than using @xmath3 instead of @xmath4 .
however , it is not hard to see that under the above assumptions on @xmath5 , the matrix @xmath36 appearing in contains a bias induced by the diagonal entries of the matrix @xmath37 whose expectations @xmath38 do not vanish .
if @xmath38 can be estimated from the data , it is natural to make a bias correction .
this leads to a new estimator @xmath39 defined as a solution of the minimization problem @xmath40 where @xmath41 is the diagonal matrix with entries @xmath42 , which are estimators of @xmath38 , and @xmath43 and @xmath44 are constants that will be specified later .
this estimator @xmath39 will be called the compensated mu selector . in this paper , we show both theoretically and numerically that the estimator @xmath39 achieves better performance than the original mu selector @xmath17 . in particular , under natural conditions given below , the bounds on the error of the compensated mu selector decrease as @xmath45 up to logarithmic factors as @xmath46 , whereas for the original mu selector @xmath17 the corresponding bounds do not decrease with @xmath9 and can be only small if the noise @xmath5 is small .
[ rem:1 ] the problem ( [ cmu ] ) is equivalent to @xmath47 where @xmath48 with the same @xmath24 and @xmath25 as in ( [ cmu ] ) ( see the proof in section [ sec : preuve ] ) .
this simplifies in some cases the computation of the solution .
an important example where the values @xmath38 can be estimated is given by the model with missing data .
assume that the elements @xmath49 of the matrix @xmath4 are unobservable , and we can only observe @xmath50 where for each fixed @xmath51 , the factors @xmath52 are i.i.d .
bernoulli random variables taking value 1 with probability @xmath53 and 0 with probability @xmath54 , @xmath55 .
the data @xmath49 is missing if @xmath56 , which happens with probability @xmath54 . we can rewrite ( [ ex1 ] ) in the form @xmath57 where @xmath58 , @xmath59 .
thus , we can reduce the model with missing data ( [ ex1 ] ) to the form ( [ 00 ] ) with a matrix @xmath5 whose elements @xmath34 have zero mean and variance @xmath60 .
so , @xmath61 in section [ sec : stoch ] below , we show that when the @xmath54 are known , the @xmath38 admit good data - driven estimators @xmath62 .
if the @xmath54 are unknown , they can be readily estimated by the empirical frequencies of 0 that we further denote by @xmath63 .
then the @xmath64 appearing in ( [ ex2 ] ) are not available and should be replaced by @xmath65 .
this slightly changes the model and implies a minor modification of the estimator ( cf .
section [ sec : stoch]).
consider the following random matrices @xmath66 @xmath67 where @xmath68 is the diagonal matrix with diagonal elements @xmath38 , @xmath51 , and for a square matrix @xmath31 , we denote by @xmath69 the matrix with the same dimensions as @xmath31 , the same diagonal elements as @xmath31 and all off - diagonal elements equal to zero . under conditions that will be specified below , the entries of the matrices @xmath70 are small with probability close to 1 .
bounds on the @xmath33-norms of the matrices @xmath70 characterize the stochastic error of the estimation .
the accuracy of the estimators is determined by these bounds and by the properties of the gram matrix @xmath71 for a vector @xmath72 , we denote by @xmath73 the vector in @xmath13 that has the same coordinates as @xmath72 on the set of indices @xmath74 and zero coordinates on its complement @xmath75 .
we denote by @xmath76 the cardinality of @xmath77 . to state our results in a general form , we follow gautier and tsybakov ( 2011 ) and introduce the sensitivity characteristics related to the action of the matrix @xmath78 on the cone @xmath79 .
for @xmath80 $ ] and an integer @xmath81 $ ] , we define the _ @xmath82 sensitivity _ as follows : @xmath83 we will also consider the _ coordinate - wise sensitivities _ @xmath84 where @xmath85 is the @xmath86th coordinate of @xmath87 , @xmath88 . to get meaningful bounds for various types of estimation errors , we will need the positivity of @xmath89 or @xmath90 .
as shown in gautier and tsybakov ( 2011 ) , this requirement is weaker than the usual assumptions related to the structure of the gram matrix @xmath78 , such as the restricted eigenvalue assumption and the coherence assumption . for completeness , we recall these two assumptions .
* assumption re(@xmath14 ) . *
_ let @xmath91 .
there exists a constant @xmath92 such that @xmath93 for all subsets @xmath77 of @xmath94 of cardinality @xmath95 . _
* assumption c. * _ all the diagonal elements of @xmath78 are equal to 1 and all its off - diagonal elements of @xmath96 satisfy the coherence condition : @xmath97 for some @xmath98 . _
+ note that assumption c with @xmath99 implies assumption re(@xmath14 ) with @xmath100 , see bickel et al .
( 2009 ) or lemma 2 in lounici ( 2008 ) . from proposition 4.2 of gautier and tsybakov ( 2011 )
we get that , under assumption c with @xmath101 , @xmath102 which yields the control of the sensitivities @xmath103 for all @xmath104 since @xmath105 by proposition 4.1 of gautier and tsybakov ( 2011 ) .
furthermore , proposition 9.2 of gautier and tsybakov ( 2011 ) implies that , under assumption re(@xmath14 ) , @xmath106 and by proposition 9.3 of that paper , under assumption re(@xmath107 ) for any @xmath108 and any @xmath109 , we have @xmath110 where @xmath111 .
in this section , we give bounds on the estimation and prediction errors of the compensated mu selector . for @xmath112 , we consider the thresholds @xmath113 and @xmath114 , @xmath115 , such that @xmath116 and @xmath117 define @xmath118 and @xmath119 where @xmath120 and @xmath12 is a given subset of @xmath23 . for @xmath112 ,
the compensated mu selector is defined as a solution of the minimization problem @xmath121 we have the following result .
[ t1 ] assume that model ( [ 0])([00 ] ) is valid with an @xmath14-sparse vector of parameters @xmath122 , where @xmath22 is a given subset of @xmath23 .
for @xmath112 , set @xmath123 then , with probability at least @xmath124 , the set @xmath125 is not empty and for any solution @xmath126 of we have @xmath127 the proof of this theorem is given in section [ sec : preuve ] . note that ( [ t1:3 ] ) contains a bound on the prediction error under no assumption on @xmath4 : @xmath128 the other bounds in theorem [ t1 ] depend on the sensitivities . using ( [ k1 ] ) ( [ k4 ] ) we obtain the following corollary of theorem [ t1 ] .
[ t2 ] let the assumptions of theorem [ t1 ] be satisfied .
then , with probability at least @xmath124 , for any solution @xmath126 of we have the following inequalities .
\(i ) under assumption re(@xmath14 ) : @xmath129 ( ii ) under assumption re(@xmath107 ) , @xmath108 : @xmath130 ( iii ) under assumption c with @xmath131 : @xmath132 where we set @xmath133 .
if the components of @xmath26 and @xmath5 are subgaussian , the values @xmath134 are of order @xmath45 up to logarithmic factors , and the value @xmath135 is of the same order in the model with missing data ( see section [ sec : stoch ] ) .
then , the bounds for the compensated mu selector in theorem [ t2 ] are decreasing with rate @xmath136 as @xmath137 .
this is an advantage of the compensated mu selector as compared to the original mu selector @xmath17 , for which the corresponding bounds do not decrease with @xmath9 and can be small only if the noise @xmath5 is small ( cf .
rosenbaum and tsybakov ( 2010 ) ) . if the matrix @xmath4 is observed without error ( @xmath15 ) , then @xmath138 , @xmath139 , and the compensated mu selector coincides with the dantzig selector . in this particular case , the results ( ii ) and ( iii ) of theorem [ t2 ] improve , in terms of the constants or the range of validity , upon the corresponding bounds in bickel et al .
( 2009 ) and lounici ( 2008 ) .
theorems [ t1 ] and [ t2 ] are stated with general thresholds @xmath134 and @xmath135 , and can be used both for random or deterministic noises @xmath140 ( in the latter case , @xmath141 ) and random or deterministic @xmath4 . in this section ,
considering @xmath142 we first derive the values @xmath134 for random @xmath26 and @xmath5 with subgaussian entries , and then we specify @xmath135 and the matrix @xmath143 for the model with missing data .
note that , for random @xmath26 and @xmath5 , the values @xmath134 and @xmath135 characterize the stochastic error of the estimator .
recall that a zero - mean random variable @xmath144 is said to be @xmath145-subgaussian ( @xmath146 ) if , for all @xmath147 , @xmath148\leq \text{exp}(\gamma^2t^2/2).\ ] ] in particular , if @xmath144 is a zero - mean gaussian or bounded random variable , it is subgaussian .
a zero - mean random variable @xmath144 will be called @xmath149-subexponential if there exist @xmath146 and @xmath150 such that @xmath151\leq \text{exp}(\gamma^2t^2/2 ) , \quad \forall \ @xmath5 satisfy the following assumption .
* assumption n. * _ let @xmath152 , @xmath153 .
the entries @xmath34 , @xmath154 of the matrix @xmath5 are zero - mean @xmath155-subgaussian random variables , the @xmath9 rows of @xmath5 are independent , and @xmath156 for @xmath157 , @xmath158 . the components @xmath159 of the vector @xmath26 are independent zero - mean @xmath160-subgaussian random variables satisfying @xmath161 , @xmath162 . _
assumption n implies that the random variables @xmath163 , @xmath164 are subexponential . indeed , if two random variables @xmath165 and @xmath166 are subgaussian , then for some @xmath167 we have @xmath168 , which implies that ( [ subexp ] ) holds for @xmath169 with some @xmath170 whenever @xmath171 , cf . , e.g. , petrov ( 1995 ) , page 56 .
next , @xmath172 is a zero - mean subexponential random variable with variance @xmath173 .
it is easy to check that ( [ subexp ] ) holds for @xmath174 with @xmath175 and @xmath176 . to simplify the notation
, we will use a rougher evaluation valid under assumption n , namely that all @xmath163 , @xmath164 are @xmath177-subexponential with the same @xmath178 and @xmath150 , and all @xmath179 are @xmath180-subexponential . here the constants @xmath181 and @xmath182 depend only on @xmath155 and @xmath160 . for @xmath183 and an integer @xmath184 , set @xmath185 [ lem : subexp ]
let assumption n be satisfied , and let @xmath4 be a deterministic matrix with @xmath186 .
then for any @xmath183 the bound ( [ pr2 ] ) holds with @xmath187 use the union bound and the facts that @xmath188 for a @xmath145-subgaussian @xmath144 , and @xmath189 for a sum of independent @xmath190-subexponential @xmath191 .
consider now the model with missing data ( [ ex1 ] ) and assume that @xmath4 is non - random .
then we have @xmath192 , which implies : @xmath193 = x_{ij}^2(1-\pi_j)\ , , \quad j=1,\dots , p.\ ] ] hence , @xmath194 is an unbiased estimator of @xmath195 .
then @xmath196 defined in ( [ ex3 ] ) is naturally estimated by @xmath197 the matrix @xmath143 is then defined as a diagonal matrix with diagonal entries @xmath198 .
it is not hard to prove that @xmath198 approximates @xmath196 in probability with rate @xmath45 up to a logarithmic factor .
for example , let the probability that the data is missing be the same for all @xmath199 : @xmath200 .
then @xmath201 where we have used the fact that @xmath202 , hoeffding s inequality and the notation @xmath203 .
this proves ( [ pr1 ] ) with @xmath204 if @xmath205 is unknown , we replace it by the estimator @xmath206 , where @xmath207 denotes the indicator function .
another difference is that @xmath64 appearing in ( [ ex2 ] ) are not available when @xmath54 s are unknown .
therefore , we slightly modify the estimator using @xmath208 instead of @xmath209 ; we define @xmath210 as a solution of @xmath211 with @xmath212 where @xmath213 and @xmath214 are suitably chosen constants , @xmath215 is the @xmath2 matrix with entries @xmath216 , and @xmath143 is a diagonal matrix with entries @xmath217 this modification introduces in the bounds an additional term proportional to @xmath218 , which is of the order @xmath219 in probability and hence is negligible as compared to the error bound for the compensated mu selector . in this section ,
we have considered non - random @xmath4 . using the same argument
, it is easy to derive analogous expressions for @xmath220 and @xmath221 when @xmath4 is a random matrix with independent sub - gaussian entries , and @xmath26 , @xmath5 are independent from @xmath4 .
the bounds of theorems [ t1 ] and [ t2 ] depend on the unknown matrix @xmath4 via the sensitivities , and therefore can not be used to provide confidence intervals . in this section ,
we show how to address the issue of confidence intervals by deriving other type of bounds based on the empirical sensitivities .
note first that the matrix @xmath222 is a natural estimator of the unknown gram matrix @xmath78 .
it is @xmath223-consistent in @xmath33-norm under the conditions of the previous section .
therefore , it makes sense to define the empirical counterparts of @xmath89 and @xmath90 by the relations : @xmath224 that we will call the _ empirical sensitivities _ can be efficiently computed for small @xmath14 or , alternatively , one can compute data - driven lower bounds on them for any @xmath14 using linear programming , cf . gautier and tsybakov ( 2011 ) .
the following theorem establishes confidence intervals for @xmath14-sparse vector @xmath7 based on the empirical sensitivities .
[ t3 ] assume that model ( [ 0])([00 ] ) is valid with an @xmath14-sparse vector of parameters @xmath122 , where @xmath22 is a given subset of @xmath23 .
then , with probability at least @xmath124 , for any solution @xmath126 of we have @xmath225 where @xmath226 , and we set @xmath227 .
set @xmath228 , and write for brevity @xmath229 using lemma [ lem3 ] in section [ sec : preuve ] , the fact that @xmath230 where @xmath77 is the set of non - zero components of @xmath231 ( cf .
lemma 1 in rosenbaum and tsybakov ( 2010 ) ) and the definition of the empirical sensitivity @xmath232 , we find @xmath233 this and the definition of @xmath234 yield ( [ t3:1 ] ) .
the proof of ( [ t3:2 ] ) is analogous , with @xmath235 used instead of @xmath234 .
note that the bounds ( [ t3:1])([t3:2 ] ) remain valid for @xmath236 .
therefore , if one gets an estimator @xmath237 of @xmath14 such that @xmath238 with high probability , it can be plugged in into the bounds in order to get completely feasible confidence intervals .
we consider here the model with missing data ( [ ex1 ] ) .
simulations in rosenbaum and tsybakov ( 2010 ) indicate that in this model the mu selector achieves better numerical performance than the lasso or the dantzig selector .
here we compare the mu selector with the compensated mu selector .
we design the numerical experiment the following way .
@xmath239 we take a matrix @xmath4 of size @xmath240 ( @xmath241 ) which is the normalized version ( centered and then normalized so that all the diagonal elements of the associated gram matrix @xmath242 are equal to 1 ) of a @xmath243 matrix with i.i.d .
standard gaussian entries .
+ @xmath239 for a given integer @xmath14 , we randomly ( uniformly ) choose @xmath14 non - zero elements in a vector @xmath7 of size @xmath244 .
the associated coefficients @xmath245 are set to @xmath246 , and all other coefficients are set to 0 .
we take @xmath247 .
+ @xmath239 we set @xmath248 , where @xmath26 a vector with i.i.d .
zero mean and variance @xmath249 normal components , @xmath250 .
+ @xmath239 we compute the values @xmath251 with @xmath208 as in , . ] and @xmath252 for all @xmath199 .
( the value @xmath205 rather than its empirical counterpart , which is very close to @xmath205 , is used in the algorithm to simplify the computations ) .
+ @xmath239 we run a linear programming algorithm to compute the solutions of and where we optimize over @xmath253 . to simplify the comparison with rosenbaum and tsybakov ( 2010 )
, we write @xmath24 in the form @xmath254 with @xmath255 .
in particular , @xmath256 corresponds to the dantzig selector based on the noisy matrix @xmath3 . in practice
, one can use an empirical procedure of the choice of @xmath257 described in rosenbaum and tsybakov ( 2010 ) .
the choice of @xmath25 is not crucial and influences only slightly the output of the algorithm .
the results presented below correspond to @xmath25 chosen in the same way as in the numerical study in rosenbaum and tsybakov ( 2010 ) .
+ @xmath239 we compute the error measures @xmath258 .
+ @xmath239 for each value of @xmath14 we run @xmath259 monte carlo simulations
. tables 15 present the empirical averages and standard deviations ( in brackets ) of @xmath260 , @xmath261 , of the number of non - zero coefficients in @xmath262 ( @xmath263 ) and of the number of non - zero coefficients in @xmath262 belonging to the true sparsity pattern ( @xmath264 ) .
we also present the total number of simulations where the sparsity pattern is exactly retrieved ( exact ) .
the lines with @xmath265 " for @xmath266 correspond to the mu selector and those with @xmath267 " to the compensated mu selector . + [ cols="^,^,^,^,^,^",options="header " , ] the results of the simulations are quite convincing .
indeed , the compensated mu selector improves upon the mu selector with respect to all the considered criteria , in particular when @xmath7 is very sparse ( @xmath268 ) .
the order of magnitude of the improvement is such that , for the best @xmath257 , the errors @xmath260 and @xmath261 are divided by @xmath269 .
the improvement is not so significant for larger @xmath14 , especially for @xmath270 when the model starts to be not very sparse .
for all the values of @xmath14 , the non - zero coefficients of @xmath7 are systematically in the sparsity pattern both of the mu selector and of the compensated mu selector .
the total number of non - zero coefficients is always smaller ( i.e. , closer to the correct one ) for the compensated mu selector .
finally , note that the best results for the error measures @xmath260 and @xmath261 are obtained with @xmath271 , while the sparsity pattern is better retrieved for @xmath272 .
this reflects a trade - off between estimation and selection .
* proof of remark [ rem:1 ] . * it is enough to show that @xmath273 where @xmath274 let first @xmath275 . using the triangle inequality , we easily get that @xmath276 .
now take @xmath276 .
we set @xmath277 and consider @xmath278 defined by @xmath279 for @xmath280 , where @xmath281 and @xmath282 are the @xmath283th components of @xmath284 and @xmath184 respectively .
it is easy to check that @xmath285 , which concludes the proof .
we first write that @xmath290 is equal to @xmath291 by definition of the @xmath134 and @xmath135 , with probability at least @xmath124 we have @xmath292 therefore @xmath289 with probability at least @xmath124 . throughout the proof
, we assume that we are on event of probability at least @xmath124 where inequalities hold and @xmath289 .
we have @xmath295 consequently , @xmath296 using that @xmath297 , we easily get that @xmath298 is not greater than @xmath299 now remark that @xmath300 finally , using that @xmath301 together with the fact that @xmath302 , we obtain the result .
we now proceed to the proof of theorem [ t1 ] .
the bounds ( [ t1:1 ] ) and ( [ t1:2 ] ) follow from lemma [ lem4 ] , the fact that @xmath230 where @xmath77 is the set of non - zero components of @xmath231 ( cf . lemma 1 in rosenbaum and tsybakov ( 2010 ) ) and the definition of the sensitivities @xmath103 , @xmath303 . to prove ( [ t1:3 ] ) , first note that @xmath304 and use ( [ t1:1 ] ) with @xmath305 and lemma [ lem4 ] .
this yields the first term under the minimum on the right hand side of ( [ t1:3 ] ) .
the second term is obtained again from ( [ p1 ] ) , lemma [ lem4 ] and the inequality @xmath306 .
* the bounds ( [ t2:1 ] ) and ( [ t2:4 ] ) follow by combining ( [ t1:1 ] ) with ( [ k3 ] ) and with ( [ k1 ] ) ( [ k2 ] ) respectively .
next , ( [ t2:2 ] ) follows from ( [ t1:3 ] ) and ( [ k3 ] ) . also , as an easy consequence of ( [ t1:1 ] ) and ( [ k4 ] ) with @xmath307 we get @xmath308 finally , ( [ t2:3 ] ) follows from this inequality and ( [ t2:1 ] ) using the interpolation formula @xmath309 , and the fact that @xmath310 .
belloni , a. , and chernozhukov , v. ( 2011 ) .
high dimensional sparse econometric models : an introduction . in : _ inverse problems and high dimensional estimation , stats in the chteau 2009 _ , alquier , p. , e. gautier , and g. stoltz , eds . , _ lecture notes in statistics _ , * 203 * 127162 , springer , berlin . | we consider the regression model with observation error in the design : @xmath0 here the random vector @xmath1 and the random @xmath2 matrix @xmath3 are observed , the @xmath2 matrix @xmath4 is unknown , @xmath5 is an @xmath2 random noise matrix , @xmath6 is a random noise vector , and @xmath7 is a vector of unknown parameters to be estimated .
we consider the setting where the dimension @xmath8 can be much larger than the sample size @xmath9 and @xmath7 is sparse . because of the presence of the noise matrix @xmath5 , the commonly used lasso and dantzig selector are unstable .
an alternative procedure called the matrix uncertainty ( mu ) selector has been proposed in rosenbaum and tsybakov ( 2010 ) in order to account for the noise .
the properties of the mu selector have been studied in rosenbaum and tsybakov ( 2010 ) for sparse @xmath7 under the assumption that the noise matrix @xmath5 is deterministic and its values are small . in this paper , we propose a modification of the mu selector when @xmath5 is a random matrix with zero - mean entries having the variances that can be estimated .
this is , for example , the case in the model where the entries of @xmath4 are missing at random .
we show both theoretically and numerically that , under these conditions , the new estimator called the compensated mu selector achieves better accuracy of estimation than the original mu selector . |
the borexino experiment @xcite , under construction at gran sasso , is designed to study @xmath5be solar neutrinos @xcite through a real - time , low - background detector , consisting of a nylon sphere ( 8.5 m in diameter ) filled with high purity organic scintillator ( pseudocumene , c@xmath6h@xmath7 ) .
the apparatus can be calibrated through external ( anti-)neutrino sources as well as through light sources @xcite .
for the first type of calibration experiment , two sources have been considered : @xmath1cr and @xmath3sr .
the @xmath1cr source generates @xmath0 through the reaction @xmath8 , with a half - life @xmath9 of 27.7 days and four energy lines : @xmath10 mev ( 9% ) , @xmath11 mev ( 81% ) , @xmath12 mev ( 1% ) , @xmath13 mev ( 9% ) .
the @xmath3sr source generates @xmath2 through the reaction @xmath14 ( @xmath15 y ) followed by @xmath16 ( @xmath17 h ) . since @xmath18
, one can simply assume that two @xmath2 s are produced for each @xmath3sr nucleus decay .
the total standard spectrum of this source is given by @xmath19 , where each @xmath20 ( @xmath21 ) is calculated using fermi theory @xcite : @xmath22 in eq .
( [ lambda ] ) , @xmath23 is a normalization factor [ so that @xmath24 , @xmath25 is the endpoint energy ( @xmath26 mev and @xmath27 mev ) , @xmath28 is the electron mass , and @xmath29 @xmath30 being the atomic number of the decaying nuclei and @xmath31=1/137.036 @xcite .
the standard @xmath3sr-@xmath3y spectrum is shown in fig . [ spectrum ] .
artificial neutrino sources of known activity and spectra can be used to probe nonstandard neutrino properties , such as flavor oscillations or magnetic moment @xcite or non - standard neutrino couplings ( for example through the exchange of an additional @xmath32 boson @xcite ) . in particular , in @xcite the combined analysis of @xmath2-@xmath4 scattering and inverse beta decay using only a @xmath3sr source has been proposed for studying both oscillation and non - standard neutrino couplings . in this paper
we study in detail how to use the @xmath1cr and @xmath3sr sources to probe flavor oscillations , electromagnetic properties , and deviations of the axial and vector couplings @xmath33 and @xmath34 from their standard value in @xmath0-@xmath4 scattering .
the sensitivity of the experiments depends on the initial activity of the sources , which is constrained by technical and economical budget limits . as a reasonable reference value
we use 5 mci@xmath35 decay / s for each source activity .
the plan of this work is the following : in sec .
ii we describe the calibration experiments and calculate the standard expectations . in sec .
iii we calculate the effect of oscillations into active or sterile states and determine the borexino sensitivity to the neutrino mass and mixing . in sec .
iv we study how borexino can probe neutrino magnetic and anapole moments . in sec .
v we show how this experiment can put interesting limits on the @xmath0-@xmath4 vector and axial couplings . finally , we draw our conclusions in sec . vi .
in this section we describe the radioactive source experiments and the standard rates induced by the sources . we assume the following technical setup : 100 tons ( @xmath36 protons , @xmath37 electrons ) of spherical ( @xmath38 m ) fiducial volume ( fv ) and 5 mci pointlike sources placed at distance @xmath39 m from the detector center . in the @xmath1cr source experiment ( @xmath0 ) the detection process is neutrino - electron elastic scattering , while in the @xmath3sr experiment ( @xmath2 ) the interactions are elastic scattering and inverse @xmath40-decay .
the scattering events are identified through the scintillation light from the electron .
the inverse @xmath40-decays are identified through the delayed coincidence between the prompt positron annihilation signal and the neutron capture @xmath41 s .
the standard ( std ) differential cross section of the scattering process @xmath42 , as a function of the the incident neutrino energy @xmath43 and of the recoil electron kinetic energy @xmath44 , is @xcite : @xmath45 \ , , \label{diffcross}\ ] ] where @xmath46 and @xmath47 , @xmath48 , with @xmath49 ( @xmath50 ) if the incident neutrino is a @xmath0 ( @xmath51 ) .
for antineutrinos , @xmath52 . we take @xmath53 @xcite . the cross section in a definite energy window
@xmath54 $ ] is given by @xmath55 where @xmath56 accounts for the finite detector resolution @xcite , @xmath57 \ , .\label{w}\ ] ] we assume @xmath58 , as obtained by the montecarlo simulations of the apparatus @xcite . our reference choice for the energy window of the @xmath0-@xmath4 scattering experiment is @xmath59 $ ] mev .
the lower limit efficiently cuts the @xmath60c decay background , and the upper limit is safely above the compton edge ( @xmath61 mev ) for the electrons scattered by @xmath1cr neutrinos . for the @xmath2-@xmath4 scattering
we choose @xmath62 $ ] mev , the upper limit now being determined by the @xmath63k contaminant [ which emits @xmath41 s ( b.r.=10% , @xmath64 mev ) and @xmath40 s ( b.r.=90% , @xmath65 mev ) ] in the scintillator . for the inverse @xmath40-decay process @xmath66 , ( characterized by a threshold @xmath67 mev ) , the cross section is @xcite @xmath68 with @xmath69 @xmath70 , and @xmath71 mev .
we actually improve eq .
( [ invb ] ) to account for the weak magnetism and bremsstrahlung corrections , as described in @xcite . for an exposure time
@xmath72 , the expected number of events , @xmath73 , is given by @xmath74 where @xmath75 is the volume density of targets in the fiducial volume fv , @xmath76 is the `` transport '' time elapsed from the source activation to the beginning of the source experiment , @xmath77 is the intensity of the source ( @xmath78 mci ) , @xmath79 is the distance between the source and the generic point @xmath80 in the detector volume , and @xmath81 given the geometry of the experiment , the integral over the volume can be performed analytically and eq .
( [ n0 ] ) can be recast in the following , compact form : @xmath82 where @xmath83 , @xmath84 , and the function @xmath85 is given by @xmath86 \label{f}\ ] ] [ where in our case , @xmath87 and @xmath88 $ ] .
we assume @xmath89 days for both sources ( cr and sr ) . for the @xmath1cr experiment , we take @xmath90 days , which maximizes the signal - to - noise ratio , as shown in @xcite . for the @xmath3sr experiment ,
the useful time limit can be determined by the condition that the statistical uncertainty of the rate reaches the size of the systematic error of the source activity ( about 1% ) @xcite .
this leads us to @xmath91 years as a realistic exposure time .
( a longer experiment would also interfere with the measurement of the solar @xmath92 rate , which is the main goal of borexino . )
now we discuss the background rates . for the scattering events , the background events are due both to solar neutrinos interactions and to the internal decays of radiocontaminants .
the background rate @xmath93 can be measured during the `` source off '' operation of the apparatus , and the number of background events is then simply given by @xmath94 .
for standard solar model expectations @xcite , the total background rate is expected to be @xmath95 events / day for events with @xmath96 $ ] mev and @xmath97 events / day when @xmath98 $ ] mev @xcite . for the inverse
@xmath40-decay events , the delayed coincidence signature allows an almost complete background rejection ( apart from 10 event / year due to antineutrinos coming from nuclear reactors @xcite ) , so we simply set @xmath99 in this case . although the background can be measured in the source off mode and subtracted from the total signal , it contributes to the uncertainties through statistical fluctuations .
the total ( signal+background ) @xmath100 uncertainty of the signal is @xmath101 where @xmath102 is the estimated uncertainty of the source activity and errors have been added in quadrature . a deviation from the standard expectations
can be evidenced at 90% c.l .
if the measured rate @xmath103 satisfies @xmath104 where @xmath105 ( for the two degrees of freedom that we will consider ) .
finally , in this paper we compare the borexino performance with munu @xcite , a reactor experiment mounted at the bugey laboratory , designed explicitly for studying @xmath2-@xmath4 scattering .
munu has been running since august 1998 . for this experiment
we consider the energy window @xmath106=[0.5,1]$ ] mev @xcite .
the energy resolution is @xmath107 @xcite .
the energy spectrum for the @xmath2 was taken from @xcite .
the standard expected rate is 5.3 events / day while the background is estimated in 6 events / day @xcite .
the uncertainty on the neutrino flux from the reactor is @xmath1085% @xcite . as reference
, we set @xmath109 y. in tab .
[ tab1 ] we report the background , the standard expectation , the uncertainty , and the @xmath110 for the three measure ( @xmath92-@xmath4 and @xmath111-@xmath4 scattering , and inverse @xmath40-decay ) of interest for this paper and for the munu experiment .
neutrino flavor oscillations @xcite represent a viable solution to the so - called solar neutrino problem @xcite and to the atmospheric neutrino anomaly @xcite .
this phenomenon can also be probed at accelerators @xcite and reactors @xcite .
the flavor survival probability of a neutrino with energy @xmath43 , at a distance @xmath112 from the source is @xmath113 \ , , \label{pveve}\ ] ] where @xmath114 is the mixing angle and @xmath115 is the difference between the square of the two neutrino masses .
( we have assumed only two neutrino families for simplicity ) .
a deficit of the measured rate in the borexino source experiments might signal neutrino oscillations , i.e. , the disappearance of the initial flavor @xmath0 into either active states ( say , @xmath51 ) or sterile states ( @xmath116 ) . we assume two - family @xmath117 or @xmath118 oscillations . in the presence of oscillations , eq .
( [ n0 ] ) transforms in the following way : @xmath119 @xmath120 \ , .\label{np}\ ] ] in eq .
( [ np ] ) , @xmath121 for @xmath117 and @xmath122 for @xmath123 and inverse beta - decay . inserting the expression for the probability ( [ pveve ] ) in eq .
( [ np ] ) , we obtain , through eq .
( [ n01 ] ) : @xmath124 \ , , \label{np1}\ ] ] where @xmath73 is the standard expectation and @xmath125 and @xmath126 \lambda(e_{\nu})g(r / d,\delta m^{2}d/2 e_{\nu } ) } { \int de_{\nu}\,\sigma_{e}(e_{\nu})\lambda(e_{\nu } ) } \ , , \label{gamma}\ ] ] where the function @xmath127 is given by @xmath128 this function can be calculated analytically ( see the appendix ) . from eqs .
( [ np1 ] ) and ( [ limit ] ) we obtain a compact form for the 90% limit in the plane @xmath129 : @xmath130 in fig . [ osc ] we show the 90% c.l .
contours in the plane @xmath131 for the @xmath3sr antineutrinos in the case of @xmath92-@xmath4 scattering ( short - dashed line ) and inverse @xmath40-decay ( dotted line ) , and the for @xmath1cr neutrinos ( long - dashed line ) , for the @xmath117 [ panel ( a ) ] and @xmath123 [ panel ( b ) ] transitions .
the gray area is the combined fit ( i.e. , the allowed zone if no deficit were found ) .
superposed we show also the 90% c.l .
bounds coming from negative evidence for oscillations ( solid thick line ) and the lsnd allowed area ( solid thin line ) .
the function @xmath132 drops rapidly to zero for @xmath133 ev@xmath134 , because of the rapid oscillation of the cosine term in eq .
( [ gdraft ] ) .
consequently , for @xmath135 ev@xmath134 , the testable value of @xmath136 tends to the constant value @xmath137 .
as expected , in the case of @xmath92-@xmath4 and @xmath111-@xmath4 scattering , the sensitivity in the @xmath123 channel is thus better , due to the absence of the @xmath138 term .
the higher statistics attainable make the @xmath3sr source more appropriate to check lower values of @xmath139 for high values of @xmath140 . on the other hand ,
the low energy of the main decay branch of the @xmath1cr source make it more appropriate to check lower values of @xmath140 , although it can not compete with the chooz sensitivity . in fig .
[ osc](a ) and ( b ) , the thick solid lines represent the 90% c.l .
exclusion contours due to the @xmath141-combination of the negative results coming from short baseline reactor experiments ( bugey @xcite , krasnoyarsk @xcite , gsgen @xcite ) and from the long baseline reactor experiment chooz @xcite , searching for the disappearance channel ( @xmath142 ) , together with the negative results from the accelerator experiments e776 ( @xmath143 ) @xcite , karmen2 ( @xmath144 ) @xcite , and cdhsw ( @xmath145 ) @xcite . in the case of @xmath123 oscillations [ panel(b ) ] only the disappearance channel can be probed ( through reactor experiments ) . in the panel
( a ) , the thin solid line defines the 90% c.l .
favored region delimited by the positive signal of the lsnd experiment @xcite as obtained in @xcite .
notice that only a very small region of the parameter space allowed by lsnd survives the comparison with negative searches . from panel ( a ) of fig .
[ osc ] we see that the zone tested by the borexino source experiments , ( in the hypothesis of @xmath146 oscillations ) is already contained in the current bounds .
therefore , if a significative difference from the standard expectation is found , it can not be interpreted as @xmath117 oscillations .
however , in the case of @xmath123 oscillations [ panel ( b ) ] the @xmath3sr source can improve the current bounds for @xmath133 ev@xmath134 and in the region around @xmath147 ev@xmath148 .
the higher sensitivity of borexino for @xmath149 with respect the reactor experiments is mainly due to the higher statistics and the lower background .
in particular , the bound on @xmath136 for @xmath150 ev@xmath134 can be shifted to @xmath151 ( @xmath152 if the combined fit is considered ) , thus improving the existing limits by about a factor 2 .
moreover , since both @xmath2 scattering and inverse @xmath40-decay can probe the same values of @xmath136 for @xmath135ev@xmath134 , cross checks are possible .
if a deficit in the counting is found in both cases , this could be interpreted as a signal for @xmath123 transition .
finally , the @xmath1cr experiment can strongly improve the present bounds on @xmath0 transitions fixed by the calibration source experiment in gallex @xcite .
in particular , in the case of @xmath117 oscillations , the bound on @xmath140 would be lowered by a factor 10 , from @xmath153 ev@xmath134 to @xmath154 ev@xmath134 .
( the lower bound on @xmath136 is now fixed by karmen to 0.052 , stronger than the borexino one . ) in the case of @xmath123 transition , borexino would fix also the bound on @xmath136 to 0.1 two times better than gallex . if an oscillation signal were find in the @xmath0 channel but not in the @xmath2 channel this would be an evidence for cp violation .
neutrinos can interact with photons through a possible magnetic dipole moment or anapole moment ( also called charge radius ) .
the effective @xmath92-@xmath41 interaction lagrangian is @xcite @xmath155 where @xmath156 is the neutrino field , @xmath157 is the anapole moment and @xmath158 the neutrino magnetic moment .
the mere existence of a neutrino dirac mass implies an effective neutrino magnetic moment equal to @xmath159 @xcite , where @xmath160 is the bohr magneton . regarding the anapole moment
, the situation is controversial .
some authors assert that the effective anapole moment coming from the radiative corrections of the neutrino vertex in the standard model is not gauge invariant and then can not be a physical observable @xcite , while in @xcite it is claimed that a gauge invariant part can be extracted , yielding a value @xmath161 cm@xmath134 for a top quark mass of 175 gev .
conservatively one can say that values of this two e.m .
form factors larger than quoted above would be an indication for non - standard neutrino physics .
stringent bounds on the neutrino e.m .
form factors come from astrophysical arguments .
for example , if the neutrino anapole moment exceeds @xmath163
cm@xmath134 , escaping neutrinos would overcool stars and hence should modify the color - magnitude diagram of globular clusters @xcite . moreover , the energy loss in red giants in globular clusters via the plasmon decay @xmath164 mediated by neutrino magnetic moment would be too large @xcite unless @xmath165 .
a rather stringent limit comes also from the sn1987a , @xmath166 @xcite .
however , the only direct experimental constraint on the @xmath0 magnetic moment ( @xmath167 ) , comes from reactor experiments sensitive to @xmath2-@xmath168 elastic scattering @xcite . as regard the electron neutrino anapole moment @xmath169
, the more stringent limit comes from the lampf collaboration which quotes @xmath170 @xcite .
improved bounds are expected from munu @xcite . in the following
, we will also discuss this experiment in comparison with borexino . the possibility to search for a neutrino magnetic moment using an external neutrino source
was addressed by the borexino collaboration in 1991 @xcite and then studied in @xcite and @xcite . in this section
we study @xmath92-@xmath4 scattering process in the general case , i.e. , with non - zero magnetic and anapole moments . from the lagrangian in eq .
( [ emlagr ] ) , the @xmath0-@xmath4 differential cross section is obtained @xcite : @xmath171@xmath172\]]@xmath173 \ , .\label{emdiffcross}\ ] ] the number of observed events as function of the e.m .
form factors is given by : @xmath174 \ , , \label{nem}\ ] ] where @xmath73 is the standard expectation , and @xmath175 , @xmath176 , and @xmath177 are the partial cross section in eq .
( [ emdiffcross ] ) integrated on the electron recoil energy [ including the corrections due to the finite resolution of the detector , according to eq .
( [ crossec ] ) ] and folded with the source spectrum . from eqs .
( [ nem ] ) and ( [ limit ] ) we obtain the equation for 90% sensitivity bound in the plane @xmath178 for a null result : @xmath179^{1/2 } \ , , \label{emlimit}\ ] ] where @xmath180 and @xmath181 . in tab .
[ tab2 ] we report our calculation of the coefficients @xmath182 , @xmath183 , and @xmath184 for the cases of @xmath1cr and @xmath3sr source experiments , and for the munu experiment , where @xmath158 is measured in units of @xmath185 and @xmath186 in units of @xmath187 cm@xmath134 . in fig .
[ munu ] we show the 90% c.l .
contours for @xmath1cr neutrinos ( dashed lines ) and @xmath3sr antineutrinos ( dotted lines ) .
the shaded line is the limit on @xmath157 set by lampf @xcite .
if no difference with the standard expectation were found , combining this limit with the borexino measurement one can put an upper limit on @xmath158 equal to @xmath188 ( 90% c.l . for 2 d.o.f . ) for neutrinos and @xmath189 for antineutrinos .
moreover , one can put an upper limit on @xmath157 equal to @xmath190 cm@xmath134 for neutrinos and @xmath191 cm@xmath134 for antineutrinos , close to the value in @xcite .
moreover , assuming that the magnetic and anapole moment are equal for neutrinos and antineutrinos , it is possible to perform a combined fit ( gray area in fig .
[ munu ] ) . in this case , one obtains a more stringent bound on the parameters : @xmath192 @xmath193 and @xmath194 . in fig .
[ munu ] we show for comparison the zone explorable by the munu experiment after one year of operation ( solid line ) . for @xmath186
unconstrained , munu can put a limit @xmath195 . for @xmath196
, we obtain a limit @xmath197 ( 90% c.l . for 1 d.o.f . ) , have to be reduced by a factor 0.767 . ] in relatively good agreement with the munu collaboration analysis @xcite . from fig .
[ munu ] we see that both the @xmath1cr and the @xmath3sr limits are more stringent then those expected in munu as a result of higher statistics , of the smaller flux uncertainties , and of the lower energy threshold of borexino . in particular , the @xmath3sr @xmath111 experiment is the most sensitive . for @xmath196 ,
the @xmath3sr limit on @xmath158 is @xmath198 ( 90% c.l . for 1 d.o.f . ) , about three times better than munu .
this limit depends weakly on assumptions about the source activity and exposure time .
for example , with 2.5 mci activity and 3 months of exposure , this limit is raised to @xmath199 about a factor two better than munu .
in fact , for @xmath200 , the limit value of @xmath158 depends on the square root of @xmath110 [ see eq .
( [ emlimit ] ) ] . this make this experiment very appropriate for magnetic moment searches .
the low energy @xmath92-@xmath4 neutral current interaction is usually parameterized by the following effective four - fermion hamiltonian : @xmath201\left[\bar{\psi}_{e}\gamma_{\alpha } \left(g^{\nu e}_{v}-g^{\nu e}_{a}\gamma^{5}\right)\psi_{e}\right ] \ , , \label{hint}\ ] ] where @xmath156 and @xmath202 are the neutrino and electron fields and @xmath203 are the vector and axial coupling of the neutrino current to the electron current .
when also charge current interactions are involved , as in the case of @xmath0-@xmath4 scattering , @xmath204 .
the standard model of electroweak interactions states that @xmath205 ) and @xmath206 , apart from small radiative corrections .
moreover , in the standard model @xmath207 , where @xmath208 ( @xmath209 ) are the couplings of the neutrinos ( electrons ) to the @xmath32 boson .
the values of @xmath208 are inferred from the `` invisible '' decay width of the @xmath32 boson @xcite .
although this gives the value of @xmath210 with great precision , it does not account for possible non - standard process occurring in the @xmath92-@xmath4 scattering .
( for example , exchange of non - standard neutral gauge boson , as proposed in @xcite ) . for this reason ,
a direct measure of the @xmath203 couplings is interesting . at present ,
the most precise _ direct _ determinations of @xmath211 , come from the charm ii experiment using @xmath51-@xmath4 scattering @xcite : @xmath212 and @xmath213 at @xmath100 , in agreement with standard model . in this section
, we investigate the possibility to probe @xmath203 using the borexino calibration experiments . from eq .
( [ hint ] ) , one obtains the differential cross section for @xmath0-@xmath4 scattering @xcite in the form of eq .
( [ diffcross ] ) with @xmath214 where @xmath203 are now independent variables . expanding eq .
( [ diffcross ] ) in term of @xmath33 and @xmath215 , and following the same procedure of the previous section we obtain @xmath216 where @xmath73 is the standard expectation and @xmath217 here @xmath218 , @xmath219 , and @xmath220 [ @xmath221 have opposite sign for @xmath111 ] . in tab .
[ tab3 ] we show the value of the coefficients @xmath223 , @xmath224 , and @xmath221 for the cases of @xmath1cr and @xmath3sr source experiments , and for the munu experiment . from eq .
( [ limit ] ) we then obtain the 90% limit in the plane @xmath225 for a null result : @xmath226 this limit is shown in fig .
[ gvga ] for the @xmath1cr experiment ( dashed line ) , the @xmath3sr experiment ( dotted line ) , and their combination ( gray area ) . also shown
are the charm ii results and the zone explorable by munu ( solid line ) . in borexino ,
a significative improvement in the measure of @xmath33 appears possible , whilst the constraints on @xmath215 are similar to charm ii . in particular , we obtain @xmath227 ( @xmath228 ) and @xmath229 at the 90% c.l .
( 2 d.o.f . ) .
the @xmath230 is thus measured with a precision of @xmath231% a factor two better than charm ii .
fixing @xmath215 to @xmath232 , we obtain a more stringent constraint on the weinberg angle : @xmath233 ( 90% c.l . for 1 d.o.f .
) , corresponding to a precision of @xmath234% . from fig .
[ gvga ] we can see that the borexino constraints are more stringent than munu as a result of higher statistics attainable and of the combination of @xmath92 and @xmath111 signal .
finally , we stress that a comparison of scattering experiment of the kind @xmath0-@xmath4 ( borexino ) and @xmath51-@xmath4 ( charm ii ) is a useful check of the universality of weak interactions at low energies .
in this paper we have explored the possibility to search for non - standard neutrino properties with the borexino cr and sr source experiments . in particular , we have considered a ) neutrino oscillations ; b ) non - zero electron neutrino e.m .
form factors @xmath158 and @xmath235 ; c ) non - standard @xmath92-@xmath4 vector and axial couplings . in case
a ) we find that , in the channel @xmath123 , borexino can extend the oscillation parameter limits for @xmath150 ev@xmath134 and @xmath236 . in case
b ) borexino can reach a sensitivity to the magnetic moment equal to @xmath188 for neutrinos and @xmath189 for antineutrinos . in additions , assuming that the e.m .
form factors are equal for @xmath92 and @xmath111 , this limit can be improved ( @xmath237 ) and a limit of @xmath238 can be put on the anapole moment the strongest limit at present . in the hypothesis that @xmath239 , the @xmath3sr experiment alone can put a limit @xmath240 , improving the munu sensitivity by a factor three . in case c ) , borexino can reduce the present uncertainty on the _ direct _ measure of the @xmath33 coupling by a factor 2 , and can check for the universality of the @xmath92-@xmath4 interactions at low energies . fixing @xmath215 to @xmath241 , the weinberg angle @xmath242 can be measured with an accuracy of @xmath243% .
we thank e. lisi for useful suggestions and for careful reading of the manuscript , and j.n .
bahcall , v.v .
sinev , and j.w.f .
valle for suggestions .
the work of a.i . and g.s .
was supported in part by ministero della ricerca scientifica ( dottorato di ricerca ) and in part by infn .
the work of d.m . was supported by infn .
in this appendix we give the analytical expression for the function @xmath127 of eq .
( [ gdraft ] ) . in polar coordinates
( choosing the @xmath244 axis as the fv center - to - source direction ) , eq .
( [ gdraft ] ) reads @xmath245 with @xmath246 . with the substitutions @xmath247 and @xmath248 ( with @xmath249 ) one has @xmath250 \ , , \label{gdraft2}\end{aligned}\ ] ] where @xmath251 and the function @xmath252 is the integral cosine : @xmath253 the integral in eq .
( [ gdraft2 ] ) can be easily evaluated with the help of the following expressions : @xmath254 the final result can be cast in the following form : @xmath255 where @xmath256 \left(z-2\beta\right)-\cos z \ , .\label{g}\ ] ] borexino collaboration , g. alimonti _ et al .
_ , _ a real time detector for low energy solar neutrinos _ , proposal to the national science foundation , edited by f.p .
( princeton , new jersey , 1996 ) . borexino collaboration , c. arpesella _ et al .
_ , infn borexino proposal , vols . 1 and 2 , edited by g. bellini , r. raghavan , _ et al . _ ( university of milano , 1992 ) ; j. bezinger , f.p .
et al . _ ,
proposal for borexino to national science foundation ( princeton , 1992 ) ; j. bezinger _
et al . _ , _ a proposal for participation in the borexino solar neutrino experiment _ , princeton university report , 1996 ( unpublished ) .
miranda , v. semikoz , and j.w.f .
valle , phys .
d * 58 * 013007 ( 1998 ) ; also in _ proceedings of the international symposium on lepton and baryon number violation , trento , 1998 _ , to appear , hep - ph/9808395 .
s. schoenert , in _ proceedings of taup97 , v international workshop on topics in astroparticle and underground physics , gran sasso laboratory , 1997 _ , edited by a. bottino , a. di credico , and p. monacelli , nucl .
b proc .
suppl . * 70 * , 195 ( 1999 ) .
bardeen , r. gastmans , and b. lautrup , nucl .
b * 46 * , 319 ( 1972 ) ; n.m .
monyonko and j.h .
reid , prog .
* 73 * , 734 ( 1985 ) ; j.l .
lucio , a. rosado , and a. zepeda , phys .
d * 39 * , 1091 ( 1985 ) ; m.j .
musolf and b.r .
holstein , phys .
d * 43 * , 2965 ( 1991 ) .
lattimer and j. coopersteim , phys .
lett . * 61 * , 23 ( 1988 ) ; a. barbieri and r.n .
mohapatra , phys .
* 61 * , 27 ( 1988 ) ; d. ntzold , phys .
d * 38 * , 1658 ( 1988 ) ; a. goyal , s. dutta , and s.r .
choudhury , phys .
b * 49 * , 312 ( 1995 ) .
.[tab1 ] time of exposure ( @xmath72 ) , expected background events ( @xmath257 ) , standard signal events ( @xmath73 ) , @xmath100 uncertainty of @xmath73 ( @xmath258 ) , and 90% c.l .
( 2 d.o.f . )
relative accuracy for the @xmath92-@xmath4 scattering , @xmath111-@xmath4 scattering , inverse @xmath40-decay , and munu experiments . [ cols="<,^,^,>,>,^ " , ] fig . 2 .
prospective 90% c.l .
sensitivity contours in the oscillation parameters plane for the borexino source experiments in the case of @xmath259 and @xmath123 transitions [ panel ( a ) and ( b ) , respectively ] .
short - dashed line : @xmath92-@xmath4 scattering of the @xmath3sr antineutrinos ; long - dashed line : @xmath92-@xmath4 scattering of the @xmath1cr neutrinos ; dotted line : inverse @xmath40-decay ; gray area : combined sr+cr experiments .
the 90% c.l .
limits coming from negative evidences of oscillations ( solid thick line ) and the lsnd allowed area ( thin solid line ) are also shown .
prospective 90% c.l .
sensitivity contours in the neutrino e.m .
form factors plane for the borexino source experiments . dashed line : @xmath1cr neutrinos ; dotted line : @xmath3sr antineutrinos
; gray area : combined sr+cr experiments .
the 90% c.l .
lampf limit on @xmath260 ( shaded line ) and the expected 90% c.l .
munu limit ( solid line ) are also shown .
prospective 90% c.l .
sensitivity contours in the @xmath92-@xmath4 vector and axial coupling plane for the borexino source experiments . dashed line : @xmath1cr neutrinos ; dotted line : @xmath3sr antineutrinos ;
gray area : combined sr+cr experiments .
the 90% c.l .
data from charm ii data and the expected 90% c.l .
munu limit ( solid line ) are also shown . | we calculate the event rates induced by high - intensity radioactive sources of @xmath0 ( @xmath1cr ) and of @xmath2 ( @xmath3sr ) , to be located near the borexino detector .
calculations are performed both in the standard case and assuming non - standard properties of neutrinos , including flavor oscillations , neutrino electromagnetic interactions , and deviations from the standard vector and axial couplings in the @xmath0-@xmath4 interaction .
it is shown that , in some cases , the current limits on non - standard neutrino properties can be significantly improved . |
a question regarding an isoperimetric problem about simple polygons was asked by brass , moser and pach ( see problem 3 on p. 437 in @xcite ) . for @xmath1 odd ,
what is the maximum perimeter of a simple @xmath0-gon contained in a euclidean unit disk ?
the authors of @xcite remarked that for @xmath0 even , the supremum of the perimeters is the trivial upper bound @xmath3 , as it can be approached by simple @xmath0-gons in which the vertices alternate between some small neighborhoods of two antipodal points of the disk . this argument can not be applied if @xmath0 is odd . in 2009 , audet , hansen and messine @xcite showed that for @xmath0 odd , the supremum is attained as the perimeter of an isosceles triangle inscribed in the disk , with an edge of multiplicity @xmath2 .
the author of @xcite gave a shorter proof of the same statement , and proved that for hyperbolic disks of any radius , the supremum is attained as the perimeter of an @xmath0-gon of the same kind ; that is , that of an isosceles triangle with a multiple edge inscribed in the disk .
he noted that for @xmath0 even and for any plane convex body @xmath4 in the euclidean plane @xmath5 or in the hyperbolic plane @xmath6 , the supremum of the perimeters of the simple @xmath0-gons contained in @xmath4 is the trivial bound @xmath7 , where @xmath8 is the diameter of @xmath4 .
he asked whether it is true that for @xmath0 odd , the supremum is the perimeter of a triangle with an edge of multiplicity @xmath2 , inscribed in @xmath4 .
in this paper we answer this question .
our main result is the following .
[ thm : oval ] let @xmath9 be an odd integer , and let @xmath4 be a plane convex body in @xmath5 or in @xmath6 . for every simple @xmath0-gon @xmath10 contained in @xmath4
there is a triangle , inscribed in @xmath4 and with side - lengths @xmath11 , such that @xmath12 . in the proof we use the following notations .
let @xmath13 and @xmath14 .
the distance of @xmath15 and @xmath16 is denoted by @xmath17 .
the closed ( respectively , open ) segment with endpoints @xmath15 and @xmath16 is denoted by @xmath18 $ ] ( respectively , @xmath19 ) . if @xmath20 , @xmath21 denotes the straight line passing through @xmath15 and @xmath16 , and @xmath22 denotes the closed ray in @xmath21 emanating from @xmath15 and not containing @xmath16 . for any set @xmath23 , we use the standard notations @xmath24 , @xmath25 , @xmath26 , @xmath27 , @xmath28 and @xmath29 for the _ interior _ , the _ boundary _ , the _ diameter _ , the _ perimeter _ , the _ area _ , or the _ convex hull _ of @xmath30 .
points are denoted by small latin letters , and sets of points by capital latin letters . in the proof
we find a triangle @xmath31 , _ contained _ in @xmath4 , with side - lengths @xmath32 and @xmath33 such that @xmath12 , as in this case we can move the vertices of @xmath31 to @xmath34 in a way that no side - length of @xmath31 decreases .
let us consider a descartes coordinate system .
if @xmath35 is an arbitrary point , by @xmath36 we mean that the @xmath15-coordinate of @xmath37 is @xmath38 , and its @xmath16-coordinate is @xmath39 .
let @xmath40 , [ a_1,a_2 ] , \ldots , [ a_{n-1},a_n]$ ] denote the edges of @xmath10 such that @xmath41 , and let @xmath42 for @xmath43 . without loss of generality , we may assume that @xmath40 $ ] is a longest edge of @xmath10 , @xmath44 is the origin @xmath45 , and that @xmath46 . for @xmath47 ,
let @xmath48 .
note that @xmath49 . as @xmath0 is odd , the sequence @xmath50 consists of an even number of elements .
thus , it has two consecutive elements , say @xmath51 and @xmath52 , that are both nonnegative or nonpositive . from this
, we have that @xmath53 , or that @xmath54 , respectively . for simplicity , we denote @xmath44 , @xmath55 , @xmath56 , @xmath57 and @xmath58 by @xmath59 , @xmath60 , @xmath61 , @xmath62 and @xmath63 , respectively , and set @xmath64 and @xmath65 . if @xmath66 , we have @xmath67 .
thus , in the following we may assume that @xmath68 and also that @xmath69 .
assume that @xmath70 \cap r_q(p , q ) \neq \emptyset$ ] .
from this , it follows that @xmath71 and that @xmath72 ( cf .
figure [ fig : figure1 ] ) .
thus , we have that @xmath73 . if @xmath74 \cap r_p[p , q ] \neq \emptyset$ ] , we may apply a similar argument .
assume that @xmath70 \cap r_p(p , q ) \neq \emptyset$ ] , which yields that @xmath75 .
if @xmath76 , then we may apply the argument in the previous paragraph , and thus , we may assume that @xmath77 . from this and from @xmath68 , we readily obtain that @xmath78 .
let @xmath79 denote the bisector of the segment @xmath80 $ ] . since @xmath81 , we have that @xmath79 separates @xmath82 from @xmath83 and @xmath84 .
observe also that if @xmath85 , then @xmath86 , and hence , we may assume that @xmath79 separates @xmath87 and @xmath82 from @xmath84 .
thus , @xmath70 \cap l \neq \emptyset \neq [ b , c ] \cap l$ ] , which implies that @xmath88 and @xmath89 . from this
, we obtain that @xmath90 .
if @xmath74 \cap r_q(p , q ) \neq \emptyset$ ] , the assertion follows by a similar argument .
we are left with the case that @xmath91 are in the same closed half - plane bounded by @xmath92 .
let @xmath93 .
if @xmath75 and @xmath94 , then @xmath66 and hence , we may assume that , say , @xmath95 .
assume that @xmath96 , and that , say , @xmath97 .
if @xmath98 , then @xmath99 and @xmath100 , which yields that @xmath101 .
if @xmath102 , then @xmath103 , from which it readily follows that @xmath104 .
assume that @xmath105 and that @xmath106 .
then the three rays , emanating from @xmath83 , that pass through @xmath107 , @xmath84 and @xmath87 are in this clockwise order around @xmath83 .
let @xmath108 denote the bisector of the segment @xmath109 $ ] .
note that as @xmath110 , @xmath108 separates @xmath111 $ ] from @xmath107 .
hence , it follows from @xmath112 that @xmath108 separates @xmath70 $ ] from @xmath107 ( cf .
figure [ fig : figure2 ] ) .
then we obtain that @xmath113 , and hence , that @xmath114 .
our last case is that @xmath105 and @xmath115 .
we may assume that @xmath116 , as otherwise the assertion clearly follows .
if @xmath117 , then @xmath118 , @xmath119 and @xmath120 .
thus , we may assume that @xmath121 , from which we obtain that @xmath122 .
assume that @xmath123 .
then @xmath124 , which yields that @xmath97 .
thus , the two legs of the right triangle @xmath125 are pairwise greater than or equal to the two legs of @xmath126 , from which we obtain that @xmath127 and that @xmath128 .
hence , we may assume that @xmath129 .
assume that @xmath130 , or in other words , that @xmath131 .
then @xmath132 , where @xmath133 and @xmath134 .
note that for any @xmath135 , the function @xmath136 is a convex function on @xmath137 .
thus , @xmath138 @xmath139 and the assertion readily follows .
finally , assume that @xmath140 , which immediately yields that @xmath141 .
if @xmath142 , then the remaining two edges of @xmath10 are @xmath109 $ ] and @xmath143 $ ] , and we obtain that @xmath144 . furthermore , if @xmath145 , then @xmath146 .
hence , we may assume that @xmath147 and that @xmath148 , which yields that @xmath149 .
set @xmath150 . since @xmath147
, it suffices to prove that @xmath151 ; that is , that @xmath152 let @xmath153 and @xmath154 denote the left - hand side and the right - hand side of ( [ eq : crucialeu ] ) , respectively , and let us regard them as functions of @xmath84 .
consider the vector @xmath155 .
note that @xmath156 , and thus , that ( [ eq : crucialeu ] ) holds for @xmath157 .
thus , we need only prove that in the direction of @xmath158 , the derivative of @xmath154 is not smaller than that of @xmath153 ; that is , using the standard differential geometric notation , that @xmath159 .
let @xmath160 and @xmath161 denote the internal angles at @xmath84 of @xmath162 , @xmath163 and @xmath164 , respectively ( cf . figure [ fig : figure3 ] ) .
we observe that @xmath165 and that @xmath166 .
note that @xmath167 , and @xmath168 .
we set @xmath169
. then an elementary calculation yields that @xmath170 @xmath171
in this section we show how to modify the argument in section [ sec : euclidean ] to obtain the proof for polygons in the hyperbolic plane .
let @xmath40 , \ldots , [ a_{n-1},a_n]$ ] be the edges of @xmath10 , where @xmath172 .
we use our notation in a way that @xmath40 $ ] is a longest edge of @xmath10 and , like in the previous section , we write @xmath173 and @xmath174 .
similarly like in section [ sec : euclidean ] , we obtain that @xmath10 has three consecutive vertices , which we denote by @xmath175 and @xmath84 , with orthogonal projections @xmath176 , @xmath177 and @xmath178 onto @xmath92 , respectively , such that @xmath179 $ ] . without loss of generality
, we may assume that , orienting the points of @xmath92 in a way that @xmath107 precedes @xmath82 , we have that @xmath83 precedes @xmath84 .
we set @xmath180 , and denote the lines , orthogonal to @xmath92 , that pass through @xmath107 and @xmath82 by @xmath181 and @xmath182 , respectively .
furthermore , we denote the closed infinite strip bounded by @xmath181 and @xmath182 by @xmath183 , and the one bounded by @xmath184 and @xmath185 by @xmath186 .
we may assume that @xmath187 and that @xmath188 and @xmath84 are in the same closed half - plane bounded by @xmath92 , since otherwise we may apply an argument similar to the one in section [ sec : euclidean ] .
thus , we obtain that one of the components of @xmath189 , say the one containing @xmath107 in its boundary , is disjoint from @xmath190 . _ case 1 _ , @xmath191 .
+ note that it yields that @xmath192 .
first , assume that @xmath193 .
without loss of generality , let @xmath194 .
then the assertion readily follows from @xmath195 .
thus , we may assume that @xmath196 let @xmath197 denote the closure of the component of @xmath198 $ ] that does not contain @xmath199 $ ] , and observe that @xmath200 .
we recall that for every hypercycle @xmath201 , there is a unique hyperbolic line such that the distances of the points of @xmath202 from @xmath79 are equal .
we call this line the _ reference line _ of @xmath202
. let @xmath203 denote the hypercycle , orthogonal to @xmath92 , that passes through @xmath107 and @xmath83 . in this case
the reference line @xmath204 of @xmath203 is also orthogonal to @xmath92 .
let @xmath205 denote the hypercycle , with the reference line @xmath204 , that passes through @xmath84 .
we observe that @xmath205 is also orthogonal to @xmath92 .
figure [ fig : figure4 ] shows these hypercycles in the conformal disk model such that @xmath107 is the centre of the model .
we note that as @xmath187 , we have that @xmath206 \neq \emptyset$ ] .
let @xmath207 and @xmath208 denote the line , orthogonal to both @xmath203 and @xmath205 , that passes through @xmath83 and @xmath84 , respectively .
observe that @xmath92 , @xmath207 and @xmath208 are pairwise disjoint .
since two hyperbolic lines intersect at most once , we obtain that the point @xmath200 and the segment @xmath199 $ ] are separated by @xmath207 or by @xmath208 .
we show that in the first case we have @xmath113 , and in the second case @xmath209 .
first , assume that @xmath87 and @xmath199 $ ] are separated by @xmath207 .
then the angle of @xmath70 $ ] and the arc of @xmath203 between @xmath83 and @xmath107 is not acute .
note that if @xmath210 is a point and @xmath201 is a hypercycle , and we move a point @xmath211 on @xmath202 farther from the orthogonal projection of @xmath15 onto @xmath202 , then the quantity @xmath17 does not decrease . thus , it follows that @xmath113 .
now we assume that @xmath87 and @xmath199 $ ] are separated by @xmath208 .
then , denoting the intersection point of @xmath205 and @xmath212 $ ] by @xmath213 , we have that the angle of @xmath74 $ ] and the arc of @xmath205 between @xmath84 and @xmath213 is not acute , and that @xmath214 .
if @xmath215 $ ] , then @xmath216 $ ] , and clearly , @xmath209 .
if @xmath217 $ ] , then the angle between @xmath74 $ ] and @xmath218 $ ] is not acute , which yields that @xmath100 ( cf .
figure [ fig : figure5 ] ) . we have obtained that @xmath113 or that @xmath209 . without loss of generality ,
let @xmath113 .
if @xmath219 , then @xmath114 , and hence , we may assume that @xmath220 .
let @xmath221 denote the line bisecting the segment @xmath143 $ ] . if @xmath222 , then @xmath223 , and the assertion readily follows .
if @xmath224 , then @xmath221 separates @xmath87 and @xmath82 , and thus , the assertion follows from @xmath209 . _ case 2 _ , @xmath225 .
+ then @xmath226 .
note that the rest of the proof of the euclidean case up to ( [ eq : crucialeu ] ) can be easily modified for the hyperbolic case , and hence we omit it .
we prove only that ( [ eq : crucialeu ] ) holds also for points in the hyperbolic plane .
let @xmath227 denote the line bisecting the segment @xmath228 $ ] , and observe that @xmath229 \neq \emptyset$ ] .
let @xmath15 and @xmath16 denote the intersection point of @xmath79 with @xmath228 $ ] and @xmath230 $ ] , respectively .
observe that @xmath231 , and thus , the assertion holds for @xmath232 .
we introduce @xmath153 , @xmath154 , @xmath233 , @xmath234 and @xmath161 as in the euclidean case .
note that the angle of @xmath235 at @xmath16 is acute , and thus @xmath165 and @xmath166 .
let @xmath158 denote the unit vector at @xmath84 ( an element of the tangent plane @xmath236 ) tangent to @xmath185 and pointing away from @xmath178 . then , as in the euclidean case
, we need only prove that @xmath159 . observe that @xmath167 , and @xmath168 . as in section [ sec : euclidean ] , we set @xmath169 . set @xmath237 and @xmath238 . by the hyperbolic pythagorean and cosine theorems , we obtain that @xmath239 note that @xmath141 and the function @xmath240 , where @xmath241 and @xmath242 , is strictly decreasing .
thus , @xmath243 now the inequality @xmath244 follows from the estimate @xmath245 and by algebraic transformations similar to those used in the euclidean case . c. audet , p. hansen and f. messine , _ simple polygons of maximum perimeter contained in a unit disk _ discrete comput
* 41 * ( 2009 ) , 208 - 215 .
p. brass , w. moser and j. pach , _ research problems in discrete geometry _ , springer , new york , 2005 .
z. lngi , _ on the perimeters of simple polygons contained in a disk _ monatsh . math .
, accepted , arxiv:1002.3922v1 [ math.gm ] february 20 , 2010 . | a simple @xmath0-gon is a polygon with @xmath0 edges such that each vertex belongs to exactly two edges and every other point belongs to at most one edge .
brass , moser and pach @xcite asked the following question : for @xmath1 odd , what is the maximum perimeter of a simple @xmath0-gon contained in a euclidean unit disk ?
in 2009 , audet , hansen and messine @xcite answered this question , and showed that the supremum is the perimeter of an isosceles triangle inscribed in the disk , with an edge of multiplicity @xmath2 . in @xcite , lngi generalized their result for polygons contained in a hyperbolic disk . in this note
we find the supremum of the perimeters of simple @xmath0-gons contained in an arbitrary plane convex body in the euclidean or in the hyperbolic plane . |
that there is a strong connection between bipartite entanglement and subsystem fluctuations is well - known .
previous studies on bipartite ground states of critical spin chains revealed that subsystem fluctuations in a conserved charge have the same ( logarithmic ) scaling with subsystem size as the entanglement entropy @xcite .
further investigations on one - dimensional free fermion systems have shown that rnyi entanglement entropies can be expressed in terms of the cumulants of the fermion number distribution within a partition @xcite .
additionally , the link between entanglement and fluctuations has been studied in the context of cosmological models @xcite and topological invariants @xcite .
simple qualitative arguments can provide insight into the deep relationship between fluctuations and entanglement .
when a pure state is partitioned into two entangled subsystems , the state of either can not be specified by a single state vector .
the complete description of a subsystem ( say @xmath1 ) requires a reduced density operator @xmath2 that is mixed @xcite . the statistical uncertainty in determining the state of this partition manifests as a non - zero entanglement entropy , as well as non - zero fluctuations in subsystem observables that originate from inter - partition quantum correlations . in fact
, the range of these correlations has been demonstrated to be a crucial factor in determining whether an entanglement area law will hold or not @xcite .
entanglement in a pure state @xmath3 that is partitioned into a subsystem @xmath1 and bath @xmath4 is completely characterized by the entanglement spectrum @xcite that consists of the eigenvalues of the reduced density operator @xmath5 .
due to normalization and the hermiticity of @xmath2 , this spectrum has @xmath6 independent real parameters when @xmath2 is of rank @xmath0 .
thus , it is reasonable to assume that a set of @xmath6 independent fluctuation measurements can fully characterize entanglement in such a bipartitioned state .
for example , take the case of two entangled qubits in an overall pure state .
the entanglement spectrum is @xmath7 so that entanglement measures @xcite , such as the von neumann entanglement entropy @xmath8 , purity @xmath9 , and concurrence @xmath10 @xcite , depend on a single real parameter @xmath11 . for qubits ,
it is well - established that these entanglement measures are related to a subsystem fluctuation or variance .
for instance , if we define an entanglement hamiltonian as @xmath12 , a direct calculation leads to a fluctuation - entanglement relation of the form @xmath13 @xcite . in this paper
we expand the notion of fluctuation - entanglement relations to the case of an arbitrary pure state partitioned into a @xmath14-state qudit and bath in such a way that the qudit is described by a reduced density operator of rank @xmath15 .
our main result is the identification of a set of @xmath6 operators that have fluctuations that collectively measure entanglement through a fluctuation - entanglement relation .
we begin with a brief introduction to some notation and terminology , especially on the bloch representation of reduced density operators @xmath2 as a linear combination of @xmath16(@xmath0 ) generators @xmath17 and the identity operator ( section [ generators ] ) .
we then express the variance of an @xmath6 component subsystem observable @xmath18 @xmath19 in terms of known entanglement measures like the linear entropy @xmath20 @xcite and a generalized concurrence @xmath21 @xcite ( section [ schmidtpolfluc ] ) : @xmath22 here we refer to the total fluctuation @xmath23 as the schmidt polarization fluctuation .
this exact fluctuation - entanglement relation is true for any qudit - bath pure state and provides a particularly clear and simple mathematical basis for the intuitive link between entanglement and subsystem fluctuations , especially in a many - body context .
furthermore , this exact relation allows ( i ) recent experimental measurements of purity in many - body systems @xcite to be interpreted as direct measures of subsystem fluctuations and ( ii ) noise measurements ( as in spin noise experiments @xcite ) to be interpreted as direct entanglement measurements .
we provide examples for cases when the subsystem partition is further decomposable into a product state ( section [ sect : comp ] ) and when it is expressible in matrix product form ( section [ sect : aklt ] ) .
a set of @xmath25 @xmath24 generators that generalizes the @xmath26 pauli operators contains the following elements : @xmath27,\end{aligned}\ ] ] for @xmath28 and @xmath29 and @xmath30 is an orthonormal set of @xmath0 state vectors that span a partition subspace @xcite . for notational convenience ,
these generators are often labeled as @xmath31 .
these @xmath25 operators are traceless and satisfy @xmath32 .
furthermore , they are hermitian and therefore represent observables on the corresponding partition . any rank-@xmath0 reduced density operator @xmath33 can be written as a linear combination of the @xmath24 generators @xmath34 and the @xmath0-dimensional identity @xmath35 : @xmath36 the constants @xmath37 form the components of an @xmath38 dimensional bloch vector ( or coherence vector ) @xmath39 @xcite .
the components of this bloch vector are @xmath40 which are expectation values of observables and thus experimentally accessible .
fluctuations in these observables have been used to construct a fluctuation - entanglement relation under the framework of generalized entanglement relative to the distinguished subset of observables @xmath17 @xcite . in this framework
, one defines a vector @xmath41 so that the total component - wise fluctuation in @xmath42 is @xmath43 each term in this expression is readily evaluated .
the first term is the expectation value of the quadratic casimir invariant @xmath44 .
the second term is the square of the bloch vector , which can be expressed in terms of the purity @xmath45 according to @xmath46 @xcite .
we thus obtain a fluctuation - entanglement relation @xmath47 calculating the fluctuation term in the left - hand side of this relation requires @xmath25 measurements .
it contains the same information as the @xmath25 independent real parameters @xmath48 needed to completely specify the reduced density operator @xmath2 .
the main goal now in the remainder of this paper is to identify a set of @xmath6 fluctuation measurements that is equivalent to the set of @xmath6 independent eigenvalues of @xmath2 .
that is , we seek a set of fluctuations that corresponds to complete knowledge of the independent elements of the entanglement spectrum @xmath49 @xcite , thereby quantifying entanglement in an arbitrary pure state .
achieving this goal requires a set of @xmath6 compatible observables that commute with the reduced density operator @xmath2 .
a particularly convenient set is formed by the @xmath6 diagonal generators @xmath50 that are constructed from the schmidt basis . in the following analysis , we find that the fluctuations in the non - diagonal generators @xmath51 and @xmath52 sum trivially in this basis .
let us consider a state vector @xmath53 that is partitioned into subsystems @xmath1 and @xmath4 with schmidt rank @xmath0 : @xmath54 the set of state vectors @xmath55 form an orthonormal basis in the hilbert subspace of partition @xmath56 .
hence , the reduced density operator for subsystem @xmath1 is diagonal in this schmidt basis : @xmath57 here the partial trace @xmath58 is performed over the @xmath59 states . for brevity
, we will often omit the partition label subscripts on the state vectors as the reduced density operator is understood to act on the hilbert subspace of a given partition . in the following discussion
we define the schmidt polarization fluctuation and obtain fluctuation - entanglement relations in terms of known entanglement measures .
we begin with the simplest case of a qubit entangled with a bath before proceeding with the general qudit case . in schmidt
decomposed form , the full density operator of a qubit @xmath1 entangled with an arbitrary environment @xmath4 is @xmath60 where @xmath61 .
the reduced density operator for the qubit in the schmidt basis @xmath62 is therefore @xmath63 it turns out that we can construct an entanglement measure that depends only on the fluctuation in the polarization of the qubit with respect to the schmidt states @xmath64 and @xmath65 .
let this polarization be represented by the diagonal @xmath26 generator @xmath66 , which corresponds to the pauli @xmath67-matrix in the schmidt basis .
we therefore have @xmath68 and @xmath69 measures the relative probability of finding the qubit in either of its schmidt states : it is equal to @xmath70 if the qubit is certainly in state @xmath64 , @xmath71 if certainly in state @xmath65 , and zero when it is in either state with equal probability .
let us define the corresponding schmidt polarization fluctuation as @xmath72 it is evident that this fluctuation is a proper entanglement measure because it is a monotonic function of the qubit concurrence @xmath21 and , upon rearrangement , the linear entropy @xmath20 : @xmath73= 2(1-\gamma ) = 2s_l .\ ] ] this simple result now sets the stage for a more general fluctuation - entanglement relation that applies to an @xmath0-level subsystem .
the main challenge addressed in this paper is to generalize the previously derived relationship between the schmidt polarization fluctuation and entanglement measures to the case of an entangled @xmath14-state qudit . to this end
, we write the qudit ( @xmath1 ) reduced density operator in terms of the schmidt basis : @xmath74 where @xmath15 .
that is , the dimension of the hilbert subspace spanned by the schmidt vectors is at most equal to the dimension of the full hilbert space of the partition .
we now define @xmath6 polarization observables in terms of the diagonal @xmath24 generators so that @xmath75 the generalized polarization expectation values are @xmath76 .\end{aligned}\ ] ] physically , the quantity @xmath77 may be interpreted as a weighted relative probability of finding the qudit in one of the @xmath78 schmidt states @xmath30 against finding it in the state @xmath79 . furthermore , the polarization operators may be used to define a polarization vector @xmath18 , where @xmath80 is a set of @xmath6 orthonormal vectors .
this construction allows the sum of polarization fluctuations to be written as @xmath81 we propose that the generalization of the qubit schmidt polarization fluctuation @xmath82 in the case of a rank @xmath0 subsystem is the variance in the polarization vector : @xmath83 . \label{wqudit}\ ] ] this sum involves fluctuations in the @xmath6 diagonal generators @xmath50 , which may be measured through appropriately generalized stern
gerlach - type experiments . to prove that the schmidt polarization fluctuation @xmath23 is indeed a proper entanglement measure for the qudit - bath system ,
we demonstrate that it is a monotonic function of known entanglement measures .
first , the sum of the mean squares of the polarizations is @xmath84 the sum of squares of the average polarizations takes some more effort to evaluate : @xmath85.\end{gathered}\ ] ] however , changing the summation order leads to simpler sums and we get @xmath86 the latter term is simplified by the normalization condition @xmath87 , which yields @xcite @xmath88 combining these results finally gives the desired fluctuation - entanglement relation between the schmidt polarization fluctuation and known entanglement measures : @xmath89 here , @xmath21 is a generalized concurrence that has been reported in ref .
when the pure state is maximally entangled , the total schmidt polarization fluctuation is equal to @xmath90 .
in the many - body context it may occur that the reduced density operator can be easily simplified into a product : @xmath91 it is therefore important to determine how the schmidt polarization fluctuations of independent subsystems relate to the total entanglement in composite subsystems .
given that the schmidt polarization fluctuation of both components are known , @xmath92 we are able to construct the fluctuation - entanglement relation for the composite system .
the total purity of @xmath2 is @xmath93 because purity is multiplicative over independent subsystems .
the relationship between the total purity of the subsystem and the component schmidt polarization fluctuations is therefore @xmath94 since @xmath95 , the polarization fluctuations are only approximately additive when higher order products of fluctuations are negligible . in a decomposition with @xmath96 qudits , @xmath97 ,
the fluctuation - entanglement relation becomes @xmath98.\ ] ] an important situation where such a decomposition is analytically tractable is encountered in the case of non - interacting spinless fermions hopping on a one - dimensional lattice @xcite . in the pure ground state of this system , the reduced density operator corresponding to a block of @xmath96 adjacent sites can be written as a product of @xmath96 qubit density matrices .
thus , the identity generalizes the determinant formulas for the entanglement entropy of free fermions @xcite that have been evaluated by the fisher hartwig relation @xcite . as an illustration
, we give the results of calculations of the block linear entropy @xmath99 in such a chain of fermions for several fermion densities @xmath100 in figure [ fig : freefermions ] . in this graph
the component schmidt polarization fluctuations are computed from @xmath101,\ ] ] where @xmath102 are the eigenvalues of the @xmath103 matrix of fermion correlation functions @xmath104 with indices @xmath105 restricted to the block . as previously reported , entanglement is greatest at half - filling @xmath106 and particle - hole symmetry leads to the linear entropy being equal for @xmath107 and @xmath108 .
future work on the @xmath96 qudit formula can extend this result to the case of multi - component atomic systems .
studies of entanglement in quantum spin chains have proved useful for the description of wires that can transmit quantum information between registers @xcite as well as for providing tractable theoretical models for investigating the relationship between entanglement and quantum critical phenomena @xcite . as an example , we consider the valence - bond solid ( vbs ) ground state of the spin-1 affleck - kennedy - lieb - tasaki ( aklt ) model @xcite : @xmath109 the vbs ground state is separated from excited states by a haldane gap @xcite and the entanglement entropy saturates for long blocks .
we impose periodic boundary conditions on the spin-1 operators @xmath110 and take the thermodynamic limit @xmath111 .
the reduced density operator @xmath112 for a contiguous block of @xmath113 spins can be calculated from the matrix product representation of the vbs state @xcite .
although the full hilbert space for the block is @xmath114-dimensional , the reduced density operator is at most rank-4 .
its eigenvalues are @xmath115 the fluctuation in the schmidt polarization vector is @xmath116 where the relevant polarization operators in the schmidt basis are @xmath117 the fluctuation - entanglement relation therefore gives the subsystem purity as @xmath118 which converges exponentially fast with @xmath113 to the maximally entangled value @xmath119 @xcite .
subsystem fluctuations have been studied as probes of entanglement , with notable results for many - body systems @xcite .
still , only few exact and general relationships have been derived between an entanglement entropy and a subsystem variance in the many - body situation .
specific results have been reported for the case of independent qubits @xcite , one - dimensional free fermions @xcite , and luttinger liquids @xcite . in this work
we have derived a model - independent fluctuation - entanglement relation that provides a general , yet simple , quantitative basis to the long - held assertion that subsystem fluctuations can measure entanglement .
our result implies that it takes @xmath6 fluctuation measurements to construct a fluctuation - based entanglement measure for a rank-@xmath0 reduced density operator .
in fact , since the fluctuation @xmath120 depends on the first @xmath121 eigenvalues of the reduced density matrix , knowledge of these @xmath6 fluctuations is equivalent to knowledge of the full entanglement spectrum @xmath122 .
it therefore becomes increasingly cumbersome to use such a measure for a general entangled many - body system with exponentially large schmidt rank .
however , there are broad classes of many - body states that have reduced density matrices with well - bounded rank as in the examples presented here : models mappable onto one - dimensional free fermions @xcite , and matrix product states and the states that are well - approximated by them @xcite .
a. v. acknowledges support from the dost science education institute through its asthrd program .
the authors thank j. p. esguerra , e. a. galapon , and k. h. villegas for helpful discussions . | we identify subsystem fluctuations ( variances ) that measure entanglement in an arbitrary bipartite pure state .
these fluctuations are of observables that generalize the notion of polarization to an arbitrary @xmath0-level subsystem .
we express this polarization fluctuation in terms of subsystem purity and other entanglement measures .
the derived fluctuation - entanglement relation is evaluated for the ground states of a one - dimensional free fermion gas and the affleck - kennedy - lieb - tasaki spin chain .
our results provide a framework for experimentally measuring entanglement using stern gerlach - type state selectors . |
* the length is sufficient to characterize the exponential growth of each cell*. various techniques have been put forth to analyze cell morphology gathered from single cell images @xcite .
recent work on image analysis of single cells has attempted to optimize two problems : separation of distinct ( but potentially overlapping ) cells and accurate determination of the edge of each cell @xcite . because crowding is not an issue in our setup
, we could focus solely on constructing an algorithm to delineate each cell contour accurately and precisely . as shown in fig .
1a , b and described in the methods section , we first segment each cell using pixel - based edge detection similar to @xcite , then perform spline interpolation to determine the cell contour at sub - pixel resolution .
the sequence of such images for each single cell constitutes a trajectory in time @xmath3 that serves as the basis for quantitative analysis .
division events are then detected in an automated fashion using custom python code , and used to divide time trajectories for each cell into individual generations .
all data shown here were obtained by observing 260 single _ c. crescentus _ cells perfused in complex medium ( peptone - yeast extract ; pye ) at 31@xmath4c over the course of 2 days ( corresponding to 9672 separate generations ) . under these conditions ,
the mean population growth rate and division time remain constant , so we treat the trajectories of individual generations as members of a single ensemble . in other words
, we segment each cell trajectory by generation and take the resulting initial frame ( i.e. , immediately following division ) as @xmath5 minutes . in order to average over the ensemble , we then bin quantitative information according to time since division , @xmath3 , normalized by the respective division time @xmath6 .
the normalized time , @xmath7 , serves as a cell - cycle phase variable . for our quantitative analysis , we focus on a set of three intuitive and independent parameters that characterize cell shape at each stage of growth : length @xmath8 , width @xmath9 , and radius of curvature @xmath10 ( fig . 1c ) .
they are calculated directly from each splined contour as follows ( see also supplementary fig .
1 ) : * we define the length , @xmath11 , as the pole - to - pole distance along the contour of the cell medial axis at the normalized time @xmath12 ( fig . 2a ) . * we assign a single radius of curvature , @xmath13 , to each cell based upon the best - fit circle to the medial axis ( fig .
although stalked ( @xmath14 ) and swarmer ( @xmath15 ) portions may be described by different radii of curvature toward the end of the cell cycle , the average radius obtained by averaging the contributions of each portion yields the same value , i.e. , @xmath16 ( see supplementary fig . 2c ) .
* we define the width , @xmath17 as the length of the perpendicular segment spanning from one side of the cell contour to the other at each position @xmath18 along the medial axis , which runs from @xmath19 at the stalked pole to @xmath20 at the swarmer pole .
furthermore , we spatially averaged the width over positions along the medial axis , @xmath21 , to obtain a characteristic width at each time point ( fig . 2c ) .
the mean division time is @xmath22 min , where @xmath23 indicates a population average .
we find that @xmath24 increases exponentially with time constant @xmath25 min , essentially the same time constant that we previously observed for the cross - sectional area @xcite , while @xmath26 and @xmath27 remain approximately constant for @xmath28 and each shows a dip for @xmath29 when cell constriction becomes prominent .
the sharp rise in @xmath27 seen for @xmath30 results from independent alignment of the stalked and swarmer portions with the microfluidic flow as they become able to move independently ( i.e. , fluctuate easily about the plane of constriction ) .
these observations confirm the assumptions in @xcite that the length is sufficient to describe the growth of cell size .
moreover , we can track the dynamics of the spanning angle , @xmath31 , using the relation @xmath32 . *
mechanical model for cell shape and growth*. there are many details of cell growth and shape that require interpretation . for example , it is not obvious _ a priori _ that growth should be almost exclusively longitudinal .
therefore , we have developed a minimal mechanical model that can explain these observations .
we parametrize the geometry of the cell wall by a collection of shape variables @xmath33 , where @xmath34 , @xmath35 , and @xmath36 are the parameters introduced above ( fig .
as the cell grows in overall size , we postulate that the rate of growth in the shape parameter @xmath37 is proportional to the net decrease in cell wall energy , @xmath38 , per unit change in @xmath37 @xcite .
assuming linear response , the configurational rate of strain , @xmath39 , is proportional to the corresponding driving force @xmath40 , in analogy with the constitutive law of newtonian flow @xcite : @xmath41 where the constant @xmath42 describes the rate of irreversible flow corresponding to the variable @xmath37 . according to equation , exponential growth occurs if @xmath43 is constant , whereas @xmath37 reaches a steady - state value if @xmath44 along with the condition @xmath45 .
it thus remains to specify the form of @xmath46 . for a _
c. crescentus _ cell of total volume @xmath47 and surface area @xmath48 , our model for the total energy in the cell wall is given by @xmath49 where @xmath50 is a constant pressure driving cell wall expansion ; @xmath51 is the tension on the surface of the cell wall ; @xmath52 is the energy required to maintain the cell width ; @xmath53 represents the mechanical energy required to maintain the crescent cell shape ; @xmath54 is the energy driving cell wall constriction .
traditionally @xmath50 was taken to be the turgor pressure @xcite ; while the importance of the turgor pressure has recently been questioned @xcite , an effective pressure must still arise from the synthesis and insertion of peptidoglycan strands that constitute the cell wall .
we note that a purely elastic description of cell wall mechanics would lead to a curvature - dependent surface tension @xcite .
however , if growth is similar to plastic deformation , the tension is uniform @xcite .
the effective tension in our model depends on the local surface curvatures through the energy terms @xmath52 and @xmath53 , that describe harmonic wells around preferred values of surface curvatures .
the mechanical energy for maintaining width is given by @xmath55 where the constant @xmath56 is the preferred radius of curvature , @xmath57 is the bending rigidity and @xmath58 is a differential area element @xcite .
contributions to @xmath57 can come from the peptidoglycan cell wall as well as membrane - associated cytoskeletal proteins like mreb , mrec , rodz , etc . , which are known to control cell width @xcite .
in addition to maintaining a constant average width , _ c. crescentus _ cells exhibit a characteristic crescent shape , which relies on expression of the intermediate filament - like protein crescentin @xcite .
although the mechanism by which crescentin acts is not known , various models have been proposed , including modulation of elongation rates across the cell wall @xcite and bundling with a preferred curvature @xcite .
we assume the latter and write the energy for maintaining the crescent shape as @xmath59 where @xmath60 is the arc - length parameter along the crescentin bundle attached to the cell wall , @xmath61 is the local curvature , @xmath62 is the preferred radius of curvature , @xmath63 is the contour length , and @xmath64 is the linear bending rigidity . equation accounts for the compressive stresses generated by the crescentin bundle on one side of the cell wall , leading to a reduced rate of cell growth , according to equation . as a result
, the cell wall grows differentially and maintains a non - zero curvature of the centerline . in the absence of crescentin ( @xmath65 )
, our model predicts an exponential decay in the cell curvature that leads to a straight morphology , consistent with previous observations @xcite . finally , one must also account for the energy driving cell wall constriction .
constriction proceeds via insertion of new peptidoglycan material at the constriction site .
this process leads to the formation of daughter pole caps @xcite .
we take constriction to be governed by an energy of the form @xmath66 , where @xmath67 is the surface area of the septal cell wall , and @xmath68 is the energy per unit area released during peptidoglycan insertion .
* there exists an optimal cell geometry for a given mechanical energy*. to apply the model introduced above ( equations and ) to interpreting the data in fig . 2
, we assume a minimal cell geometry given by a toroidal segment with uniform radius of curvature @xmath13 , uniform cross - sectional width @xmath21 and the spanning angle @xmath69 . to this end
, we estimate as many mechanical parameters as we can from the literature and then determine the rest by fitting our experimentally measured values .
turgor pressure in gram - negative bacteria has been measured to be in the range @xmath70 mpa @xcite .
we use a value for the effective internal pressure close to the higher end of the measured values for turgor pressure , @xmath71 mpa , in order to account for peptidoglycan insertion .
we estimate the surface tension as @xmath72 nn@xmath73 m ( see supplementary model section ) and multiply it by the cell surface area @xmath74 to obtain the cell wall surface energy .
first , we neglect cell constriction ( setting @xmath75 ) and assume that the crescentin structure spans the length of the cell wall ( excluding the endcaps ) @xcite , with a contour length @xmath76 . the mechanical properties of mreb and crescentin are likely similar to those of f - actin and intermediate filaments , respectively @xcite .
however , due to a lack of direct measurements , we obtain the mechanical parameters @xmath57 and @xmath64 by fitting the model to the experimental data .
as desired , we find that the total energy @xmath46 has a stable absolute minimum at particular values of the cross - section diameter @xmath77 and the centerline radius @xmath10 , given by solution of @xmath78 ( see supplementary fig . 4 ) .
the measured values are @xmath79 m and @xmath80 m ( @xmath28 ) , and , as indicated by the red solid curves in fig .
2b , c , the model reproduces them with @xmath81 nn@xmath82 m and @xmath83 nn@xmath82m@xmath84 .
while the fitted value for @xmath64 is numerically close to the estimate based on the known mechanical properties of intermediate filaments ( @xmath85 nn@xmath82m@xmath84 ) , the value for @xmath57 is much higher than the bending rigidity of mreb bundles ( see supplementary information ) .
this indicates that @xmath57 is only determined in part by mreb and can have contributions from the cell wall .
given stable values for @xmath77 and @xmath10 , growth is completely described by the dynamics of the angle variable @xmath69 .
consequently , we write the total energy in the scaling form @xmath86 , with @xmath87 the energy density along the longitudinal direction .
the condition for growth then becomes @xmath88 , such that the energy is minimized for increasing values of @xmath69 . from our experimental data ,
the angle spanned by the cell centerline increases by an amount @xmath89 during the entire cycle . using our parameter estimates and fitting the data in fig . 2
, we obtain a numerical value for the energy density @xmath90 nn@xmath82 m .
we relate the angle dynamics to the length by @xmath91 where @xmath92 ( @xmath88 ) is the rate of longitudinal growth , which can be interpreted as resulting from remodeling of peptidoglycan subunits with a mean current @xmath93 , across the cell surface area @xmath94 . from an exponential fit to the data for cell length ( fig .
2a ) , we obtain @xmath95 ( nn @xmath82 m min)@xmath96 , which gives us an estimate of the friction coefficient , @xmath97 nn @xmath82m@xmath96 min , associated with longitudinal growth ; e.g. , mreb motion that is known to correlate strongly with the insertion of peptidoglycan strands @xcite .
our results are consistent with previous observations of _ c. crescentus _ cells with arrested division but continued growth @xcite .
* constriction begins early and proceeds with the same time constant as exponential growth*. having characterized the dynamics of growth , we now turn to constriction at the division plane .
as mentioned above , we obtain the experimental width at each point along each cell s medial axis .
the typical width profile is non - uniform along its length , exhibiting a pronounced invagination near the cell center ( with width @xmath98 ; fig .
this invagination , which ultimately becomes the division plane , is readily identifiable early in the cell cycle , even before noticeable constriction occurs .
we discuss the kinetics of constriction in this section , and focus on its location later in the manuscript . as shown in fig .
3a ( black points ) , @xmath99 progressively decreases towards zero until pinching off at @xmath100 . due to the limited spatial resolution of our imaging ( phase contrast microscopy ) , the pinch - off process occurring for @xmath30 could not be captured , but @xmath99 at earlier times ( i.e. , @xmath101 ) is precisely determined as a function of @xmath12 . to model the dynamics of constriction , we assume as in ref .
@xcite ( fig .
3a , inset ) : ( i ) the shape of the zone of constriction is given by two intersecting and partially formed hemispheres with radii @xmath102 ; and ( ii ) constriction proceeds by completing the missing parts of the hemisphere such that the newly formed cell wall surface maintains the curvature of the pre - formed spherical segments .
as a result , a simple geometric formula is obtained that relates the width of the constriction zone , @xmath98 , to the surface area @xmath103 of the newly formed cell wall , @xmath104 where @xmath105 is the maximum surface area achieved by the caps as the constriction process is completed , i.e. , when @xmath106 .
we assume that the addition of new cell wall near the division plane initiates with a rate , @xmath107 , and thereafter grows exponentially with a rate , @xmath108 , according to , @xmath109 subject to the initial condition @xmath110 . the first term on the right - hand side of equation follows from equation , using @xmath103 as the shape variable , after incorporating the constriction energy @xmath111 .
the rate of septal peptidoglycan synthesis , @xmath108 , is thus directly proportional to the energy per unit area released during constriction , @xmath68 .
the solution , @xmath112 , can then be substituted into equation to derive the time - dependence of @xmath98 , whose dynamics is controlled by two time scales : @xmath113 and @xmath114 .
fitting equation with the data for @xmath99 , we obtain @xmath115 min and @xmath116 min .
the fitted values for the time constants controlling constriction dynamics ( @xmath113 and @xmath114 ) are remarkably similar to that of exponential cell elongation ( @xmath117 min ) .
this shows that septal growth proceeds at a rate comparable to longitudinal growth .
therefore , one of the main conclusions that we draw is that cell wall constriction ( fig .
3a ) is controlled by the same time constant as exponential longitudinal growth ( fig .
2a ) . having determined the dynamics of @xmath98 , we compute the average width across the entire cell @xmath21 using the simplified shape of the constriction zone as shown in fig . 3a ( inset ) .
the resultant prediction ( blue solid curve in fig .
2c ) is in excellent agreement with the experimental data and captures the dip in @xmath26 seen for @xmath118 .
constriction also leads to a drop in the average radius of curvature of the centerline , as shown by the experimental data in fig .
in the supplementary material we derive a relation between the centerline radius of curvature @xmath13 and the minimum width @xmath98 , given by @xmath119 , predicting that cell curvature increases at the same rate as @xmath98 drops . using this relation , we are able to quantitatively capture the dip in @xmath27 seen for @xmath118 ( solid blue curve in fig .
2b ) without invoking any additional fitting parameters . *
origin of the asymmetric location of the primary invagination*. we now consider the position of the division plane and its interplay with cell shape . as shown in fig .
3b , the distance of the width minimum from the stalked pole ( @xmath120 ) increases through the cell cycle at the same rate as the full length of the growing cell ( @xmath11 ) , such that their ratio remains constant with time - averaged mean @xmath121 .
the presence of the primary invagination early in the cell cycle is reiterated in fig .
3c , which shows the width profile constructed by ensemble - averaging over each cell at the timepoint immediately following division .
in addition to the width minimum @xmath98 , there are two characteristic maxima near either pole , @xmath122 and @xmath123 , respectively ( fig . 3c , inset ) . as evident in fig . 3c
, the stalked pole diameter @xmath124 is on average larger than its swarmer counterpart @xmath125 ( also see supplementary fig .
we show that the asymmetric location of the invagination ( and the asymmetric width profile ) can originate from the distinct mechanical properties inherent to the pole caps in _ c. crescentus_. the shapes of the cell poles can be explained by laplace s law that relates the pressure difference , @xmath50 , across the cell wall to the surface tensions in the stalked or the swarmer pole , @xmath126 .
the radii of curvature of the poles then follow from laplace s law @xmath127 where the superscript ( @xmath128 ) denotes the stalked or the swarmer pole .
thus a larger radius of curvature in the poles has to be compensated by a higher surface tension to maintain a constant pressure difference @xmath50 . assuming that the poles form hemispheres , we have @xmath129 .
our data indicate that the early time ratio for @xmath130 ( @xmath131 ) shows a strong positive correlation with the ratio @xmath132 , with an average value @xmath133 ( see supplementary fig 3a ) .
laplace s law then requires that the stalked pole be mechanically stiffer than the swarmer pole ; @xmath134 .
this observation suggests that the asymmetry in the lengths of the stalked and swarmer parts of the cell depends upon different mechanical properties of the respective poles .
to quantitatively support this claim , we investigate an effective contour model for the cell shape . to this end , we assume that the fluctuations in cell shape relax more rapidly than the time scale of growth .
this separation of timescales allows us to derive the equation governing the cell contour by minimizing the total mechanical energy ( equation ) . from the solution we compute the resultant width profile for the entire cell ( see supplementary model section ) . as shown in fig .
3d , the model with asymmetric surface tensions of the poles causes the primary invagination to occur away from the cell mid - plane .
the spatial location of the invagination relative to the cell length depends linearly on the ratio @xmath135 .
symmetry is restored for @xmath136 , as shown in fig .
3d ( blue dashed curve ) .
we note that a gradient in @xmath51 along the cell body would imply differences in longitudinal growth rates between the stalked and the swarmer portions of the cell ( eq . ) .
our data exclude this possibility since both @xmath120 and @xmath137 grow at the same rate @xmath138 , as evidenced by the constancy of their ratio ( fig .
3b and supplementary fig .
because _ c. crescentus _ does not exhibit polar growth , the _ polar stiffness model _ is consistent with the observed uniformity in longitudinal growth rate .
in addition , the non - uniformity in cell width comes from the differences in mechanical response in the cell wall due to preferential attachment of crescentin along the concave sidewall . for a cres mutant cell ( where @xmath65 ) , our model predicts a uniform width profile before the onset of constriction .
* cell shape evolution during wall constriction*. the experimental width profiles show that the growing and constricting cells typically develop a second minimum in width ( fig .
these secondary invaginations are observed in both the stalked and swarmer portions of single cells in the predivisional stage ( @xmath139 ) , although they are more common in the stalked portions ( fig .
we show here that these secondary minima become the primary minima in each of the daughter cells . to study the dynamics of the development of the secondary minimum we introduce a new quantity , @xmath140 , defined as the distance from the stalked pole to the secondary minimum in the stalked part ( see fig . 5c , inset )
we find that the ratio @xmath141 has a mean value of @xmath142 0.55 at later points in the cell cycle ( fig .
5b ) , equal to the constant ratio maintained by the distance from the stalked pole to the primary minimum , @xmath143 .
in fact the kymograph of width profiles ( shown over 2 generations for a representative single cell ) in fig .
5c demonstrates that the predivisional secondary invaginations are inherited as primary invaginations after division .
this mechanism provides continuity and inheritance of the invaginations across generations and is an intrinsic element of the mechanism for cell division in _ c. crescentus_. to quantitatively explain the experimental width profiles during constriction , we use our mechanical model to determine the instantaneous cell shape by minimizing the total energy ( equation ) at the specified time points ( see supplementary model section ) . to take constriction into account
, we impose the constraint that @xmath144 , where @xmath145 is determined by equations and .
in addition , we assume non - uniform materials properties in the cell wall by taking the tension in the cell poles ( @xmath146 ) and the septal region to be higher than the rest of the cell . as constriction proceeds and @xmath145 decreases , we compute the shape of the cell contours ( fig .
4c ) and the corresponding width profiles ( fig .
the computed width profiles faithfully reproduce the secondary invaginations , which become more pronounced as the daughter pole caps become prominent .
an example of the experimental width profiles is shown in fig .
4b at evenly - spaced intervals in time for a single generation , and the corresponding model width profiles are shown in fig .
we note that the experimental cell contours in the predivisional stage ( @xmath30 ) bend away from the initial midline axis and develop an alternate growth direction ( fig . 4a , blue contour ) .
these bend deformations are induced by the microfluidic flow about the pinch - off plane ; the cells become increasingly `` floppy '' as the constriction proceeds .
the consistent propagation of a specific shape through the processes of growth and division relies upon an intricate interplay between the controlled spatiotemporal expression and localization of proteins , and cytoskeletal structural elements .
the high statistical precision of our measurements allows us to gain new insights into cell morphology . from precise determination of cell contours over time , we observe that a typical cell width profile is non - uniform at all times with a pronounced primary invagination appearing during the earliest stages of the cell cycle . during cell constriction ,
the decrease in the minimum width is governed by the same time constant as exponential axial growth ( fig .
furthermore , the location of the primary invagination divides the cell contour into its stalked and swarmer compartments , such that the ratio of the length of the stalked part @xmath147 to the total pole - to - pole length @xmath11 remains constant during the cycle with a mean value @xmath148 ( fig .
these observations and our mechanical model lead to two important conclusions : first , _ the dynamics of cell wall constriction and septal growth occur concomitantly _ , and second , _ the asymmetric location of the primary invagination can be explained by the differences in mechanical properties in the stalked and swarmer poles_. a corollary of the first conclusion is that the size ratio threshold at division occurs naturally without requiring a complex timing mechanism @xcite .
in addition to the primary septal invagination , the cell contours exhibit a pronounced secondary invagination during the predivisional stages ( fig .
remarkably , the secondary invaginations develop at a precise location relative to the total length of the stalked compartments , @xmath149 ( fig .
the data thus allow a third conclusion : _ these secondary invaginations are inherited as primary invaginations in each of the daughter cells , directing the formation of the division plane in the next generation_. thus , through consistent and controlled nucleation of invaginations across generations , _ c. crescentus _ cells maintain a constant ratio of the sizes of stalked and swarmer daughter cells .
our experimental observations and the parameters in the cell shape model can be related to the current molecular understanding for gram - negative bacteria , in particular _ c. crescentus_. before the onset of noticeable constriction , cell shape is dictated by the mechanical properties of the peptidoglycan cell wall in addition to various shape - controlling proteins such as mreb , mrec , rodz and cres .
single molecule tracking studies have revealed that mreb forms short filamentous bundles anchored to the inner surface of the cell wall and moves circumferentially at a rate much faster than the rate of cell growth @xcite . _ in vitro _
experiments show that mreb filaments can induce indentation of lipid membranes , suggesting that they may have a preferred radius of curvature @xcite .
thus on time scales comparable to cell growth , @xmath52 is determined in part by the energy cost of adhering mreb bundles to the cell wall ( see supplementary model section ) .
bacterial cell division is driven by a large complex of proteins , commonly known as divisomes that assemble into the z - ring structure near the longitudinal mid - plane of the cell @xcite .
the z - ring contains ftsz protofilaments that are assembled in a patchy band - like structure @xcite .
ftsz protofilaments are anchored to the cell membrane via ftsa and zipa , and play a crucial role in driving cell wall constriction @xcite . during constriction
, the divisome proteins also control peptidoglycan synthesis and direct the formation of new cell wall via the activity of penicillin - binding proteins ( pbps ) @xcite .
thus the divisome plays a two - fold role by concomitantly guiding cell wall constriction and growth of the septal peptidoglycan layer . according to our model the constriction of the cell wall
is driven by the synthesis of septal cell wall at a rate @xmath108 ( @xmath150 ) , which can be directly related to the activity of pbps triggered by the divisome assembly .
furthermore , in our model it is sufficient that the divisome guide the curvature of cell wall growth in the septal region ( see fig .
3a , inset ) .
while the mechanism behind the precise asymmetric location of the division plane in _ c. crescentus _ cells is not well understood , it is likely that the atpase mipz helps division site placement by exhibiting an asymmetric concentration gradient during the predivisional stage @xcite .
mipz activity inhibits ftsz assembly ; as a result of polar localization of mipz , z - ring assembly is promoted near the mid - cell @xcite .
our cell shape model suggests that the early time asymmetric location of the primary invagination , which develops into the division plane , is controlled by the differences in surface tensions maintained in the poles .
the presence of this invagination at @xmath151 , as inherited from the secondary invaginations in the previous generation , aids in z - ring assembly at the site of the invagination .
the curvature - sensing capability of the z - ring may be enabled by the minimization of the ftsz polymer conformational energy that is determined by the difference between cell surface curvature and ftsz spontaneous curvature @xcite .
a higher tension in the stalked pole can be induced by asymmetric localization of polar proteins , such as popz , early in the cell cycle .
experiments have shown that popz localizes to the stalked pole during the initial phase of the cell cycle and increasingly accumulates at the swarmer pole as the cell cycle proceeds @xcite .
consistent with this observation , our data show that the correlation between the pole sizes ( determined by the ratio of surface tension to pressure ) and the stalked and swarmer compartment lengths tend to disappear later in the cycle ( supplementary fig .
3 ) , as cell constriction proceeds .
a recent experimental study also demonstrates that molecular perturbation of clp proteases can destroy the asymmetry of cell division in _ c. crescentus _
@xcite , suggesting the interplay of subcellular protease activity with the physical properties of the cell wall .
earlier theoretical models have predicted that a small amount of pinch - off force from the z - ring ( @xmath152 pn ) is sufficient to accomplish division by establishing a direction along which new peptidoglycan strands can be inserted @xcite . in contrast , our data combined with the mathematical model allows the interpretation that _ the early time asymmetric invagination in the cell wall can set the direction for the insertion of new peptidoglycan strands_. constriction results from exponential growth of surface area in the septum ( at the same rate as longitudinal extension ) .
the instantaneous cell shape is determined by minimizing the energy functional at given values of the cell size parameters .
finally , from our estimate of the cell wall energy density @xmath87 ( @xmath153 nn@xmath82 m ) , we predict that a net amount @xmath154 nn@xmath82 m of mechanical energy is used by the peptidoglycan network for cell wall growth . for a _
c. crescentus _ cell of surface area @xmath155m@xmath84 , layered with glycan strands of length @xmath15 nm and cross - linked by peptide chains with maximally stretched length @xmath14 nm @xcite , there are roughly @xmath156 peptidoglycan subunits . thus on average , each peptidoglycan subunit can consume mechanical energy of @xmath12.4@xmath157 nn@xmath82 m , or @xmath10.6 @xmath158 at a temperature @xmath159c .
cell wall remodeling and insertion of new peptidoglycan material can likely create defects in the peptidoglycan network @xcite .
one thus expects cellular materials properties to change over time , as a result of these molecular scale fluctuations .
although we neglect such variations in our mean field model , it nonetheless quantitatively captures the average trends in cell shape features . in future
work we plan to more closely connect the energy terms of the continuum model with molecular details .
* acquisition of experimental data*. data were acquired as in ref . @xcite .
briefly , the inducibly - sticky _ caulobacter crescentus _ strain fc1428 was introduced into a microfluidic device and cells were incubated for one hour in the presence of the vanillate inducer .
the device was placed inside a homemade acrylic microscope enclosure ( @xmath160 ) equilibrated to 31@xmath4c ( temperature controller : csc32j , omega and heater fan : hgl419 , omega ) . at the start of the experiment ,
complex medium ( peptone - yeast extract ; pye ) was infused through the channel at a constant flow rate of 7 @xmath82l / min ( phd2000 , harvard apparatus ) , which flushed out non - adherent cells . a microscope
( nikon ti eclipse with the `` perfect focus '' system ) and robotic xy stage ( prior scientific proscan iii ) under computerized control ( labview 8.6 , national instrument ) were used to acquire phase - contrast images at a magnification of 250x ( emccd : andor ixon+ du888 1k @xmath161 1k pixels , objective : nikon plan fluor 100x oil objective plus 2.5x expander , lamp : nikon c - hfgi ) and a frame rate of 1 frame / min for 15 unique fields of view over 48 hours . in the present study we use a dataset consisting of 260 cells , corresponding to 9672 generations ( division events ) .
* analysis of single cell shape*. the acquired phase - contrast images were analyzed using a novel routine we developed ( written in python ) .
each image was processed with a pixel - based edge detection algorithm that applied a local smoothing filter , followed by a bottom - hat operation .
the boundary of each cell was identified by thresholding the filtered image .
a smoothing b - spline was interpolated through the boundary pixels to construct each cell contour .
each identified cell was then tracked over time to build a full time series .
we chose to include only cells that divided for more than 10 generations in the analysis .
a minimal amount of filtering was applied to each growth curve to remove spurious points ( e.g. , resulting from cells coming together and touching , or cells twisting out of plane ) .
the timing of every division was verified by visual inspection of the corresponding phase contrast images , so that the error in this quantity is approximately set by the image acquisition rate of 1 frame / min .
10 url # 1`#1`urlprefix[2]#2 [ 2][]#2 & . _ _ * * , ( ) . & . _ _ * * , ( ) . &
_ _ * * , ( ) .
_ _ ( , , ) . , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . , & .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . , & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . , & .
_ _ * * , ( ) . , , , & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ et al . _ . _ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . , , , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) .
_ _ ( , , ) . , & . _ _ * * , ( ) . & . _ _ * * , ( ) . .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , , , & .
_ _ * * , ( ) . , , &
. _ _ * * , ( ) . , , & .
_ _ * * , ( ) . & _ _ * * , ( ) . & _ _
* * , ( ) . , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) .
_ _ * * , ( ) . , , , & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . , , & .
_ _ * * , ( ) .
_ et al . _ . _ _ ( ) . , , & . _
_ * * , ( ) . , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , , , &
_ _ * * , ( ) . & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ et al . _ .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) .
we gratefully acknowledge funding from nsf physics of living systems ( nsf phy-1305542 ) , nsf materials research science and engineering center ( mrsec ) at the university of chicago ( nsf dmr-1420709 ) , the w. m. keck foundation and the graduate program in biophysical sciences at the university of chicago ( t32 eb009412/eb / nibib nih hhs / united states ) .
n.f.s . also thanks the office of naval research ( onr ) for a national security science and engineering faculty fellowship ( nsseff ) .
c.s.w . , s.i.b . , a.r.d . , and n.f.s .
designed the experiments ; s.b .
, a.r.d . , and n.f.s .
designed the model ; c.s.w . and s.i.b .
performed the experiments and observed the phenomena reported ; c.s.w . designed and implemented custom software to automate cell shape image analysis ; c.s.w . and s.b . analyzed the data ; s.c .
contributed reagents and materials ; c.s.w .
, a.r.d . , and n.f.s .
wrote the manuscript ; all authors discussed the results and commented on the manuscript .
the authors declare no competing financial interests . * supplemental material * biblabel[1][s#1 ]
a number of quantities can be immediately calculated from each splined cell contour ( supplementary fig .
1a ) , including the cross - sectional area .
the cell medial axis was determined by calculating the voronoi diagram of the cell contour [ s@xcite ] , pruning the branches ( supplementary fig .
1b ) , and extending this skeleton to the edges of the cell contour in a manner that preserves average curvature of the medial axis ( supplementary fig .
the intersections between the cell medial axis and contour represent the stalked and swarmer poles , respectively .
the cell length was calculated by evaluating the distance along the medial axis between either pole .
the cell widths were determined by creating ribs perpendicular to the medial axis along its length and determining the distances between their intersections with opposite sides of the cell contour ( supplementary fig .
to calculate time - averaged quantities , we normalized trajectories for each generation by the respective division time @xmath6 ( thus converting each variable to a function of @xmath7 ) .
we then split these data into 73 bins , as @xmath162 min under these conditions , and ensemble - averaged each of these bins over every generation .
we can define five specific values of the width according to the local minima and maxima in the width profile ( @xmath163 , @xmath164 , @xmath165 , @xmath166 , @xmath167 ) , not all of which may be present in any given cell . where we identify a primary minimum in the width profile , @xmath163 , we can also determine two local maxima , @xmath164 and @xmath165 , corresponding respectively to the stalked and swarmer portions ( supplementary fig .
these values are approximately constant throughout the cell cycle , with @xmath168 . however , at later times ( @xmath139 ) the value of @xmath123 increases until @xmath169 . in some cases , additional secondary local minima
are observed , @xmath170 and @xmath171 , corresponding respectively to the stalked and swarmer portions ( supplementary fig .
although we note the value of these quantities for early times here ( where they are approximately equal to their respective local maxima ) , these minima can only be determined with certainty at later times ( @xmath139 ) .
there , we observe the presence of a secondary minimum in the stalked portions of most cells .
p5.5em p6.5em p3em l + & mean @xmath172 s.d . & & @xmath12 + + @xmath173 & 0.74 @xmath172 0.02 & @xmath82 m & @xmath174 $ ] + @xmath175 & 0.85 @xmath172 0.02 & @xmath82 m & @xmath174 $ ] + @xmath176 & 0.84 @xmath172 0.03 & @xmath82 m & @xmath174 $ ] + @xmath177 & 0.83 @xmath172 0.02 & @xmath82 m & @xmath174 $ ] + @xmath178 & 0.82 @xmath172 0.02 & @xmath82 m & @xmath174 $ ] + @xmath179 & 4.44 @xmath172 2.12 & @xmath82 m & @xmath174 $ ] + @xmath180 & 4.34 @xmath172 2.50 & @xmath82 m & @xmath174 $ ] + @xmath181 & 3.94 @xmath172 2.42 & @xmath82 m & @xmath174 $ ] + @xmath182 & 54.3 @xmath172 5.2 & % & @xmath183 $ ] + @xmath184 & 45.7 @xmath172 5.2 & % & @xmath183 $ ] + + we split each cell into stalked and swarmer portions according to the location of the primary minimum in the width profile .
the radius of curvature ( @xmath10 ) was calculated as the radius of the best - fit circle to the cell medial axis .
we tested two methods of determining the radius of curvature : ( i ) fitting the whole cell to a circle or ( ii ) fitting the stalked and swarmer portions separately . at early times in the cell cycle , ( ii ) gives poorer results because of fewer data points . at later times in the cell cycle , ( i ) gives poorer results because the stalked and swarmer portions can indeed have differing radii of curvature , which we attribute to the alignment of the swarmer portion of the cell with the direction of fluid flow after the division plane has narrowed enough that it becomes mechanically decoupled from the stalked portion , i.e. , like a flexible hinge .
however , in the mean the results of ( ii ) are equal to the value calculated by ( i ) for earlier times .
therefore , we use only data from method ( i ) but exclude later time points ( supplementary fig .
because of the variation in the values of the calculated radius of curvature was large , such that we can not define a reasonable arithmetic mean without arbitrarily filtering the dataset , we first averaged the corresponding unsigned curvature ( equivalent to @xmath185 ) and then converted to radius of curvature ( i.e. , what we report as @xmath186 is actually the harmonic mean , calculated as @xmath187 ) .
the length was also split according to the locations of the local minima , into @xmath188 ( distance along cell medial axis from stalked pole to @xmath163 ) , @xmath189 ( distance along cell medial axis from swarmer pole to @xmath163 ) , @xmath190 ( distance along cell medial axis from stalked pole to @xmath166 ) , and @xmath191 ( distance along cell medial axis from swarmer pole to @xmath167 ) .
the value of @xmath188 and @xmath190 are compared in supplementary fig .
note that the length of the stalked portion @xmath188 grows exponentially , with the same time constant as the length @xmath8 ( as does @xmath189 , although it is not shown here for clarity ) , which is a necessary condition for the addition of peptidoglycan material along the entire length of the cell when the location of @xmath163 is set at early times . at later times ( @xmath139 ) , the length from stalked pole to secondary minimum @xmath190 also starts to increase .
we relate the asymmetry in the length of stalked and swarmer portions at early times to asymmetries in the stalked and swarmer poles .
the model predicts a linear relationship between the ratio of lengths @xmath192 and the ratio of the tensions at either pole @xmath193 . we can not directly measure the latter quantity , but from laplace s law it is equal to the inverse ratio of the mean curvatures at either pole , which we approximate as the inverse of half the maximum pole width , assuming that the poles are hemispheres with diameter @xmath164 and @xmath165 , respectively .
supplementary fig .
3 shows scatter plots comparing @xmath194 to @xmath192 for three different time intervals .
the red best - fit line is shown only for supplementary fig .
3a , for which @xmath195 ( linear fits to all other plots produced values of @xmath196 ) .
this line runs from the point @xmath197 , corresponding to the dashed blue curve in fig .
3c of the main text ( which is the case of a symmetric cell ) to @xmath198 , corresponding to the red solid curve in fig .
3d of the main text ( which is the case of the average asymmetric _ c. crescentus _ cell ) . note that at times @xmath199 , any correlation disappears . in order to quantify the error in our width profiles , we imaged a single field of view of 24 _ c. crescentus _ cells perfused in complex medium at 31@xmath4c at a frame rate of 5 frames per second ( 300 times faster than the frame rate used to acquire all other data ) , and calculated the splined contours for each cell .
we focused in particular on a single `` dead '' ( non - growing and non - dividing ) cell , and found the root - mean - square deviation ( rmsd ) of nearest points along the cell contour between subsequent frames to be 12 nm .
in addition , we found the rmsd of equivalent points along the width profile between subsequent frames to be 28 nm , or a 3.2% pinch depth at the average value of @xmath164 = 0.85 @xmath82 m .
* cell wall mechanics*. the total energy @xmath46 for the bacterial cell wall is given as the sum of contributions from an active internal pressure @xmath50 driving cell volume ( @xmath47 ) expansion , mechanical energy @xmath200 in the cell wall , and the mechanical energy of interactions with cytoskeletal proteins @xmath201 : @xmath202 the bacterial cell wall consists of a network of glycan strands cross - linked by peptide chains know as the peptidoglycan network .
growth occurs via the insertion of new peptidoglycan strands into the existing network along with the breaking of existing bonds due to turgor pressure induced stretching .
we assume that elastic equilibrium is reached rapidly as compared to the rate of synthesis of new material [ s@xcite ] . as a result of cell wall remodeling and irreversible elongation
, growth can be understood as resulting from plastic deformations [ s@xcite ] .
to understand the origin of the cell wall tension , @xmath51 , in the model introduced in the main text , we consider the cell wall as a thin elastic shell that deforms plastically when stretched beyond a maximum strain @xmath203 , the yield strain .
a thin shell has two modes of elastic deformations , bending and stretching [ s@xcite ] , such that @xmath204 . in the limit of small thickness of the shell
@xmath205 as compared to its radii of curvature , one can neglect the bending energy ( that scales as @xmath206 ) whereas the stretching energy @xmath207 is given by , @xmath208 where @xmath209 is the mechanical stress tensor and @xmath210 is the strain tensor . as yield strain
is reached at the onset of growth , we have @xmath211 ( assuming isotropic stretching ) , where @xmath212 is the kronecker delta .
furthermore , assuming a hookean constitutive relation for the stress tensor [ s@xcite ] @xmath213\;,\ ] ] where @xmath214 is the young s modulus and @xmath215 is the poisson ratio , we have @xmath216 , where , @xmath217 for a gram - negative bacterial cell wall of thickness @xmath218 nm , elastic modulus @xmath219 mpa [ s@xcite ] and average yield strain @xmath220 , the wall tension is estimated to be @xmath72 nn@xmath73 m .
while the actual value for the turgor pressure counteracting this tension can be contested , our choice for the numerical value for internal pressure @xmath50 can be justified using a simple mechanical argument .
radial force - balance dictates that in order to maintain an average cross - sectional radius @xmath221 @xmath82 m , a cell wall with surface tension @xmath222 50 nm/@xmath82 m has to balance an internal pressure of magnitude @xmath223 0.25 mpa , which is numerically very close to our choice for @xmath71 mpa .
* cytoskeletal bundle mechanics*. next , we model the mechanical energy in the cell wall due to interactions with cytoskeletal proteins .
our minimalist approach considers two crucial protein bundles that are directly responsible for maintaining the shape of _ c. crescentus
mreb protein bundles contribute an effective energy @xmath52 , which favors a rod - like shape , and crescentin filament bundles contribute an energy @xmath53 , which favors a crescent - like shape .
we thus have @xmath224 .
mreb subunits form patchy filamentous bundles adherent to the cell wall and oriented perpendicular to the long axis of the cell [ s@xcite ] . the elastic energy stored in an adherent mreb subunit
is given by @xmath225 where @xmath226 labels the subunit , @xmath227 is the circumferential radius of curvature of the cell wall where the subunit is attached , @xmath228 is the length of the subunit , @xmath56 is the intrinsic radius of curvature and @xmath229 is the bending rigidity of the associated mreb bundle .
the total energy imparted by a collection of @xmath230 attached mreb subunits is given by @xmath231 .
next , employing a continuum mean field assumption , we replace individual subunit lengths by their average length @xmath232 nm [ s@xcite ] and assume a uniform number density @xmath233 of mreb subunits in the cell surface to obtain @xmath234 where , @xmath235 is the effective bending modulus due to mreb induced traction forces .
the bending rigidity of an mreb bundle is given by @xmath236 , where @xmath237 is the flexural rigidity of mreb filaments ( assumed to be similar to f - actin ) , @xmath238 is the number of mreb protofilaments per bundle , and @xmath239 is an exponent in the range 12 depending on the strength of crosslinking or bundling agents [ s@xcite ] .
mreb filaments in a bundle appear to have strong lateral interaction with negligible filament sliding , so we assume @xmath240 .
the diameter of an mreb protofilament is @xmath241 nm [ s@xcite ] and the average width of an mreb bundle has been determined from super - resolution imaging to be in the range 60 - 90 nm [ s@xcite ] , giving the estimate @xmath242 15 - 22 .
estimating the surface number density of mreb subunits as @xmath243 @xmath82m@xmath244 and using @xmath245 nn@xmath82m@xmath84 [ s@xcite ] , we obtain @xmath57 in the range 5.612.1 nn@xmath82 m .
crescentin proteins form a cohesive bundled structure anchored to the sidewall of _ c. crescentus _ cells .
the energetic contribution due to crescentin is given by @xmath246 where @xmath64 is the bending rigidity , @xmath247 is an arc - length parameter , @xmath248 is the longitudinal curvature of the cell wall and @xmath62 is the preferred radius of curvature of crescentin bundles . the bending rigidity of crescentin can be expressed as @xmath249 , where @xmath250 is the young s modulus and @xmath251 is the area moment of inertia of the bundle ( with width @xmath252 ) given by @xmath253 . since crescentin is an intermediate filament homologue , we assume that @xmath250 is similar to the young s modulus of intermediate filament bundles given by @xmath254 mpa [ s@xcite ] . assuming @xmath255 @xmath82 m , we estimate the bending rigidity of a crescentin bundle to be @xmath256 nn@xmath82m@xmath84 .
* mean field model for cell shape and size dynamics*. as described in the main text , the total mechanical energy in the cell wall of _ c. crescentus _ cells can be given as a sum of contributions from internal pressure , wall surface tension , mechanical energy of interactions with bundles of cytoskeletal proteins such as crescentin and mreb , and the constriction energy @xmath54 during cell division .
in the mean field description , we neglect the contributions from @xmath54 and disregard any spatial variations in cell geometry .
the mean field description is a good approximation of the cell shape dynamics for @xmath257 , where the average width and the midline radius of curvature remains constant ( fig . 2 in the main text ) .
we approximate the shape of _ c. crescentus _ cells as the segment of a torus with radius of curvature @xmath258 , cross - sectional width @xmath259 and spanning angle @xmath260 .
we also neglect the pole caps that are mechanically rigid and do not remodel during wall growth .
the total energy is then given by @xmath261=-p ( \pi { \bar{w}}^2 r\theta)/4 + \gamma ( \pi r { \bar{w}}\theta ) + e_\text{width } + e_\text{cres}\;,\ ] ] where @xmath262 ^ 2 \;,\ ] ] @xmath263 ^ 2\;.\ ] ] from the above expressions , we see that the total energy has the scaling form @xmath264=\theta u[{\bar{w}},r]$ ] . according to eq .
( 1 ) in the main text , the dynamics of the radius of curvature @xmath10 , the spanning angle @xmath31 and the width @xmath9 are given by @xmath265 where @xmath266 , @xmath267 and @xmath268 are the rate constants .
the steady - state values for the radius of curvature of the centerline and the cross - sectional width is given by the solutions to @xmath269 and @xmath270 .
as shown in supplementary fig .
4a the energy density @xmath87 admits a stable absolute minimum in width , controlled by the stiffness parameter @xmath57 such that for lower values of @xmath57 ( red curve in supplementary fig .
4a ) , @xmath87 does not have a minimum and the width grows in time rendering the rod - like shape unstable .
similarly , the parameter @xmath64 controls cell curvature , such that the energy density has a stable absolute minimum in @xmath10 beyond a critical stiffness @xmath64 ( supplementary fig .
4b ) . at lower values of @xmath64 the cell
does not have a stable radius of curvature and the energy is minimized by increasing @xmath10 .
this leads to straightening of the cell s medial axis . in the parameter range where the cell maintains a stable value for @xmath77 and @xmath10 ,
the energy density @xmath87 becomes a constant during cell elongation . from the growth law introduced in the main text
, the spanning angle @xmath31 evolves in time according to the equation @xmath271 such that the condition for exponential growth becomes @xmath88 .
* constriction dynamics*. as described in the main text , we assume that the shape of the constriction zone is given by two intersecting hemishperical segments with diameter @xmath272 . the total surface area of the septum is given by @xmath273 , where @xmath274 is the total length of the spherical segments ( see supplementary fig .
5a ) . using elementary geometry , we obtain the following relation between @xmath274 and the minimum width , @xmath275 : @xmath276 if septal growth occurs by addition of new peptidoglycan strands while maintaining the curvature of the spherical segments , then @xmath274 is the shape variable controlling the growth of septal surface area @xmath67 as well as the constriction dynamics of @xmath275 . to describe the dynamics of @xmath274
, we assume that initial phase of elongation occurs with a velocity @xmath277 , and thereafter growth follows an exponential law with rate @xmath108 , @xmath278 multiplying both sides of the equation by @xmath279 , we derive the growth dynamics of the surface area s(t ) , @xmath280 where @xmath281 .
having determined the dynamics of @xmath282 , one can evaluate the time - dependence of the average width @xmath77 defined as , @xmath283 , where @xmath60 is the coordinate along the centerline . using the simplified geometry of the constricting cell , given by two toroidal segments ( peripheral regions ) connected by two intersecting hemispheres ( septal region ) , one gets , @xmath284\;,\ ] ] where @xmath285 , is
the width in the septal region and @xmath77 is the average width of the stalked and swarmer components as determined by eq . .
as shown in fig . 2c in the main text ( blue solid line ) , eq .
is in excellent agreement with the experimental data and captures the dip in @xmath259 seen for @xmath118 . fig .
2b in the main text also shows that the radius of curvature of the centerline ( @xmath258 ) drops for @xmath118 when cell constriction is prominent .
therefore we examine the role of constriction dynamics on the time - dependence of the centerline radius of curvature .
the centerline spans a distance @xmath274 and an angle @xmath286 in the septal region , as shown in supplementary fig .
consider a small segment @xmath287 along the centerline with a local radius of curvature @xmath10 and spanning angle @xmath288 .
we then have the relation @xmath289 .
this leads to the following identity , @xmath290 now the geometry of the septal region ( supplementary fig .
5a ) directly relates the rate of increase in the spanning angle @xmath291 to the rate of drop in the tangent angle ( @xmath292 ) at the constriction site , @xmath293 .
furthermore , using the relations , @xmath294 and @xmath295 , we get @xmath296 .
the last equality follows after variations of eq . .
as a result we have the following kinetic relation , @xmath297 combining the identities in eq .
and we get , @xmath298 , showing that the radius of curvature drops at the same rate as @xmath275 shrinks .
* contour model for cell shape*. to quantitatively capture the experimentally observed spatial variations in cell shape we study an effective two - dimensional contour model for the cell shape .
this approach facilitates closer comparison with the two - dimensional splined contours obtained from our experimental data .
the contour description of the rod - like cell shape negelects circumferential variations in cell geometry .
the model incorporates non - uniform materials properties and mechanical constraints across the cell wall , with the poles and the septal region being mechanically stiffer than the rest of the cell wall . by exploiting the rod - like geometry
, one can use the centerline curve to divide the contour into two parts , the _ upper _ and the _ lower _ cell wall . as shown in supplementary fig .
we further subdivide the cell contour into the stalked and swarmer poles ( @xmath299 ) , the upper ( @xmath300 ) and lower cell wall ( @xmath301 ) , and the septal region ( sep ) .
we parametrize the instantaneous shape of the cell contour using the two - dimensional centerline curve @xmath302 , the length of the centerline @xmath303 and the width @xmath304 , where @xmath60 is the absolute distance along the centerline from the stalked pole such that @xmath305 $ ] . if @xmath306 denotes the outward unit normal vector on the centerline , the curves defining the upper and lower parts of the cell contour , @xmath307 , are given by the relation , @xmath308 , where @xmath309 represent the perpendicular distances of the top and bottom curves from the centerline .
the total cell width is then given by @xmath310 .
it is convenient to switch to polar coordinates , where the shape of the cell contour is given by the re - parametrized curve @xmath311 , where @xmath312 is the angular coordinate spanning the centerline , which can be approximated as the arc of a circle with radius @xmath258 .
since the ratio @xmath313 is small at all times , one can approximate the local curvature as @xmath314 where prime denotes derivative with respect to @xmath315 and the subscripts @xmath172 represent the upper and the lower part of the cell contour respectively .
furthermore , in the linear regime , the differential arc length can be approximated as @xmath316 .
the dynamics of the shape parameters @xmath303 , @xmath260 , and @xmath258 are determined from the kinetic law in eq .
( 1 ) of the main text .
the instantaneous width profile @xmath304 results from minimizing the total energy functional , which leads to the following shape equation , @xmath317 where , @xmath318 is the tension on the cell contour and @xmath319 and @xmath320 are the linear force densities on the contour due to maintenance of width and the crescent shape , respectively . in the contour model
we simplify the energetic contribution due to maintenance of width ( acting on the sections @xmath300 and @xmath301 ) as @xmath321 where the elastic constant @xmath322 depends on the bending rigidity @xmath57 introduced in the main text as @xmath323 .
this linear approximation holds if @xmath324 . from our data and mean field model
fits we get @xmath325 .
the resultant force density is given by @xmath326 .
the bending energy induced by crescentin protein bundles anchored onto the cell wall ( regions @xmath300 ) is given by @xmath327 ^ 2\;,\ ] ] where @xmath328 is the spontaneous curvature of the crescentin bundle , and s is the arc - length parameter along the upper part of the cell contour ( @xmath300 ) which is related to @xmath60 as @xmath329 . to obtain the force density we consider an infinitesimal deformation , @xmath330 , of the upper contour as @xmath331 , where @xmath332 is the outward unit normal .
accordingly the curvature and the differential arc - length changes , @xmath333 and @xmath334 , where @xmath335 and @xmath336 [ s@xcite ] .
the resultant force density is obtained after variations of the energy functional , @xmath337 , where @xmath338 $ ] .
this leads to the following non - linear force contribution : @xmath339 using eq .
, @xmath340 can be linearized and expressed using the angular coordinate @xmath315 as @xmath341\;.\ ] ] * compartmentalizing the cell contour*. ( 1 ) pole caps : the cell poles are assumed to be mechanically inert in the sense that they do not interact with the active cytoskeletal proteins such that @xmath342 .
the mechanical forces acting on the cell poles come from turgor pressure @xmath50 , and the tension @xmath343 .
the shape of the cell poles are then described by the two - dimensional laplace s law , @xmath344 where the superscripts @xmath345 and @xmath346 denote respectively the stalked and swarmer poles .
\(2 ) upper cell contour ( @xmath300 ) : the region @xmath300 in the upper cell contour obeys the force - balance equation : @xmath347 at the endpoints of the segments @xmath300 we impose the boundary condition that the curvature @xmath348 must equal the longitudinal curvature of the poles .
\(3 ) bottom cell contour ( @xmath301 ) : in the absence of crescentin , the region @xmath301 in the bottom cell contour obeys the force - balance equation : @xmath349 \(4 ) septal region : we incorporate the effect of constriction in the cell shape equation by imposing the constraint ( boundary condition ) that @xmath350 , where @xmath275 evolves according to the kinetics described in eq .
( 6 ) and ( 7 ) of the main text .
furthermore , the curvatures at the end points of the septal segments must conform to the curvature of the newly formed poles .
subject to these boundary conditions , the width profile in the upper septal region ( @xmath351 ) is given by , @xmath352 with a surface tension ( @xmath353 ) of the newly formed poles chosen to be much higher than the peripheral region , @xmath354 .
the contour below the centerline obeys the equation , @xmath355 the governing cell shape equation given in eq . , is then solved numerically in each part of the cell contour with matching boundary conditions in @xmath356 and its derivatives .
10 url # 1`#1`urlprefix[2]#2 [ 2][]#2 | we investigate the intergenerational shape dynamics of single _ caulobacter crescentus _ cells using a novel combination of imaging techniques and theoretical modeling .
we determine the dynamics of cell pole - to - pole lengths , cross - sectional widths , and medial curvatures from high accuracy measurements of cell contours .
moreover , these shape parameters are determined for over 250 cells across approximately 10000 total generations , which affords high statistical precision .
our data and model show that constriction is initiated early in the cell cycle and that its dynamics are controlled by the time scale of exponential longitudinal growth . based on our extensive and detailed growth and contour data ,
we develop a minimal mechanical model that quantitatively accounts for the cell shape dynamics and suggests that the asymmetric location of the division plane reflects the distinct mechanical properties of the stalked and swarmer poles .
furthermore , we find that the asymmetry in the division plane location is inherited from the previous generation .
we interpret these results in terms of the current molecular understanding of shape , growth , and division of _ c. crescentus_. cell shape both reflects @xcite and regulates @xcite biological function .
the importance of cell shape is exemplified by bacteria , which rely on specific localization of structural proteins for spatiotemporal organization @xcite .
bacteria take forms resembling spheres , spirals , rods , and crescents .
these shapes are defined by cell walls @xcite consisting of networks of glycan strands cross - linked by peptide chains to form a thin peptidoglycan meshwork @xcite .
super - resolution imaging is now revealing the internal positions of associated proteins @xcite .
these include cytoskeletal proteins such as mreb , a homolog of actin @xcite , intermediate filament - like bundles of cres ( crescentin ) @xcite , and ftsz , a homolog of tubulin @xcite .
however , due to the inherently stochastic nature of molecular processes , understanding how these proteins act collectively to exert mechanical stresses and modulate the effects of turgor pressure and other environmental factors requires complementary methods such as high - throughput , quantitative optical imaging .
multigenerational imaging data for bacterial cells can now be obtained from microfluidic devices of various designs @xcite .
still , a common limitation of most devices is that the environmental conditions change throughout the course of the experiment , particularly as geometric growth of the population results in crowding of the experimental imaging spaces .
we previously addressed this issue by engineering a _ c. crescentus _ strain in which cell adhesion is switched on and off by a small molecule ( and inducible promoter ) @xcite , allowing measurements to be made in a simple microfluidic device @xcite .
this technology allows imaging @xmath0100 generations of growth of an identical set of 250500 single cells distributed over @xmath125 fields of view .
thus cell density is low and remains constant .
these studies afforded sufficient statistical precision to show that single _
c. crescentus _ cells grow exponentially in size and divide upon reaching a critical multiple ( @xmath21.8 ) of their initial sizes @xcite .
satisfaction of a series of scaling laws predicted by a simple stochastic model for exponential growth indicates that these dynamics can be characterized by a single time scale @xcite . in this paper
, we use more advanced image analysis methods to extract cell shape contours from these data .
the resulting geometric parameters , together with mathematical models , provide insights into growth and division in _ c. crescentus _ and the plausible role of cell wall mechanics and dynamics in these processes .
specifically , we identify natural variables for tracking cell dynamics , and develop a minimal mechanical model that shows how longitudinal growth can arise from an isotropic pressure .
we then examine the dynamics of cell constriction and unexpectedly find that it is governed by the same time constant as exponential growth .
this important finding can be understood in terms of an intuitive geometric model that relates the constriction dynamics to the kinetics of the growth of septal cell wall .
we further suggest that the site of constriction can arise from differences in materials properties of the poles and show that it is established in the previous generation i.e . , the location of the site of division can be predicted before formation of the divisome .
we relate our results to the known dynamics of contributing molecular factors and existing models for bacterial growth and division . |
one of the major challenges in the study of the cosmic microwave background ( cmb ) at present , is to overcome the problem of separating the cosmological signal from the galactic foreground emissions , the noise and the extragalactic point sources , in the future high resolution cmb maps .
such missions as map ( bennett et al .
1996 ) and planck ( mandolesi et al . 1998 ; puget et al . 1998 ) will produce cmb maps that will be seriously affected by this problem .
several methods have already been tested to perform the separation of the different emissions in cmb observations , as the ones based on the wiener filter ( tegmark and efstathiou 1996 , bouchet and gispert 1999 ) and on maximum entropy methods ( hobson et al .
1998 , 1999 ) .
recently , wavelet techniques have been introduced in the analysis of cmb data .
denoising of cmb maps has been performed on patches of the sky of @xmath5 using multiresolution techniques ( sanz et al .
1999a ) and 2d wavelets ( sanz et al .
1999b ) , as well as on the whole celestial sphere ( tenorio et al .
1999 ) .
extragalactic point sources should be removed from the maps before any analysis is performed .
there are still many uncertainties regarding the composition of this population , the characteristics of the sources , and their time variability as well as the abundance and emission at different frequencies .
simulations of the emission of extragalactic sources at the observing frequencies of the planck mission have been worked out by toffolatti et al .
1998 and guiderdoni 1999 . in general
there is agreement between these extrapolations . in this paper
we use simulations based on the predictions of toffolatti et al .
the number counts of bright objects at frequencies above 300 ghz are dominated by far - ir selected sources .
lower frequencies will get most of the emission from radio selected sources .
removal of point sources in the observed maps is usually done by removing all pixels above a threshold of @xmath6 .
the error in rejecting pixels not corresponding to point sources is very small .
however , large number of point sources will still remain undetected .
tegmark and oliveira - costa 1998 ( toc98 hereinafter ) presented a method to locate and detect point sources based on the minimization of the variance of the map .
the application of this technique assumes that the cmb , noise and galactic foregrounds behave as gaussian fields and it requires knowledge of the power spectrum of these components . as a product of
the mem used to separate all the foreground components from the cmb , hobson et al .
1999 are able to recover the point sources present in the simulated maps included as part of the noise in the method .
however they are not able to remove the brightest sources that can still be observed in the residuals .
as will be shown these will be the point sources that the method presented here is able to extract .
tenorio et al .
1999 applied daubechies wavelets to detect and remove point sources from cmb plus noise flat maps .
however daubechies wavelets are not the optimal ones to detect point sources as explained below . as part of the effort to explore the possibilities of using wavelets to extract the information contained in cmb observations
, we are studying in this paper isotropic wavelets in the context of this data .
in particular , taking advantage of the spherical symmetry , these wavelets are very appropiate for locating and detecting point sources .
the method does not assume any behaviour for the rest of the signals and noise present in the maps .
the basic point source characteristics are its gaussian shape and scale , defined by the beam used in the observations .
convolution of the cmb map with a wavelet of the same scale and similar shape of the point source will produce wavelet coefficients with maxima ( amplification ) at the position of point sources ( location ) .
the most appropiate wavelet to detect gaussian shaped sources is the so called mexican hat .
this wavelet has been used to detect sources in the presence of noise in x - ray images by damiani et al .
the paper is organized as follows .
an analytical approach to the use of isotropic wavelets to analyse cmb maps is presented in section 2 .
section 3 is dedicated to the application of isotropic wavelets to cmb maps in order to detect point sources . a comparison with the toc98 method is also included in this section .
we present the flux estimation of the detected point sources in section 4 . wherever flux magnitudes are given
they refer to the total integrated flux under the beam .
conclusions are included in section 5 .
the continous isotropic wavelet transform of a 2d signal @xmath7 is defined as @xmath8 where @xmath9 is the wavelet coefficient associated with the scale @xmath10 at the point with coordinates @xmath11 .
@xmath12 is the `` mother '' wavelet that is assumed to be isotropic and satisfies the constraint @xmath13 and the admissibility condition : @xmath14 , where @xmath15 is the fourier transform of @xmath16 .
this is a necessary and sufficient condition in order to synthesise the function with the wavelet coefficients @xmath17 our aim is to locate point sources in microwave maps .
the wavelet coefficients given by equation ( 1 ) , provide information about the contribution of different scales ( @xmath10 ) to the value of the analyzed function ( @xmath18 ) at a certain location ( @xmath19 ) .
moreover , they will increase as the shape of the wavelet ( @xmath20 ) gets closer to the shape of the analyzed function .
we want to maximize the wavelet coefficients at the location of point sources .
as observations are performed through antennas , the observed point sources will be the result of a convolution with a certain antenna response function .
the response is most frequently modeled as a gaussian .
the most appropiate isotropic wavelet to use in this case is the so called mexican hat wavelet given by : @xmath21 the convolution of the mexican hat wavelet with an image consistent of point sources filtered with a gaussian beam , is a functional proportional to the second derivative of the analyzed function and therefore related to the maxima and minima in the analyzed image .
moreover the wavelet coefficients corresponding to a constant background will be identically zero .
we would also like to point out the resemblance between the mexican hat wavelet and the filter used in toc98 who obtained the optimal filter to detect point sources .
observations of the cmb contain not only the cosmological cmb signal but also galactic emission ( synchrotron , free - free and dust ) , noise and extragalactic emission ( sunyaev - zeldovich effect and point sources ) .
we are interested in removing point sources ( some of the sunyaev - zeldovich sources will be extended and we will deal with the removal of these in a future paper ) from the underlying signal ( cmb or galactic emissions ) plus noise image . if one assumes that the sources that will appear in real maps at different frequencies correspond to the convolution of a point source with a gaussian of dispersion @xmath22 ( antenna beamwidth ) , then any source ( we are only concerned with the detection of point sources in this paper and will therefore use the word source to refer to the filtered point sources unless otherwise indicated ) can be represented by a function : @xmath23 , @xmath24 being the flux of the point source and @xmath25 the area under the beam
. at the position of the peak of the source one has the following wavelet coefficient at scale @xmath10 : @xmath26 whereas the wavelet variance of the signal plus noise at scale @xmath10 is given by the expression @xmath27\ ; , \ ] ] where @xmath28 are respectively the signal dispersion in real space and the pixel scale @xmath29 ( @xmath30 being the pixel size ) , and it is assumed that @xmath31 .
@xmath32 is the signal to noise ratio of the signal plus noise map .
the signal has been represented by a power spectrum with effective index @xmath33 .
consider the amplification suffered by the flux of a source as the ratio of the detection level in wavelet space @xmath34 to the detection level in real space @xmath35 .
this quantity will be maximum at @xmath36 as the wavelet at that scale selects the contribution of scales around the size of the source whereas the signal and noise appear at a higher or lower scale , respectively . in theory
, one can calculate the amplification taking into account equations ( 4 ) and ( 5 ) .
the amplification factor is obtained to be greater than 2.3 for cmb dominated maps and larger in the case of dust dominated maps .
this result is also obtained numerically as will be shown in the next section .
isotropic wavelets are defined as a function of position @xmath19 and scale @xmath10 . as pointed out in the previous section , the optimal wavelet scale in order to detect point sources in cmb maps
is approximately the dispersion of the antenna @xmath37 .
convolution with a mexican hat of that scale will produce peaks at the positions of point sources and will allow for point source detection .
our aim is to illustrate the performance of the detection technique considering representative maps of high sensitivity experiments .
we have chosen to simulate maps following the planck mission characteristics .
the detection method is applied to simulations of flat patches of the sky of @xmath38 at two different frequencies , 100 and 857 ghz , with angular resolution ( arcmin)=10 ( 100 ghz ) , 5 ( 857 ghz ) and pixel size ( arcmin)=3 ( 100 ghz ) , 1.5 ( 857 ghz ) . as previously pointed out
, cmb observations will contain the cosmological signal , galactic foregrounds emissions , extragalactic emissions and instrumental noise . in the frequency range that the planck mission will sample ,
the observed maps will be dominated either by the cosmological cmb signal at low frequencies ( below @xmath39 ghz ) or by dust at high frequencies .
as two representative extreme cases we have chosen @xmath40 and @xmath41 ghz to present the detection method based on isotropic wavelets
. therefore simulations at 100 ghz include cosmological signal , point sources and instrumental noise and simulations at 857 ghz consist of dust emission , point sources and instrumental noise .
cosmological signal simulations are performed of cold dark matter ( cdm ) model with @xmath42 .
dust simulations are made using the predicted full - sky maps of finkbeiner et .
al . 1999 .
these predictions assume that dust emission corresponds to two grey bodies , with an emissivity of 1.67 for the cold one and 2.70 for the hot one ( this has been checked in the frequency range of firas ) .
point source simulations at the two frequencies considered are based on the code developed by toffolatti et al .
non correlated gaussian noise is added to the signal with a @xmath43 per resolution element of @xmath44 at 100 ghz and @xmath45 at 857 ghz as in the corresponding two channels of the planck mission . to show how different mexican hat widths affect the value of the wavelet coefficients at positions in the simulated map with or without a point source , we have plotted in figure 1 the value of the wavelet coefficient normalized to the dispersion of the wavelet coefficient map as a function of the mexican hat width @xmath10 .
it is clear that pixels with a point source have a characteristic curve with a maximum at @xmath36 . taking this behaviour into account , the detection method proceeds in three steps .
first the simulated map is convolved with mexican hat wavelets of several widths ( r ) obtaining maps of wavelet coefficients at each r. in the second step we look at the map of wavelet coefficients obtained for @xmath46 ( scale at which the amplification is maximum ) and find the spots ( connected pixels ) appearing above @xmath47 .
the location of the maxima of these spots will correspond to the location of point sources as previously discussed . in order to make sure that all detections correspond to real point sources we determine , in the third step , the numerical curve of @xmath48 as a function of @xmath10 ( being @xmath49 the position of a maximum identified in the second step ) and compare it with the theoretical one given by equations ( 4 ) and ( 5 ) .
a reduced @xmath50 is calculated and only maxima positions with values of @xmath51 are accepted as point source detections . we have plotted in figure 2 the number of point sources detected above a certain flux together with the number of point sources that exist in the simulation after filtering ( notice that the number of point sources above a certain flux can change when a gaussian filter is applied to the simulated map , in particular at high frequencies and low fluxes ) . as one can see the detected point sources are the ones with the highest fluxes .
the method is best suited to detect the bright sources leaving the majority of the faint ones undetected . at @xmath41 ghz
the number of point sources detected using the wavelet method is @xmath1 of the ones that exist in the simulation with fluxes above @xmath0 jy . at 100 ghz
we are able to detect @xmath3 of all the point sources existing above @xmath2 jy .
we have also looked at the amplification suffered by the detected point sources ( once the point source is located we compare the amplitude of that pixel in the wavelet coefficients map relative to the dispersion of that map with the amplitude of that pixel in the original ( signal plus noise plus point sources ) map relative to the dispersion of that map ) obtaining average values of 11 and 4 at 857 ghz and 100 ghz respectively .
we have compared our results with the method used in toc98 . to illustrate how both methods perform on the same data we have considered the same simulations at 100 ghz and 857 ghz .
the number of point sources detected by the two methods are given in table 1 ( also included in the second column is the number of point sources directly found above @xmath52 in the signal plus noise plus point sources map ) .
as one can see , the performance of both methods is comparable .
moreover the average amplification suffered by the detected point sources is also comparable .
the detection is clearly improved in relation to considering only peaks above @xmath6 .
this improvement is a consequence of the amplification suffered by the amplitude of the point sources after applying the mexican hat wavelet ( with the appropiate scale ) .
it is important to note that no assumptions about the nature of the underlying signal plus noise field are made in the wavelet method presented in this paper .
however , the detection method of toc98 is based on modeling the signal plus noise present in the analyzed maps as gaussian random fields and requires a knowledge of the power spectrum of all of the signal plus the noise components .
c|c|c|c frequency ( ghz ) & above @xmath52 & wavelet & toc98 857 & 8 & 35 & 32 100 & 0 & 3 & 3
in this section we want to show how the flux of the detected point sources can be recovered using wavelet coefficients . from equation ( 4 ) we know the analytical expression of the wavelet coefficient at the position of the detected point source , being proportional to the filtered flux of the point source @xmath24 .
moreover , to take into account the information at different scales we obtain the final amplitude value from the @xmath50 fit presented in section 3 . defining the error in the flux estimation as the ratio of the difference between the real and the recovered value divided by the real value , we have plotted in figure 3 these errors for the point sources detected in the two simulated maps . at 857 ghz point sources with fluxes above @xmath0 jy have errors with absolute values below @xmath53 , and @xmath54 of all detections have errors with absolute values below @xmath55 .
the three point sources detected at 100 ghz have errors with absolute values below @xmath4 .
as one can see from the figure the overestimated amplitudes have larger errors than the underestimated ones ( this can only be seeing in the 857 ghz results as the statistics of detected point sources is very poor at 100 ghz ) .
there is also a slight bias towards negative error values .
the @xmath50 fit used to recover the amplitude does not take into account possible correlations between the different scales . including these correlations
improves the amplitude determination method so that no bias appears .
this will be presented in a future paper .
once the position and the flux of the detected point sources is known , one can simulate a map with only point sources filtered with the corresponding gaussian beam ( depending on the frequency ) and subtract it from the original map . a map with point sources removed
will be obtained .
this will be an improvement on the map obtained by setting to 0 the value of the pixels affected by the detected point sources as is usually done .
wavelet coefficients provide information at each location of an analyzed image of the contributions from the different scales . moreover , as they are obtained from the convolution of the image with the wavelet , in the case that the shape of the wavelet coincides with the shape of the field at a fixed position
, the wavelet coefficient will be maximum .
since we are interested in extracting point sources from cmb maps , isotropic wavelets have the appropiate shape .
cmb observations are performed using antennas with a characteristic response .
modeling the antenna as a gaussian beam the observed point sources will be the result of convolving with such a beam . in this case ,
the best suited isotropic wavelet to detect point sources is the mexican hat .
this wavelet has a very similar shape to the optimal filter used by toc98 . as a demonstration of the method
, the study is performed on simulations at 100 ghz and 857 ghz dominated by cmb and dust respectively and with the characteristics of the future planck mission .
we have used the mexican hat wavelet to detect and determine the flux of point sources in @xmath56 square degree simulations .
the detection method is based on the maxima reached by the wavelet coefficients at @xmath36 at point source positions .
the reason for this is that the signal and the noise have characteristic scales larger and smaller than @xmath57 respectively and the wavelet used has the appropiate shape to resemble that of a point source filtered with a gaussian .
we consider detections above 5 @xmath58 .
making use of the information provided by the wavelet coefficients obtained convolving with mexican hat wavelets of different widths ( r ) as well as of the analytical expression we make sure that all detections correspond to real point sources by a reduced @xmath50 fit .
this method detects point sources in the simulated maps above @xmath59 jy at 857 ghz and above @xmath60 jy at 100 ghz .
@xmath1 of the existing point sources above @xmath0 jy are detected in the simulation at 857 ghz whereas all the existing point sources above @xmath2 jy are detected at 100 ghz .
the detections are achieved due to the amplification suffered by the ratio of the flux of the point source in the wavelet coefficient map ( at @xmath46 ) relative to the dispersion of that map in comparison with the flux in the original simulated map relative to its dispersion .
average amplification values of 11 and 4 are found at 857 ghz and 100 ghz .
it is very important to note that the detection method is not based on any assumption about the nature of the signal plus noise .
the method presented in toc98 provides comparable results as presented in section 3 .
toc98 method uses the optimal filter to locate point sources in cmb maps however a price has to be paid : the signal and noise are assumed to be gaussian random fields and a knowledge of the power spectrum is required .
we would also like to point out the possibility of using this method as complementary to the mem presented in hobson et al .
the bright point sources could be removed from the maps before mem is applied .
the rest of the point sources still in the analyzed maps will be recovered as a product of the separation method as explained in their paper .
finally , based on the analytical knowledge of the wavelet coefficients at positions dominated by point sources we recover the flux of the detected point sources . for the extracted point sources with fluxes above @xmath0 jy at 857 ghz
the errors ( absolute value ) are below @xmath53 .
the flux of the point sources detected in the simulation at 100 ghz have errors ( absolute value ) below @xmath4 .
removal of these point sources from the original map can therefore be done by simulating maps of only point sources at the locations found in the detection procedure and subtracting them from the initial ones .
we would also like to point out that an analysis of the whole sphere could also be performed by just dividing the sphere in small flat patches .
this paper is aimed to serve as a presentation of the power of isotropic wavelets to detect sources in cmb maps .
no information on different frequencies has been used in this work . in a future paper
we will add this information and will study the separation of sunyaev - zeldovich sources from point sources in planck simulated maps .
this work has been supported by the comisin conjunta hispano - norteamericana de cooperacin cientfica y tecnolgica ref .
98138 , spanish dges project no .
pb95 - 1132-c02 - 02 , spanish cicyt accin especial no .
esp98 - 1545-e and the eec project intas - open-97 - 1992 .
thanks the cfpa hospitality during may 1999 .
j.l.s . acknowledges partial financial support from a spanish mec fellowship and cfpa and thanks berkeley astronomy dept .
hospitality during year 1999 .
thanks finantial support from spanish mec in the form of a predoctoral fellowship and from the pparc in the form of a research grant .
pv acknowledges support from an universidad de cantabria fellowship .
lt acknowledges partial financial support from the italian asi and cnr .
jmd acknowledges support from a spanish mec fellowship fp96 20194004 .
bennett , c. et al .
1996 ; map homepage http://map.gsfc.nasa.gov bouchet , f.r .
& gispert , r. 1999 , astro - ph/9903176 damiani , f. , maggio , a. , micela , g. & sciortino , s. 1997 , apj , 483 , 350 finkbeiner , d.p . ,
davis , m. & schlegel , d.j .
1999 , astro - ph/9905128 guiderdoni , b. 1999 , astro - ph/9903112 hobson , m.p . ,
jones , a.w . ,
lasenby , a.n . & bouchet , f.r .
1998 , mnras , 300 , 1 .
hobson , m.p . ,
barreiro , r.b . ,
toffolatti , l. , lasenby , a.n . ,
sanz , j.l .
, jones , a.w . & bouchet , f.r .
1999 , mnras , 306 , 232 .
mandolesi , n. et al .
1998 ; proposal submitted to esa for the planck low frequency instrument puget , j.l .
et al . 1998 ; proposal submitted to esa for the planck high frequency instrument sanz , j.l . ,
argeso , f. , cayn , l. , martnez - gonzlez , e. , barreiro , r.b . & toffolatti , l. 1999a , mnras , 309 , 672 .
sanz , j.l . ,
barreiro , r.b .
, cayn , l. , martnez - gonzlez , e. , ruiz , g.a . ,
daz , f.j . ,
argeso , f. , silk , j. & toffolatti , l. 1999b , a&as , 140 , 99 .
tegmark , m. & efstathiou , g. 1996 , mnras , 281 , 1297 tegmark , m. & oliveira - costa , a. 1998 , apj , 500 , 83 tenorio , l. , jaffe , a.h . , hanany , s. & lineweaver , c.h . 1999 , astro - ph/9903206 .
toffolatti , l. et al .
1998 , mnras , 297 , 117 | it is the aim of this paper to introduce the use of isotropic wavelets to detect and determine the flux of point sources appearing in cmb maps .
the most suited wavelet to detect point sources filtered with a gaussian beam is the mexican hat .
an analytical expression of the wavelet coefficient obtained in the presence of a point source is provided and used in the detection and flux estimation methods presented . for illustration
the method is applied to two simulations ( assuming planck mission characteristics ) dominated by cmb ( 100 ghz ) and dust ( 857 ghz ) as these will be the two signals dominating at low and high frequency respectively in the planck channels .
we are able to detect bright sources above @xmath0 jy at 857 ghz ( @xmath1 of all sources ) and above @xmath2 jy at 100 ghz ( @xmath3 of all ) with errors in the flux estimation below @xmath4 .
the main advantage of this method is that nothing has to be assumed about the underlying field , i.e. about the nature and properties of the signal plus noise present in the maps .
this is not the case in the detection method presented by tegmark and oliveira - costa 1998 .
both methods are compared producing similar results .
= = = = = = = = # 1 # 1 # 1 # 1 @mathgroup@group @mathgroup@normal@groupeurmn @mathgroup@bold@groupeurbn @mathgroup@group @mathgroup@normal@groupmsamn @mathgroup@bold@groupmsamn = `` 019 = ' ' 016 = `` 040 = ' ' 336 = " 33e =
= = = = = = = # 1 # 1 # 1 # 1 = = = = = = = = [ firstpage ] cosmology : cmb data analysis |
galaxies at @xmath1 exhibit very diverse properties : they range from the blue lyman break galaxies ( lbgs ) which are bright in the rest - frame ultraviolet ( steidel et al 1996a , b ) to the distant red galaxies ( drgs ) which are generally faint in the rest - frame uv and have fairly red rest - frame optical colors .
the drgs are selected in the observers near - infrared ( nir ) by the simple criterion @xmath17 ( franx et al .
2003 , van dokkum et al .
the variety in the galaxy population at @xmath1 is comparable to that seen in the local universe , where colors range from very blue for young starbursting galaxies to very red for old elliptical galaxies .
an urgent question is what causes the red colors of drgs at @xmath1 .
are they `` old and dead '' , or are they actively star forming and more dusty than u - dropout galaxies ? first analyses of the optical - to - nir spectral energy distributions ( seds ) and spectra have suggested that both effects play a role : they have higher ages , contain more dust , and have higher mass - to - light ( @xmath18 ) ratios in the rest - frame optical than lbgs ( franx et al .
2003 , van dokkum et al .
2004 , forster schreiber et al . 2004 ; f04 ) .
furthermore , many have high star formation rates @xmath19yr@xmath20 ( van dokkum et al .
2004 , f04 ) .
unfortunately , the number of drgs with rest - frame optical spectroscopy is very small .
inferences inevitably depend on sed fitting , which has large uncertainties ( see e.g. , * ? ? ?
the sed constraints on the stellar and dust content are expected to improve significantly by extending the photometry to the rest - frame near - infrared . here
we present the first results on rest - frame nir photometry of drgs and lbgs in the hubble deep field south ( hdfs ) as observed with the infrared array camera ( irac ; fazio et al .
2004 ) on the _ spitzer space telescope_. where necessary , we assume an @xmath21 cosmology with @xmath22 km s@xmath20mpc@xmath20 , and a salpeter initial mass function ( imf ) between 0.1 and @xmath23 .
magnitudes are expressed in the ab photometric system unless explicitly stated otherwise .
we have observed the 5 arcmin@xmath24 hdfs / wfpc2 field with the irac camera integrating 1 hour each in the mid - infrared ( mir ) 3.6 , 4.5 , 5.8 , and 8@xmath25m@xmath16bands .
the observations , data reduction , and photometry will be described by labb et al .
( in preparation ) . briefly
, we reduced and calibrated the data using standard procedures ( e.g. , * ? ? ? * ) .
the limiting depths at 3.6 , 4.5 , 5.8 , and 8@xmath25 m are 25.0 , 24.6 , 22.6 , and 22.5 mag , respectively ( 5@xmath26 , 3@xmath27 diameter aperture ) .
we psf - matched the irac images and a very deep @xmath28image ( labb et al .
2003 ) to the @xmath4m@xmath16band .
nearby sources were fitted and subtracted to avoid confusion ( see labb et al .
we measured fluxes in @xmath29 diameter apertures .
the @xmath30 colors were then combined with the hdfs - catalog of labb et al .
( 2003 ) resulting in 11-band photometry from 0.3 to 8@xmath25 m .
in addition to the photometric errors , we add in quadrature an error of 10% to reflect absolute calibration uncertainties of irac .
we selected drgs in the field of the hdfs using the criteria of @xcite yielding 14 galaxies with photometric redshifts ( rudnick et al .
2003 ) in the range @xmath31 .
one blended source was excluded .
in addition , we selected isolated lbgs in the same field to the same @xmath32-band limit and in the same redshift range , using the u - dropout criteria of @xcite on the wfpc2 imaging @xcite .
the two samples are complementary , with only 1 source in common . for both samples , the mean
redshift is @xmath33 with a standard deviation of 0.5 ( f04 , wuyts et al . in preparation ) .
the estimate is based on direct comparison of 16 galaxies with both photometric and spectroscopic redshifts .
uncertainties in rest - frame magnitudes and model parameters are based on monte carlo simulations that take into account uncertainties in the observed fluxes and redshift estimates . ] .
in fig . [ fig.a ] we analyze the @xmath5 versus the @xmath6 m colors .
the @xmath34 m color has a sufficiently large baseline and much higher signal - to - noise ratio than the @xmath35 m or @xmath36 m colors .
the drgs and lbgs lie in very different regions , with little overlap .
the mean @xmath34 m color of the drgs is significantly redder than that of lbgs , confirming that drgs have higher @xmath18 ratios in the rest - frame optical .
the red @xmath5 and @xmath34 m colors of the drgs imply that they must be prominent in irac selected samples . indeed , the majority of red @xmath37 m sources selected by yan et al .
( 2004 ) in the goods survey satisfy the drg selection criteria .
we show color tracks for stellar population models ( bruzual & charlot 2003 ; bc03 ) , redshifted to a fixed @xmath33 : a single - age burst model ( ssp ) and two models with constant star formation ( csf ) , and reddenings typical for lbgs ( @xmath38 , shapley et al 2001 ) , and typical for drgs ( @xmath39 , f04 ) . the @xmath6 m colors of lbgs are consistent with low reddening models , whereas most of the drgs lie in the area of models with substantial extinction .
three of 13 drgs ( indicated by star symbols in fig .
1 ) lie well outside the area of constant star - forming models .
their @xmath5 colors are too red for their @xmath34 m color and they lie close to the line of an old single - age burst model with ages of @xmath40 gyr . the candidate old
, passive galaxies are shown in fig .
[ fig.b ] .
their seds are very well fit by an old single burst population .
the gray model curves are predictions based on fits to the exceptionally deep optical / nir data only , demonstrating that the spitzer fluxes lie very close to the expected values .
dead galaxies have blue @xmath34 m colors and with spitzer we can now effectively separate them from dusty star - forming drgs , which are red in @xmath34 m .
models with ongoing star - formation and dust - reddening fit the seds badly .
star - formation histories ( sfhs ) with exponentially declining star - formation rates ( sfrs ) and timescales @xmath41 myr can be ruled out at the 99.9% confidence level .
for the marginally acceptable models , the ratio of on - going to past - average sfr is @xmath42 indicating that these 3 galaxies are truly `` red and dead '' .
the best - fit ages and implied formation redshifts depend on the assumed metallicity @xmath43 , as expected .
super - solar @xmath44 models give a mean age of @xmath45 gyr and mean formation redshift @xmath46 , solar @xmath47 models yield @xmath48 gyr and @xmath49 , while sub - solar metallicity models fail to provide good fits to the irac fluxes as they are too blue .
hence we infer from the models that the `` dead '' galaxies had formed most of their stellar mass by @xmath50 ( @xmath44 ) or much earlier @xmath51 ( @xmath47 ) .
the number density of `` dead '' galaxies is 1.9 @xmath52 mpc@xmath53 , assuming a top hat redshift distribution between @xmath54 and 3 . for stellar masses
@xmath55 and the same imf , that number density is @xmath56 lower than that of early - type galaxies in the nearby universe ( bell et al .
2003 ) . obviously , the sampled volume is too small to allow firm conclusions , but the result is indicative of strong evolution of passive galaxies from @xmath57 to @xmath58 . in fig .
[ fig.b ] we show the seds for 3 of the 8 drgs whose optical / nir fluxes were better fit with csf models and reddening by a calzetti et al .
( 2000 ) dust law .
the observed mir flux points are often different from predictions based on shorter wavelengths , especially where the flux density @xmath59 was still rising at the observed @xmath28band .
hence , for the very dusty drgs ( @xmath60 ) the mir fluxes help to better constrain the age and dust in the models . the average ages and extinction of the fits changed mildly after inclusion of the irac data : from @xmath611.1 to 1.3 gyr , and @xmath621.9 to 1.5 mag , respectively .
finally , two remaining galaxies could not be fit well by any model , most likely due to emission line contamination .
one of them is spectroscopically confirmed as a very strong lyman-@xmath63 emitter ( wuyts et al . in preparation ) .
however , the generally good fits to the uv - to-8@xmath25 m seds of the drgs indicate that the red @xmath64 colors were caused by old age and dust - reddened star light , and not other anomalies .
we note that one of the dead galaxies ( see fig .
1 ) has an apparent flux excess at @xmath4 m , suggesting the presence of an obscured agn which starts to dominate the flux in the rest - frame @xmath14band ( see e.g. , @xcite ) .
we next study the @xmath18 ratios inferred from the sed fits described above , adopting the best fitting sfh ( ssp or csf ) .
figure 3a shows the modeled @xmath15 ( rest - frame @xmath13 ) versus rest - frame @xmath65 .
the curves show the dust - free bc03 model @xmath15 ratios .
galaxies generally lie to the red of the model curves as a result of dust attenuation .
the fits imply a large range in @xmath15 for drgs and lbgs together ( a factor of 6 ) .
furthermore , the @xmath15 ratio correlates with rest - frame @xmath65 color , where galaxies red in the rest - frame optical have higher @xmath15 ratios than blue galaxies .
the average @xmath15 of the drgs is @xmath66 , about three times higher than that of the lbgs .
figure 3b shows the derived @xmath15 against the derived age .
the @xmath15 correlates well with the age , as expected from the models .
the role of extinction is greatly reduced in the rest - frame @xmath14band , implying that the differences in the @xmath15 ratios for drgs and lbgs are driven by age differences .
figure 3c shows the @xmath15 ratio against stellar mass @xmath67 .
drgs dominate the high - mass end .
the highest - mass galaxies ( @xmath55 ) in this sample all have high @xmath15 ratios , and here the @xmath15 ratio does not depend strongly on @xmath67 . at lower masses galaxies have much lower @xmath15 ratios .
this may be a selection effect caused by our magnitude cut - off , as we would miss low - mass galaxies with high @xmath15 .
however , high - mass galaxies with _ low _ @xmath15 would be detected if they existed , hence the lower envelope of the @xmath15 versus @xmath67 distribution is real .
two intriguing possibilities are that the mean @xmath15 decreases towards lower stellar mass , or that the intrinsic scatter in @xmath15 increases .
our drgs and lbgs were selected in the observed @xmath28band ( rest - frame @xmath68band ) .
we find a large variation in @xmath69 ( a factor of 25 ) , hence selection effects play a major role in nir studies at high redshift .
selection using mir observations with spitzer would improve the situation , but the wide range in @xmath15 ratios and ages found here indicates that even a irac @xmath4m@xmath16selected sample would still be very different from a mass - selected sample .
we have presented rest - frame uv - to - nir photometry of a sample of drgs and lbgs at @xmath3 in the hdfs .
these galaxies span a wide range in properties , similar to low - redshift galaxies .
the rest - frame nir photometry from irac helps significantly : first , by allowing us to separate `` old and dead '' from dusty star forming drgs using only the observed @xmath70 , and @xmath71m - band , and second , by improving model constraints on the heavily obscured drgs .
the wide range in galaxy properties at @xmath57 raises several important issues .
first , it demonstrates it is impossible to obtain mass - selected samples photometrically . even in the rest - frame @xmath13-band
, the @xmath18 ratio varies by a factor of 6 for the drgs and lbgs in the hdfs .
second , we need to understand what produces these variations . if a relation between total stellar mass and @xmath15 exists ( see fig .
3c ) then stellar mass might be driving the variations .
deeper irac data is needed to establish this well , as incompleteness effects may play a role .
it is tantalizing that kauffmann et al .
( 2003 ) find a similar correlation at @xmath72 , with the most massive galaxies being the oldest and having the highest @xmath18 ratios .
the authors also found that above a stellar mass @xmath73 galaxy properties correlate only weakly with @xmath67 , similar to what we find at @xmath3 .
the simplest explanation is that we observe the same galaxies at @xmath72 and @xmath57 , although we note that hierarchical formation scenarios predict significant merging for galaxies at @xmath1 .
alternatively , we observe merging processes occurring at a critical mass of about @xmath74 .
a partial explanation may be a simple relation between mass and metallicity : higher mass galaxies might have higher metallicity , and thus be redder .
it is very noteworthy that there are `` red and dead '' galaxies at @xmath3 .
previous authors found such galaxies at @xmath75 ( e.g. , * ? ? ?
* ; * ? ? ?
whereas the number density of `` dead '' galaxies at @xmath57 is probably much lower than at @xmath58 , the mere existence of these systems at such high redshift raises the question what caused such a rapid and early decline in star formation .
our model fits imply that their star formation stopped by @xmath76 or higher , close to or during the epoch of reionization .
possibly a strong feedback mechanism caused this , such as an agn or galactic - scale outflow .
we note that these galaxies are among the most massive galaxies at @xmath57 , and hence were probably at the extreme tail of the mass function at @xmath76 .
obviously , many questions remain unanswered by this study .
only a very small field has been studied , and similar studies on wider fields are necessary .
finally , the @xmath18 ratios derived here remain model dependent and vary with the assumed sfh , metallicity , and imf ( e.g. , bell et al .
direct mass determinations are presently very challenging , but are becoming more important to understand the @xmath3 galaxy population .
this research was supported by the carnegie institution of washington , the netherlands foundation for research ( nwo ) , the leids kerkhoven - bosscha fonds , the lorentz center , and the smithsonian institution .
barmby , p. , et al .
2004 , , 154 , 97 bell , e. f. , et al .
2003 , , 149 , 289 bruzual , g. & charlot , s. 2003 , , 344 , 1000 ( bc03 ) casertano , s. et al . , 120 , pp . 27472824 , 2000 calzetti , d. , et al .
2000 , , 533 , 682 cimatti , a. , et al .
2004 , , 430 , 184 fazio , g. g. , et al .
2004 , , 154 , 10 franx , m. et al .
2003 , , 587 , l79 frster schreiber , n. m. , et al .
2004 , , 616 , 40 ( f04 ) kauffmann , g. , et al .
2003 , , 341 , 54 labb ' e , i. , et al .
2003 , , 125 , 1107 labb ' e , i. , 2004 , doctoral thesis , leiden university madau , p. , et al .
1996 , , 283 , 1388 mccarthy , p. j. , et al .
2004 , , 614 , l9 papovich , c. , dickinson , m. , & ferguson , h. c. 2001 , , 559 , 620 rudnick , g. , et al .
2003 , , 599 , 847 steidel , c. c. , et al .
1996 , , 112 , 352 steidel , c. c. , et al . 1996b , , 462 , l17 stern , d. , et al .
2005 , , submitted ( astro - ph/0410523 ) van dokkum , p. g. et al .
2003 , , 587 , l83 van dokkum , p. g. , et al .
2004 , , 611 , 703 yan , h. , et al .
2004 , , 616 , 63 | we present deep @xmath0 m imaging of the hubble deep field south with irac on the _ spitzer space telescope_. we study distant red galaxies ( drgs ) at @xmath1 selected by @xmath2 and compare them to a sample of lyman break galaxies ( lbgs ) at @xmath3 .
the observed uv - to-@xmath4 m spectral energy distributions are fit with stellar population models to constrain star formation histories and derive stellar masses .
we find that 70% of the drgs are best described by dust - reddened star forming models and 30% are very well fit with old and `` dead '' models .
using only the @xmath5 and @xmath6 m colors we can effectively separate the two groups .
the dead systems are among the most massive at @xmath7 ( mean stellar mass @xmath8 ) and likely formed most of their stellar mass at @xmath9 . to a limit of @xmath10
their number density is @xmath11 lower than that of local early - type galaxies .
furthermore , we use the irac photometry to derive rest - frame near - infrared @xmath12 and @xmath13 fluxes .
the drgs and lbgs together show a large variation ( a factor of 6 ) in the rest - frame @xmath14band mass - to - light ratios ( @xmath15 ) , implying that even a spitzer @xmath4m@xmath16selected sample would be very different from a mass - selected sample .
the average @xmath15 of the drgs is about three times higher than that of the lbgs , and drgs dominate the high - mass end .
the @xmath15 ratios and ages of the two samples appear to correlate with derived stellar mass , with the most massive galaxies being the oldest and having the highest mass - to - light ratios , similar as found in the low - redshift universe . |
deterministic photon - photon phase gate is a key building block to construct circuits for the scalable all - optical quantum information processing .
cross - phase modulation ( xpm ) between slow pulses in media under electromagnetically induced transparency ( eit ) conditions @xcite ( or with similar properties ) is the main route towards such a gate .
considerable theoretical developments ( see , e.g. , @xcite ) as well as experimental studies ( see , e.g. , @xcite ) have been undertaken to explore the eit - based xpm and alternative approaches aiming at realizing photonic two - qubit gates . in optics - based quantum computing , a conditional phase of @xmath0 radians needs to be implemented by photon - photon phase gates .
for achieving such a large phase there are the proposals @xcite of making pulses co - propagate , so that the generic weak pulse interaction could be compensated by the prolonged interaction time .
another requirement for an ideal phase gate is the uniformness of conditional phase , as given by the mapping @xmath1 , i.e. , the same phase @xmath2 is induced for each mode @xmath3 of a continuous - mode photon in the state @xmath4 , where @xmath5 is the pulse profile in wave - vector space .
the currently dominant understanding is that a homogeneous phase could be possible if the interaction between two pulses is averaged out by letting one pulse completely go through the other @xcite .
since a photon is not a point particle , photon - photon interactions in nonlinear media should be modeled as interacting quantum fields of continuous modes . using this picture ,
we show that the above - mentioned notions are generally invalid .
the essential effects in realistic single - photon xpm are clarified here for the first time .
we first provide a theoretical framework for the xpm between photons . the interaction between two pulses in a medium of any type of atomic structure realizing eit can be translated into that between two dark - state polariton fields @xmath6 ( @xmath7 ) with @xmath8=\delta_{i , j}\delta(k - k')$ ] @xcite . by neglecting the pulse loss and deformation , as well as the possible self - phase modulation term which has no effect on photon - photon interactions @xcite
, one has the following equations of motion for the slowly varying and transversely well confined polariton fields @xcite : @xmath9 where @xmath10 are the pulse group velocities .
the term @xmath11 could come from a general interaction potential @xmath12 .
the pulse interaction in the experimentally studied xpm thus far @xcite , for instance , can be modeled by a contact potential @xmath13 , where @xmath14 approximated by a real quantity is the nonlinear rate determined by the specific system parameters .
the potential @xmath12 considered here acts instantaneously ; see @xcite for a study on the non - instantaneous effects . from the field - theoretic viewpoint ,
( [ motion ] ) is obtained by the equation of motion @xmath15 for non - relativistic fields , where the hamiltonian @xmath16 consists of the kinetic term @xmath17 and the interaction term @xmath18 .
the pulse interaction would evolve the input state to @xmath19 , where @xmath20 ( @xmath21 denotes the time - ordering operation and @xmath22 is adopted hereafter ) .
it is convenient to use the _ two - particle function _ , @xmath23 , to study the evolution of the initial state @xmath24 . to indicate how close a realistic xpm is to the ideal one @xmath1 for phase gates
, we use the _ fidelity _ @xmath25 . as figures of merit to characterize a realistic xpm , the fidelity @xmath25 and conditional phase @xmath2 are determined by the overlap @xcite @xmath26 where @xmath27 is the two - particle function from the field operator equations @xmath28 , corresponding to the freely evolved state @xmath29 .
a similar formula in the discrete form is given in @xcite .
the output state of a realistic xpm , @xmath30 , is generally entangled between the wave - vector modes @xmath3 and @xmath31 of the individual photons .
such field modes entanglement was widely neglected in the previous researches on photon - photon gates , though its effect on xpm was conjectured in @xcite .
it is conceivable that the entanglement would lower the fidelity @xmath25 . yet , for clarifying the issue , a relation between the amount of entanglement generated in pulse interaction and the corresponding gate operation fidelity has to be found .
here we quantify the field mode entanglement with the linear entropy @xmath32 ( @xmath33 are the reduced density matrices of the bipartite state @xmath34 ) , which takes the following closed form : @xmath35 see appendix for the proof . compared with the schmidt decomposition method to characterize bi - photon entanglement @xcite , this formula provides an exact measure of such entanglement . in reality
the medium for pulse interaction carries a finite bandwidth @xmath36 , which is connected to the width of eit transparency window @xcite .
hence , the equal - time commutator for the field operator @xmath37 after being imposed a cut - off by the bound of the wave - vector mode @xmath38 becomes @xmath39&=&\delta_{ij}(k_s/\pi ) \mbox{sinc}\big ( k_s(z_1-z_2)\big)\nonumber\\ & \equiv & \delta_{ij}c(z_1-z_2 ) , \label{com}\end{aligned}\ ] ] where @xmath40 .
substituting the formal solution of ( [ motion ] ) , @xmath41 , where @xmath42 originates from the contact potential @xmath43 , into @xmath44 , while considering the commutator in ( [ com ] ) , yields the general two - particle function @xmath45 at the time @xmath46 , where @xmath47 and @xmath48 . the second term on the right side of eq .
( [ 2-p ] ) arises from photon - photon interaction .
the norm of the two - particle function @xmath49 is not preserved as a consequence of the deviation of the field operator commutator @xmath50 from the delta function @xmath51 , so the output two - photon function @xmath49 has to be normalized prior to calculating the fidelity and linear entropy using eqs .
( [ over - lap ] ) and ( [ entropy ] ) .
the system bandwidth @xmath36 effectively results in a non - unitary evolution , though the wave - vector modes @xmath3 , @xmath31 of two photons are still continuous after imposing the cut - off @xmath52 .
this could be understood by the restriction @xmath53 on the matrix elements @xmath54 of the evolution operator @xmath55 , causing the loss of the orthonormal relation for the matrix elements @xmath54 ; c.f . the inequality @xmath56 .
an important situation we analyze here is the xpm between two co - propagating pulses @xcite . with the relative velocity @xmath57 , the two - particle function in eq .
( [ 2-p ] ) reduces to ( see also ref .
@xcite ) @xmath58 where @xmath59 , @xmath60 and @xmath61 . using the normalized form of this two - particle function ,
one will obtain from eq .
( [ over - lap ] ) the following relations to determine the conditional phase and fidelity : @xmath62 @xmath63 where @xmath64 $ ] , @xmath65)^2 $ ] .
the dimensionless variables of the integrals are @xmath66 , where @xmath67 is the pulse size in medium ; the system parameter is defined as @xmath68 , in proportion to the ratio of the system bandwidth @xmath69 to the pulse bandwidth @xmath70 ( the reciprocal of the pulse duration ) .
the factor @xmath71 in the parameter @xmath72 reflects the pulse compression in eit media . in fig .
[ fig : fhi_vs_phi ] obtained from eq .
( [ c - phase ] ) , one sees two different xpm patterns depending on the values of @xmath73 and separated at the transitional point @xmath74 .
this @xmath73 value is the threshold above which the denominator @xmath75 on the right - hand side of eq .
( [ c - phase ] ) may become zero or negative with increasing @xmath76 , and it is universal with respect to any pulse shape . as @xmath73 decreases or increases across this point , the gate performance characterized by the achievable @xmath2 values will undergo a transition to the different pattern . for @xmath77 ,
the conditional phase @xmath2 can never assume values greater than @xmath78 no matter how large @xmath76 is . without loss of generality , in our discussion below we consider the interaction between two identical pulses in the gaussian profile @xmath79 .
then the boundary value @xmath74 of the two patterns in fig .
1 corresponds to the system parameter @xmath80 .
as @xmath72 increases ( decreases ) from this value , the coefficient @xmath73 will become smaller ( larger ) into the respective pattern .
( [ entropy ] ) and ( [ fidelity ] ) yield the evolution of the field - mode entanglement and gate operation fidelity with the phase @xmath76 , which is proportional to the pulse interaction time @xmath46 .
the results are illustrated in fig . 2 for various @xmath72 values .
the expected relation between @xmath81 and @xmath25 manifests in the lower regime of @xmath77 in fig .
1 , where a weaker entanglement , at any fixed @xmath76 value , is accompanied by a higher fidelity . in this regime
the entanglement between the field modes totally vanishes in the limit @xmath82 . in the upper regime of @xmath83
, the field mode entanglement disappears in the limit @xmath84 .
the fidelity value , however , goes down to zero at this point , for any non - zero @xmath76 value ; c.f .
( [ fidelity ] ) with a diverging @xmath85 at @xmath84 .
the non - unitary evolution due to a finite system bandwidth @xmath36 accounts for this phenomenon .
note that , as the consequence of non - unitary evolution , both factors @xmath86 , @xmath87 in the second term of eq .
( [ 2-p - e ] ) carry the same variable @xmath88 , thus deviating from the ideal output two - particle function @xmath89 in the neighborhood of @xmath84 .
the xpm between two evenly distributed square pulses shows the same characteristics except for the different relations between the @xmath90 and @xmath72 values , indicating that the effects described above differ , by nature , from the inhomogeneity in pulse interactions . an interesting observation from fig .
2 is the feature that , around the transitional point @xmath80 of the two operation patterns , the plots of @xmath81 , e.g. , the thin solid line , reaches the highest value at some @xmath91 .
increasing @xmath76 beyond the point leads to a continual distortion of the two - photon state without increasing its entanglement .
irrespective of the pulse shapes , the linear entropy plateau at the transitional point is correlated to the fidelity valley like that in fig .
3 , which goes down to @xmath92 at @xmath80 and @xmath93 . furthermore
, we comment on the notion that pulses moving with matched group velocities make large conditional phase possible @xcite . in the previous theoretical studies , the variables @xmath94 and @xmath95 of the two - particle function in ( [ 2-p - e ] ) were often mixed up with the pulse - center coordinates , and then the phase @xmath76 would be regarded as the conditional phase @xmath2 for the reason that the overlapped pulse centers with @xmath96 could lead to the ideal output @xmath97 from eq .
( [ 2-p - e ] ) .
in fact , the variable @xmath98 of the @xmath99 function in ( [ 2-p - e ] ) assumes any value even if the two pulses co - propagate , because @xmath100 are the field coordinates over the whole @xmath101 axis rather than those of the pulse centers .
the conditional phase @xmath2 should be determined by eq .
( [ over - lap ] ) giving its relation with the xpm phase @xmath76 in fig .
1 . the conditional phase value could reach @xmath0 in the upper regime of fig .
1 , where the gate operation fidelity is , however , rather low .
a high fidelity is possible only in the lowest region in fig .
1 , where the state evolution is close to unitary but the peak of the conditional phase @xmath2 is vanishing . the vanishing conditional phase in the regime of near unitary evolution also exists in case of interaction between co - propagating single photon and coherent state ; see @xcite .
the trade - off between conditional phase and fidelity in xpm between co - propagating photons is also discussed recently in @xcite , where a finite mode approximation for pulses is adopted .
one problem with the finite mode approximation is its underestimation of xpm intensity compared with the contribution from the second term in our eq .
( [ 2-p - e ] ) , the term arising from xpm , the discrete sum with the corresponding term in @xcite ( the second term in eq .
( 22 ) of @xcite ) contributes much less significantly to a coefficient similar to @xmath73 in this paper , limiting its value to less than @xmath102 .
then only the lower regime in our fig .
1 can be obtained in the finite mode approach . for two ultraslow pulses the small ratio @xmath103 in the parameter
@xmath72 makes the upper regime in fig . 1 more relevant .
the effect of non - unitary evolution of quantum states dominating in this regime is beyond the description by the finite mode approximation .
next we examine the xpm between pulses colliding head - on @xcite via a contact potential @xmath43 . for two ultraslow pulses satisfying @xmath104 , there is the approximation @xmath105 in ( [ 2-p ] ) , so the infinite sum in eq .
( [ 2-p ] ) can be approximated by a closed form in this regime . substituting the normalized two - particle function into ( [ over - lap ] ) , we obtain the fidelity evolution in the course of pulse interaction ; see the example in fig .
it shows that the fidelity value will decline once a pulse touches the other and stabilize again after they pass through each other .
the stronger the interaction ( indicated by the @xmath76 values ) is , the lower the fidelity will become after collision . against the intuitive notion that an averaged interaction on pulses could generate a uniform conditional phase @xmath2 , a realistic xpm between pulses of a non - zero relative velocity can be far away from the ideal process @xmath1 .
an ideal performance of phase gates based on the xpm described above exists in the limit of an infinite system bandwidth @xmath36 . in this limit
the two - particle function after two pulses completely going through each other is @xmath106 , where @xmath107 is a fixed value from the contact potential @xmath43 .
this result is also true to the xpm between single photon and coherent state @xcite .
the time - dependent two - particle functions for such idealized unitary evolution under a general interaction potential @xmath12 take the form @xmath108 @xcite .
the field mode entanglement exhibited by the possibly non - factorisable @xmath109 with respect to its spatial variables is therefore the main factor that determines the gate operation fidelity in a regime of approximately unitary evolution , which could be realized under the condition @xmath110 .
we have illustrated the effects of continuous field mode entanglement arising from pulse interaction and non - unitary evolution caused by finite system bandwidth , which drastically impair photon - photon phase gate performance .
these effects induce more complexity in xpm than what was previously understood . due to their possible existence in any device working with quantum objects of continuous degrees of freedom , the proper handling of the effects could be a major concern in quantum technology .
we provide a brief derivation for the linear entropy formula in eq .
( [ entropy ] ) .
the elements of one reduced density matrix @xmath111 for a general bipartite state @xmath112 can be obtained by the following @xcite ( its discrete form is given in @xcite ) : @xmath113 the matrix elements of @xmath114 can thus be obtained by substituting the above into @xmath115 , which leads to a closed form of the linear entropy @xmath116 .
99 h. schmidt and a. imamoglu , opt
* 21 * , 1936 ( 1996 ) .
m. fleischhauer , a. imamoglu , and j. p. marangos , sec .
phys . * 77 * , 633 ( 2005 ) .
s. e. harris and l. v. hau , , 4611 ( 1999 ) m. d. lukin and a. imamoglu , phys .
lett . * 84 * , 1419 ( 2000 ) .
d. petrosyan and g. kurizki , phys .
a * 65 * , 033833 ( 2002 ) .
m. masalas and m. fleischhauer , , 061801(r ) ( 2004 ) .
d. petrosyan and y. p. malakyan , , 023822 ( 2004 ) .
a. andre , m. bajcsy , a. s. zibrov , and m. d. lukin , phys .
. lett . * 94 * , 063902 ( 2005 ) .
i. friedler , d. petrosyan , m. fleischhauer , and g. kurizki , phys .
a * 72 * , 043803 ( 2005 ) .
s. rebic , d. vitali , c. ottaviani , p. tombesi , m. artoni , f. cataliotti , and r. corbalan , 032317 ( 2004 ) .
wang , k .-
marzlin , and b. c. sanders , phys .
lett . * 97 * , 063901 ( 2006 ) .
marzlin , z .- b .
wang , s. a. moiseev , and b. c. sanders , j. opt .
b * 27 * , a36 ( 2010b .
he , a. macrae , y. han , a. i. lvovsky , and c. simon , phys .
a * 83 * , 022312 ( 2011 ) .
e. shahmoon , g. kurizki , m. fleischhauer , and d. petrosyan , phys .
a * 83 * , 033806 ( 2011 ) .
b. he , q. lin , and c. simon , , 053826 ( 2011 ) .
a. rispe , b. he , and c. simon , , 043601 ( 2011 ) .
l. v. hau _
_ , nature ( london ) * 397 * , 594 ( 1999 ) . h. kang and y. zhu , phys . rev
91 * , 093601 ( 2003 ) .
chen , c .- y .
wang , s .- h .
wang , and i. a. yu , phys .
lett . * 96 * , 043603 ( 2006 ) .
s. li , x. yang , x. cao , c. zhang , c. xie , and h. wang , phys .
lett . * 101 * , 073602 ( 2008 ) .
lo , p .- c .
su , and y .- f .
chen , , 053829 ( 2010 ) .
lo , y .- c .
chen , p .- c .
su , h .- c .
chen , j .- x .
chen , y .- c .
chen , i. a. yu , and y .- f .
chen , phys .
a * 83 * , 041804(r ) ( 2011 ) .
shiau , m .- c .
wu , c .- c .
lin , and y .- c .
chen , , 193006 ( 2011 ) .
our treatment as well applies to the situations with one of the pulses being a photon at the speed @xmath118 .
j. h. shapiro , phys .
a * 73 * , 062305 ( 2006 ) .
j. gea - banacloche , , 043823 ( 2010 ) . c. k. law , i. a. walmsley , and j. h. eberly , , 5304 ( 2000 ) .
c. k. law and j. h. eberly , , 127903 ( 2004 ) .
m. srednicki , , 666 ( 1993 ) .
b. he and j. a. bergou , , 062328 ( 2008 ) . | the effects arising from the inherent continuous - mode nature of photonic pulses were poorly understood but significantly influence the performance of quantum devices employing photonic pulse interaction in nonlinear media .
such effects include the entanglement between the continuous wave - vector modes due to pulse interaction as well as the consequence of a finite system bandwidth .
we present the first analysis on these effects for interactions between single - photon pulses , demonstrating their impact on the performance of quantum phase gates based on such process .
our study clarifies a realistic picture of this type of quantum devices . |
the reionization of the intergalactic medium ( igm ) by the first stars and quasars was likely a complex process .
current evidence from observations of polarization of the cosmic microwave background implies a redshift for reionization of @xmath16 @xcite , while tracers at lower redshift suggest that the tail end of the process was still ongoing at redshifts @xmath17 @xcite .
gamma - ray bursts ( grbs ) offer one of the most promising probes of this epoch due to their bright afterglows , which can be seen at cosmological distances and lack some of the complications of quasar observations ( e.g. , @xcite ) . in this _ letter _ , we present our gemini spectrum of the afterglow of the long - duration grb 140515a . it reveals a sharp decrement in flux shortward of @xmath188900 , implying a redshift of @[email protected] @xcite , which we refine below .
this makes grb 140515a the burst with the third largest spectroscopic redshift . only grbs 080913 ( @xmath19=6.733 ; @xcite ) and 090423 ( @[email protected] ; @xcite ) are known to be more distant , although grb 090429b has a photometric redshift of @xmath189.4 @xcite .
it is important to note that at the time of our gemini observations , no optical observations or photometric redshift estimates had been reported in the gcn circulars , so the high-@xmath19 nature of the burst was purely serendipitous .
we describe our spectroscopic observations in section 2 and analyze the spectrum to set constraints on the opacity of the igm at @xmath21 in section 3 .
we fit the red damping wing of ly@xmath0 with three fiducial models in section 4 to set limits on the neutral fraction of the igm .
we conclude in section 5 by placing these observations in the context of other constraints on reionization .
grb 140515a was detected by _
swift _ at 09:12:36 ut on 2014 may 15 @xcite .
we initiated followup observations with the gemini multi - object spectrograph ( gmos ; @xcite ) on the 8 m gemini - north telescope at 10:48 ut and discovered a faint optical source consistent with the x - ray afterglow position @xcite in our @xmath22 and @xmath19 acquisition images @xcite .
we obtained a dithered pair of 900 s spectra starting at 11:42 ut on 2015 may 15 ( 2.5 hr after the bat trigger ; @xcite ) . during the observations , the source position increased in airmass from 2.38 to 3.00 .
we oriented the 1@xmath23-wide slit at a position angle of 78@xmath243 , which was within 5@xmath25 of the parallactic angle , so differential slit losses should be minimal .
we used the r400 grating with an og515 filter to cover the wavelength range 588810162 .
the gmos focal plane has three abutting ccds and we were only able to obtain observations with a single grating angle , so our data have two gaps at 72757298 and 87098737 .
in addition to the high airmass , the moon was @xmath2699% illuminated , so the sky background was extremely high .
fortunately , the seeing was exceptional , with a full - width at half - maximum ( fwhm ) of 0@xmath2745 in the acquisition images @xcite and even in our final spectrum , the spatial fwhm of the object trace was only 0@xmath276 .
the spectral fwhm of night sky emission lines are @xmath187 , but due to the excellent seeing , the object did not fill the slit and the resolution of the object spectrum is @xmath185
. we use iraf to perform basic two - dimensional image processing , remove cosmic rays @xcite , and shift and stack the two - dimensional frames prior to spectral extraction .
we apply flux calibrations and correct for telluric absorption using our own idl procedures and archival observations of standard stars taken with a similar setup .
the airmass of our data was substantially higher than any suitable standard star observations that we are able to locate in the gemini archive , so the overall spectral shape of the final spectrum is somewhat uncertain , although this does not affect any of our analysis .
in addition , we are careful to manually scale the correction for telluric h@xmath28o absorption near 9400 because of the large airmass difference with the available standards .
we also correct the spectrum for @xmath29 mag of galactic extinction @xcite .
the final extracted spectrum is shown in figure [ specfig ] .
the sharp decrement in flux shortward of 8910 is evidence of ly@xmath0 absorption at @xmath1 .
we fit a power law to the continuum flux at 90209870 , while excluding the region of the most uncertain telluric correction near 9400 and find a best fit of @xmath30@xmath31@xmath32 , which is shown as the solid green line .
the value for the slope has additional uncertainty due to low - level telluric absorption over most of the available continuum .
we search the afterglow continuum for narrow absorption features from the host galaxy interstellar medium but find no plausible matches at wavelengths compatible with the approximate redshift of ly@xmath0 , so we are unable to make a precise measurement of the redshift of grb 140515a ( @xmath33 ) , a point that we will return to below .
instead , we find unresolved absorption lines at 8985.2 , 9000.0 , and 9697.9 that are consistent with the @xmath34 1548 , 1551 doublet and @xmath351670 from an intervening absorber at @xmath36 .
a possible weak absorption feature at 9170.4 , with an observed - frame equivalent width ( @xmath37 ) of [email protected] , does not match any other expected line from the absorber or the host galaxy3@xmath275 away from the grb afterglow .
it exhibits a single spectrally - resolved emission line at 8200 , which we identify as [ ] @xmath353727 at z=1.200 .
no strong absorption lines from that galaxy are expected in our limited continuum wavelength interval . ] .
we first examine the transparency of the igm to ly@xmath0 photons over the redshift range @xmath38 . the signal - to - noise ratio ( s / n ) of these data ( peaking in the continuum at @xmath1820 per 1.38 pixel near 9000 ) are not as high as for grb 130606a @xcite or for typical high-@xmath19 quasar observations but still place meaningful constraints .
we inspect the sky - subtracted two - dimensional spectra and find no clear transmitted continuum over the wavelength range 81858905 ( corresponding to redshifts of 5.7336.325 relative to ly@xmath0 ) .
we follow @xcite and define an effective optical depth to ly@xmath0 , @xmath39 , where @xmath40 is the average transmission relative to the continuum in a bin . to compare to previously reported results from @xmath17 quasars and grb 130606a ,
we compute @xmath41 in bins of width @xmath42 , taking care to define our wavelength intervals to avoid ly@xmath43 and the ccd gaps .
we also measure the transmission in the ly@xmath43 forest and convert these to effective ly@xmath0 opacities using the same correction factors as @xcite .
we present the results in figure [ taufig ] along with the numerical values in table [ transtab ] .
we do not detect any transmission at @[email protected] , but we are able to set lower limits on @xmath41 of @xmath184 . collectively , the results from the two grb sightlines are generally consistent with those seen towards quasars .
lccc [email protected] & ly@xmath0 & 0.082 & 2.51 + [email protected] & ly@xmath0 & 0.094 & 2.37 + [email protected] & ly@xmath0 & 0.062 & 2.78 + [email protected] & ly@xmath0 & 0.029 & 3.53 + [email protected] & ly@xmath0 & @xmath110.014 & @xmath454.3 + [email protected] & ly@xmath0 & @xmath110.019 & @xmath454.0 + [email protected] & ly@xmath0 & @xmath110.016 & @xmath454.2 + [email protected] & ly@xmath0 & @xmath110.022 & @xmath453.8 + [email protected] & ly@xmath43 & @xmath110.103 & @xmath455.1 + [email protected] & ly@xmath43 & @xmath110.117 & @xmath454.8 [ transtab ] the large redshift bins used above effectively over - weight the transmission to regions of lower opacity and are a crude tracer of ly@xmath0 opacity when the optical depth is large . instead
, higher - order statistics , such as the distribution of length scales of dark gaps ( e.g. , @xcite ) or pixel - scale transmission statistics @xcite are more sensitive probes of the structure of reionization .
future work with additional grb sightlines , in combination with numerical simulations , will enable systematic tests of the statistics of ly@xmath0 opacity @xcite with a less - biased tracer of the end of reionization than quasars , which reside in more massive dark matter halos @xcite .
we now analyze the red damping wing of ly@xmath0 to set constraints on the neutral hydrogen column density responsible for the absorption .
we exclude the two regions shown in gray in the right panel of figure [ specfig ] , which include the foreground absorber and the residuals from an unfortunately placed strong night sky line at 8919 .
because we do not detect any narrow lines from the host galaxy , @xmath33 is a free parameter in all fits .
our first model assumes that all of the absorption is from a single absorber , presumably the host galaxy , at redshift @xmath33 .
this model is shown as the blue dashed line in figure [ specfig ] .
our best - fit parameters are @[email protected] at @[email protected] .
only 510% of grbs at @xmath19@xmath262 have measured host hydrogen column densities this low @xcite .
@xcite argued that the distribution of @xmath14 for a complete sample of grbs offers a statistical sampling of sightlines from regions of massive - star formation and can be used to infer the escape fraction of ionizing photons in high-@xmath19 galaxies . in figure
[ nhfig ] , we show the distribution of available @xmath14 measurements for @xmath19@xmath264 grbs ( mostly compiled by @xcite ) .
it is striking that the four highest - redshift grbs include the three lowest values for @xmath14 in the sample .
lower neutral hydrogen column densities imply a higher escape fraction .
an anderson - darling k - sample test gives a probability of 1.8% that the distribution of column densities at @xmath46 comes from the same distribution as that at lower redshift .
the redshift cut was not made blindly , so more data are needed to test whether the neutral hydrogen column distribution evolves to lower values at higher redshift .
one concern is selection biases , particularly that sightlines with higher columns will potentially be associated with more dust extinction and thus less likely to have afterglows identified and spectra taken with adequate s / n , but it is not clear how much of a differential effect this could be between the @xmath47 and @xmath17 samples . this result may be consistent with recent observations of a decreasing fraction of ly@xmath0-emitters among lyman - break galaxies at @xmath48 , which have also been modeled in terms of an increase in escape fraction ( e.g. , @xcite ) , although an increase in the neutral hydrogen fraction ( @xmath10 ) of the igm ( e.g. , @xcite ) is an alternative hypothesis .
these are complementary observations , as grbs are superior tracers of star formation at high redshift compared to flux - limited samples of galaxies @xcite .
our non - detection of narrow absorption lines from the host galaxy may be simply a consequence of the low neutral hydrogen column density .
some of the strongest lines we expect in the limited wavelength interval of our continuum are @xmath351260 , @xmath351302 , and @xmath351334 .
we use the noise in the spectrum to estimate conservative 3@xmath12 upper limits on the observed - frame ews of these lines of 1 , 2.5 , and 2.5 , respectively . assuming the optically - thin limit and the measured value of @xmath14 , these correspond to upper limits on the gas - phase abundances of [ z / h]@xmath11@xmath49 , @xmath50 , and @xmath51 , respectively @xcite .
if the igm contributes to the neutral hydrogen column , these limits would have to be relaxed .
nevertheless , they are compatible with typical sub - solar abundance estimates for high-@xmath19 grb host galaxies ( e.g. , @xcite ) and we do not require unusually low abundances to explain the lack of detected absorption lines from the host .
our second model ignores the likelihood of an absorber associated with the host galaxy and embeds a `` naked '' grb in a uniform medium of constant @xmath10 .
we use the model of @xcite to approximate the effects of a partially neutral igm .
the model requires a lower redshift bound for neutral hydrogen absorption , which we fix to @xmath19=6.0 because it has no effect on the results , and an upper redshift bound that we assume is equal to @xmath33 .
the best fit is shown as the red dot - dashed line in figure [ specfig ] and has @xmath33=6.3282@xmath52 and @xmath10=0.056@xmath53 . at the 3@xmath12 level ,
the maximum value allowed for @xmath10 is 0.09 .
the two models above represent possible extremes of absorption from only the host or igm . if both are present , the situation is more complex
. if we simply combine the models , the limits will relax and favor somewhat smaller values for @xmath10 and @xmath14 as the two sources of opacity are both allowed to contribute . however , more properly treating the inhomogeneity of the igm during of the end of reionization will considerably relax these constraints @xcite .
we address this issue by fitting a hybrid model with absorption from both the host and the igm .
this model has three free parameters ( @xmath14 , @xmath10 , and @xmath33 ) .
we place the grb in an ionized bubble and fix the radius of the bubble along our sightline to @xmath54=10 comoving mpc .
this choice is motivated by the simulations of @xcite , who found regions of approximately this scale when @xmath10 was globally equal to 0.5 .
the best - fit values are @xmath33=6.3273 , @xmath14=18.43 , and @xmath10=0.12 .
the marginalized contours from our fit are shown in figure [ igmfig ] . as expected ,
this significantly relaxes the constraint on @xmath10 . even with the favorable assumption about the existence of an ionized bubble , @[email protected] at the 2@xmath12 level .
we do not take into account the effect of inhomogeneity on the expected shape of the damping wing from the igm @xcite .
a larger bubble size would formally allow for even higher @xmath10 , but such bubbles are rare unless @xmath10 is low @xcite .
smaller ionized bubbles would only tighten this constraint .
the three fiducial models sample the range of variation of plausible reionization scenarios .
all three models provide acceptable fits to the data , with reduced-@xmath56 values near 1.0 , although the igm - only model appears to be less adequate at the sharp edge of ly@xmath0 .
the patchiness of the end of reionization will result in significant stochasticity along different sightlines @xcite and a statistical ensemble of grb sightlines will be necessary to measure cosmic variance and jointly constrain the models in more detail .
we have presented an optical spectrum of the afterglow of grb 140515aat @xmath1 and used it to provide constraints on the opacity and ionization state of the igm .
this is now the fourth high-@xmath19 grb for which a damping - wing analysis has been performed .
@xcite found that for grb 050904 at @xmath19=6.295 @xcite , @xmath57 at the 95% confidence level , but that the high host hydrogen column density inhibited tight constraints .
@xcite examined the same spectrum and argued that proper treatment of inhomogeneity will relax the constraint even further .
@xcite were unable to place meaningful constraints on the sightline to the @xmath19=6.733 grb 080913 due to the low s / n of the data , but quoted @xmath58 at the 90% confidence level .
the high-@xmath19 grb afterglow with the highest s / n observations was grb 130606a at @xmath19=5.913 . @xcite
set a 2@xmath12 upper limit of @[email protected] for a model with igm absorption extending up to @xmath33 .
@xcite argued that their spectra of the same burst disfavored a pure host - galaxy absorption model .
if they allowed igm absorption at redshifts up to @xmath33 , they found @xmath10=0.086@xmath59 . lowering the upper redshift to @xmath19=5.83 allowed values for @xmath10 up to @xmath180.5 .
however , their claims are based on a subtle 0.6% deviation from the pure host - galaxy model .
quasars are the other major observational probe of the ionization state of the igm at these redshifts , but variations in intrinsic properties and difficulties in modeling quasar near zones make the interpretation more controversial ( e.g. , @xcite ) .
the highest - redshift quasar currently known , ulas j1120@xmath600641 at @xmath19=7.085 , exhibits a red damping wing ascribed to @[email protected] @xcite , although the exact value depends on the unknown details of the quasar s lifetime @xcite . in our toy model with an @xmath61=10 mpc ( comoving ) bubble , we set a 2@xmath12 upper limit of @[email protected] , with a best - fit value of @xmath10=0.12 .
models with smaller or nonexistent regions will prefer even smaller values for @xmath10 .
the sharp decrement in flux at ly@xmath0 is incompatible with a large neutral fraction in the igm at @[email protected] in almost any model .
we compare this value to the others described above in the righthand panel of figure [ taufig ] . the grb constraints on
@xmath10 are still above predictions from models of the igm tuned to reproduce the ly@xmath0 forest at @xmath62 ( e.g. , @xcite ) .
the reasonable quality of the constraint on @xmath10 from these data despite the short integration time is a result of the brightness of the afterglow and the low hydrogen column density of the host . a future sample of grbs at even higher redshift ,
if it can be collected , would serve as an excellent record of the end of reionization .
we thank the gemini staff , particularly lucas fuhrman , for their superb assistance in obtaining these observations .
the berger grb group at harvard is supported by the national science foundation under grant ast-1107973 . based in part on observations obtained under program
i d gn-2014a - q-38 ( pi : berger ) 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 science and technology facilities council ( united kingdom ) , the national research council ( canada ) , conicyt ( chile ) , the australian research council ( australia ) , ministrio da cincia , tecnologia e inovao ( brazil ) and ministerio de ciencia , tecnologa e innovacin productiva ( argentina ) . | we present the discovery and subsequent spectroscopy with gemini - north of the optical afterglow of the _ swift _ gamma - ray burst ( grb ) 140515a .
the spectrum exhibits a well - detected continuum at wavelengths longer than 8915 with a steep decrement to zero flux blueward of 8910 due to ly@xmath0 absorption at redshift @xmath1 .
some transmission through the ly@xmath0 forest is present at @xmath2 , but none is detected at higher redshift , consistent with previous measurements from quasars and grb 130606a .
we model the red damping wing of ly@xmath0 in three ways that provide equally good fits to the data : ( a ) a single host galaxy absorber at @xmath3 with @[email protected] ; ( b ) pure intergalactic medium ( igm ) absorption from @xmath6 to @xmath7 with a constant neutral hydrogen fraction of @xmath8 ; and ( c ) a hybrid model with a host absorber located within an ionized bubble of radius 10 comoving mpc in an igm with @xmath9 ( @[email protected] at the 2@xmath12 level ) .
regardless of the model , the sharpness of the dropoff in transmission is inconsistent with a substantial neutral fraction in the igm at this redshift .
no narrow absorption lines from the host galaxy are detected , indicating a host metallicity of [ z / h ] @xmath13 . even if we assume that all of the hydrogen absorption is due to the host galaxy , the column is unusually low for a grb sightline , similar to two out of the other three highest - redshift bursts with measured @xmath14 .
this is possible evidence that the escape fraction of ionizing photons from normal star - forming galaxies increases at @xmath15 . |
multi - resolution representations of discrete data are useful tools in several areas of application as image compression or adaptive methods for partial differential equations . in these applications ,
the ability of these representations to approximate the input data with high accuracy using a very small set of coefficients is a central property .
moreover , the stability of these representations in presence of perturbations ( generated by compression or due to approximations ) is a key point . + in the last decade , several attempts to improve the property of classical linear multi - resolutions have lead to nonlinear multi - resolutions . in many cases ,
this nonlinear nature hinders the proofs of convergence and stability .
+ in @xcite , in the context of image compression , a new multi - resolution transform has been presented .
this multi - resolution is based on an univariate nonlinear multi - resolution called pph multi - resolution ( see @xcite in the context of convexity preserving ) .
it has been analyzed in terms of convergence and stability of an associated subdivision scheme following an approach for data dependent multi - resolutions introduced in @xcite .
due to nonlinearity , the stability of the pph multi - resolution is not a consequence of the convergence of the associated subdivision scheme .
it has been established in @xcite , presenting the pph subdivision scheme as some perturbation of a linear scheme following @xcite , @xcite , @xcite and @xcite .
+ the aim of the present paper is to generalize the results presented in @xcite for a general family of nonlinear multi - resolution schemes associated to an interpolatory subdivision scheme @xmath0 of the form : @xmath1 where @xmath2 is a nonlinear operator defined on @xmath3 , @xmath4 is a linear and continuous operator on @xmath5 and @xmath6 is a linear and convergent subdivision scheme . considering two subdivision schemes @xmath7 and @xmath6
, it is always possible to introduce the difference @xmath8 . if one assume some properties of polynomial reproduction ( see section [ sec3 ] ) , as shown in @xcite , @xmath2 is in fact a function of differences , i.e. of @xmath9 defined by @xmath10 .
theorems [ th1 ] and [ th5 ] , that are the main results of this paper , establish that if @xmath11 and @xmath4 satisfy some natural properties , then the subdivision scheme is convergent and the multi - resolution is stable . + the paper is organized as follows . in section [ sec2 ]
we recall briefly the harten s interpolatory multi - resolution framework which is the natural setting for our work .
we precise the class of schemes under consideration and we establish the main results in section [ sec3 ] .
various applications are presented in section [ sec4 ] .
in harten s interpolatory multi - resolution , one considers a set of nested bi - infinite regular grids : @xmath12 where @xmath13 stands for a scale parameter and @xmath14 controls the position .
+ the point - value discretization operators ( or sampling operators ) are defined by @xmath15 where @xmath16 is the space of real sequences and @xmath17 the set of continuous functions on @xmath18 .
+ the reconstruction operators @xmath19 associated to this discretization are any right inverses of @xmath20 on @xmath21 , that is , any operators @xmath22 satisfying : @xmath23 for any @xmath13 , the operator defined by @xmath24 acts between a fine scale @xmath25 and a coarser scale j. here , it is a sub - sampling operator from @xmath26 to @xmath27 . + the operator defined by @xmath28 acts between a coarse scale @xmath13 and a finer scale @xmath25 and is called a prediction operator . a prediction operator can be considered as a subdivision scheme @xcite from @xmath27 to @xmath26 .
we say that the subdivision scheme @xmath6 defined by @xmath29 is uniformly convergent if : @xmath30 we note @xmath31 . since for most function @xmath32 , @xmath33 , details , called @xmath34 and defined by @xmath35 ,
should be added to @xmath36 to recover @xmath37 from @xmath38 . the multi - resolution decomposition ( see @xcite , @xcite , @xcite for precisions ) of @xmath39 is the sequence @xmath40 . moreover , the multi - resolution transform is said to be stable if : @xmath41 where @xmath42 is the multi - resolution decomposition of @xmath43 . when the prediction operator @xmath36 is linear , the convergence of the associated subdivision scheme implies the stability of the multi - resolution analysis . in the non linear case ,
it is not the case and there is no general result for the multi - resolution analysis stability .
introducing @xmath6 a linear , reproducing polynomials reproduces polynomials of degree @xmath44 if , for any polynomials @xmath45 of degree less or equal to @xmath44 , if @xmath46 then @xmath47 up to degree @xmath44 and convergent interpolatory subdivision scheme we consider nonlinear interpolatory subdivision schemes that write @xmath48 where @xmath2 is a nonlinear operator defined on @xmath49 and @xmath4 is a continuous linear operator on @xmath49 .
+ we have the following theorem related to the convergence of the nonlinear subdivision scheme @xmath7 : : [ th1 ] if @xmath50 and @xmath4 verify : @xmath51 then the subdivision scheme @xmath7 is uniformly convergent .
moreover , if @xmath6 is @xmath52 convergent ( i.e for all @xmath53 , @xmath54 . for @xmath55 with @xmath56 , @xmath57 and @xmath58 , @xmath59 ) then , for all sequence @xmath60 with @xmath61 .
+ * proof * using hypotheses ( [ h1 ] ) and ( [ h2 ] ) and the definition of @xmath7 , we get : @xmath62 that can be rewritten as : @xmath63 writing : @xmath64 the convergence of the subdivision scheme @xmath7 can be obtained applying theorem 3.3 of @xcite .
+ in our context , this theorem applies as follows : + if @xmath6 is a linear @xmath52 convergent subdivision scheme reproducing polynomials up to degree @xmath44 and if @xmath7 is a perturbation of @xmath6 in the sense that , calling @xmath65 for all @xmath66 , @xmath67 then @xmath7 is @xmath68convergent with @xmath69 .
it follows that if @xmath6 is @xmath52 convergent then @xmath7 is at least @xmath68 convergent with @xmath61 .
+ @xmath70 + when @xmath2 is linear , theorem [ th1 ] is a consequence of theorem 6.2 in @xcite .
[ regularite ] in many of our examples , @xmath6 is the two point centered linear scheme defined by @xmath71 which limit function is in @xmath72 .
therefore , as soon as the non linear scheme @xmath7 verifies hypothesis ( [ h1 ] ) and ( [ h2 ] ) with @xmath73 , the @xmath7 is @xmath74 convergent .
hypothesis [ h2 ] can be weakened as : @xmath75 the proof remains the same except that : @xmath76 becomes : @xmath77 that can be rewritten , for @xmath78 $ ] , as : @xmath79 the conclusion is reached applying theorem 3.3 of @xcite .
[ deuxoperateurs ] a straightforward generalization of theorem [ th1 ] can be obtained introducing two linear operator @xmath80 , @xmath81 and a perturbation of the form @xmath82 . under the following hypotheses : @xmath83 for all @xmath84 , the scheme @xmath7 is uniformly convergent .
[ r2 ] we can also apply theorem [ th1 ] to bi - variate schemes written as @xmath85 if the following conditions are satisfied for @xmath86 @xmath87 for all @xmath88 , the scheme @xmath7 is uniformly convergent . + we now consider the multi - resolution analysis associated to the subdivision scheme ( [ 75 ] ) recalling that , for any sequence @xmath38 , the details @xmath89 are defined by @xmath90 .
+ we have the following theorem concerning the stability of the multi - resolution : + [ th5 ] if @xmath91 and @xmath4 verify : @xmath92 such that @xmath93 , @xmath94 then the multi - resolution transform associated to the non linear subdivision scheme @xmath7 is stable .
+ * proof * + + we first prove ( [ s1 ] ) : + due to the interpolatory property , we only consider @xmath95 .
since @xmath6 is a convergent linear scheme , we have , using the stability of the linear scheme @xmath6 : @xmath96 such that @xmath97 from ( [ eq30 ] ) : @xmath98 concentrating on the last right hand side term we get : @xmath99 from ( [ eq31 ] ) we get : @xmath100 since @xmath101 we get finally : @xmath102 and , using the continuity of @xmath103 , we get ( [ s1 ] ) with a constant @xmath104 + we now establish ( [ s2 ] ) et ( [ s3 ] ) . + equation ( [ s2 ] ) is a direct consequence of the interpolatory properties .
+ for ( [ s3 ] ) , we have , for @xmath105 : @xmath106 using the property ( [ s3 ] ) for the multi - resolution associated to @xmath6 , hypothesis ( [ eq30 ] ) and the continuity of @xmath4 , we have : @xmath107 from ( [ s2 ] ) for the multi - resolution associated to @xmath7 we have @xmath108 and therefore we get ( [ s3 ] ) with @xmath109 .
+ @xmath70 + + as previously , we can again consider a weaker formulation for hypothesis ( [ eq31 ] ) such as : @xmath110 under this hypothesis , the stability of the subdivision scheme can still be established .
however , the multi - resolution stability is not ensured .
to get it , a stronger hypothesis like : @xmath111 is required .
this sections is devoted to applications of the previous results to three specific subdivision schemes ( linear and nonlinear ) available in the literature .
we provide for each of them , the proofs of convergence and stability .
in all this section , given @xmath112 we note : @xmath113 and , more generally @xmath114 with : @xmath115 as it has been said before , for linear scheme , the stability of the multi - resolution analysis is a consequence of the convergence of the subdivision scheme ( see @xcite ) .
therefore , we only consider here the convergence of the subdivision scheme .
+ the convergence of centered linear interpolatory schemes is well know since delauriers and debuc @xcite .
+ for linear but non centered schemes there is no general results of convergence .
moreover the general tools proposed in @xcite are very fastidious to apply and do nt provide general results . in this subsection , we focus on completely decentred lagrange interpolatory linear schemes . in order to apply our theoretical results , we consider @xmath6 the two point centered linear scheme and express any right hand side excentred scheme @xmath116 ( where @xmath44 stands for the number of point of the considered stencil ) as a perturbation of it .
precisely , if we write @xmath117 we get : if @xmath44 is even , @xmath118 and , if @xmath44 is odd + @xmath119 numerical evaluation of the perturbation terms for @xmath120 is given in table [ tabsub ] .
c|l p & @xmath121 + @xmath122 & ' '' '' @xmath123 + @xmath124 & ' '' '' @xmath125 + @xmath126 & ' '' '' @xmath127 + @xmath128 & ' '' '' @xmath129 + @xmath130 & ' '' '' @xmath131 + @xmath132 & ' '' '' @xmath131 + & ' '' '' @xmath133 + + it also appears that @xmath134 can be written naturally as a perturbation of @xmath135 , for even values of @xmath44 and as a perturbation of @xmath136 for odd values of @xmath44 .
indeed , we have : when @xmath44 is even : @xmath137 and when @xmath44 is odd : @xmath138 in both cases , its is easy to check that the function @xmath2 defined by @xmath139 when @xmath44 is even and by @xmath140 when @xmath44 is odd , is linear and continuous
. therefore , the convergence can be reached as soon as the contractivity hypothesis ( [ h2 ] ) for @xmath141 or @xmath142 is satisfied .
+ + direct calculations provide the estimates gathered in table [ table2 ] .
c|c|c p & perturbation term & contractivity estimate + 4 & ' '' '' @xmath143&@xmath144&@xmath145 + 6 & ' '' '' @xmath146&@xmath147 + 7 & ' '' '' @xmath148&@xmath149 + 8 & ' '' '' @xmath150&@xmath151 + 9 & ' '' '' @xmath152&@xmath153 + it then follows from theorem [ th1 ] that all the fully excentred interpolatory subdivision schemes for @xmath154 points converge .
+ the following comments can be made for the other situations : figure [ decentre ] represents the completely decentred 9 and 10 points scheme iterated functions at scale @xmath130 starting from @xmath155 . from the zooming in the oscillating region
, one can guess that the @xmath132 point scheme converges while the @xmath156 point does nt .
in fact , following @xcite , and using the so called iterative formalism one observes numerically that the spectral radius of the iterated matrix for @xmath156 points overshoots the critical value @xmath157 while the spectral radius of the iterated matrix for @xmath132 points does nt , that confirms the guess
. + obviously , for these linear schemes , theorem [ th5 ] applies as soon as theorem [ th1 ] does .
therefore stability is ensured for @xmath158 .
+ + @xmath159 subdivision schemes @xcite , are constructed using convex combination of different interpolatory polynomials of fixed degree . for degree @xmath160 and therefore a 6 point stencil ,
the @xmath161 subdivision is given by : @xmath162 where the coefficients @xmath163 control the convex combination and therefore satisfy @xmath164 and @xmath165 .
+ + in @xcite , these coefficients are defined as : @xmath166 with @xmath167 where @xmath168 , defined as a function of the first difference @xmath9 is an indicator of smoothness while @xmath169 and @xmath170 are fixed positive constants .
a set of possible values for these constants is suggested in @xcite .
+ + the convergence of the associated subdivision scheme has been studied in @xcite .
we present an alternative proof , using theorem [ th1 ] .
+ first , the @xmath161 subdivision scheme is written as a perturbation of the linear two point interpolation scheme as : @xmath171 with @xmath172 , @xmath173 and @xmath174 .
+ + + we then have the following proposition : + the @xmath161 subdivision scheme is convergent and , for any initial sequence @xmath38 , the limit function belongs to @xmath175 .
+ * proof * + according to remark [ deuxoperateurs ] , the proof can be performed in three steps considering that @xmath2 is a function of @xmath176 and @xmath177 .
first , according to the definition of @xmath2 and to the properties of @xmath178 , we have @xmath179 second , we prove ( [ deux1 ] ) for the first difference operator @xmath180 : we have , for @xmath181 : @xmath182 + we have to consider two cases , according to the parity of @xmath183 .
we give the details for @xmath184 , the even case being similar .
@xmath185 since @xmath186 and @xmath187 , we have : @xmath188 third , we prove inequality ( [ h2 ] ) for the second difference operator @xmath189 .
+ again , two cases have to be considered : + @xmath190 for k=2n+1 , then : + @xmath191 using @xmath192 , @xmath187 , @xmath193 and @xmath194 , we get : @xmath195 + + @xmath190 for k=2n , then : + @xmath196 using @xmath194 , we get : + @xmath197 then , with @xmath192 , @xmath187 and @xmath193 : @xmath198 + therefore , from ( [ eq23 ] ) , ( [ eq24 ] ) and ( [ eq25 ] ) , we obtain the inequality : @xmath199 finally , using remarks [ deuxoperateurs ] and [ regularite ] we get the convergence of the @xmath161 subdivision scheme to a @xmath175 function . @xmath70 + + + in the same vein as the pph scheme ( @xcite ) , the power p scheme is a four point scheme based on a piecewise degree @xmath160 polynomial prediction .
considering @xmath200 the centered four point lagrange interpolation prediction that reads : @xmath201 the definition of the power - p subdivision scheme is based on the substitution of the arithmetic mean of second order differences , @xmath202 , by a general mean @xmath203 defined in @xcite for any integer @xmath204 , and any couple @xmath205 as : @xmath206 note that it coincides for @xmath207 , with the arithmetic mean and for @xmath208 with the geometric mean.the power - p subdivision scheme then naturally appears as a perturbation of the linear two point interpolation scheme since it is defined by @xmath209)_{2n+1}&=&\frac{f_n+f_{n+1}}{2}-\displaystyle{\frac{1}{8}}power_p(df^j_n , df^j_{n+1}).\end{aligned}\ ] ] before establishing the convergence theorem we first prove the following lemma : [ lem1 ] for any @xmath210 , the function @xmath211 satisfies the following properties : 1 .
@xmath212 2 .
@xmath213 3 .
@xmath214 4 .
@xmath215 5 .
@xmath216 + * proof * + + claims of @xmath217 are obvious ; + inequality @xmath124 comes from the equality @xmath218.\end{aligned}\ ] ] @xmath70 + we then have the following proposition : the power p subdivision scheme is uniformly convergent and , for any initial sequence @xmath38 the limit function belongs to @xmath219 for @xmath220 and @xmath175 for @xmath221 .
+ * proof * + here again , the hypotheses of the general theorem [ th1 ] must be checked : we first check hypothesis ( [ h1 ] ) . using property @xmath122 of lemma [ lem1 ] , we obtain for @xmath222 : @xmath223 + then we consider hypothesis ( [ h2 ] ) : we study as before two different cases : + + @xmath190 for @xmath184 : @xmath224 + from property @xmath122 of lemma [ lem1 ] we get : latexmath:[\ ] ] the contractivity hypothesis ( [ h2 ] ) is therefore satisfied as soon as @xmath268 , 1-\frac{\sqrt{2}}{\pi }
< h'(x)<1+\frac{\sqrt{2}}{\pi}$ ] .
for instance , the following function h : @xmath269 which is in agreement with the criteria proposed in @xcite leads to a scheme satisfying ( [ h2 ] ) . since the same sketch of proof also provides the contractivity for @xmath270 we get the convergence applying theorem [ th1 ] .
we have formulated convergence and stability conditions for non linear subdivision schemes and associated multi - resolutions .
these conditions deal with the difference with a suitable linear and convergent subdivision scheme .
many examples show that this formulation lead to simple proofs of convergence and stability . | this paper is devoted to the convergence and stability analysis of a class of nonlinear subdivision schemes and associated multi - resolution transforms .
these schemes are defined as a perturbation of a linear subdivision scheme . assuming a contractivity property , stability and convergence are derived .
these results are then applied to various schemes such as uncentered interpolatory linear scheme , weno scheme @xcite , power - p scheme @xcite and a non linear scheme using local spherical coordinates @xcite .
jacques liandrat non linear subdivision schemes , non linear multi - resolutions , convergence , stability * key words . * non linear subdivision schemes , convergence , multi - resolution , interpolation , stability , convergence 41a05 , 41a10 , 65d05 , 65d17 |
* samples * details of the pld growth of the yig layer can be found in ref.@xcite .
its dynamical properties have been determined by broadband fmr measurements .
the transport parameters of the 8 nm thick pt layer deposited on top by magnetron sputtering have been determined in a previous study @xcite .
the yig@xmath4pt microdiscs are defined by e - beam lithography , as well as the au(80 nm)@xmath4ti(20 nm ) electrodes separated by 1 @xmath0 m from each other which contact them .
this electrical circuit is insulated by a 300 nm thick sio@xmath75 layer , and a broadband microwave antenna made of 250 nm thick au with a 5 @xmath0 m wide constriction is defined on top of each disc by optical lithography .
* measurements * the samples are mounted between the poles of an electromagnet which can be rotated to vary the angle @xmath36 shown in fig .
two 50 @xmath76 matched picoprobes are used to connect to the microwave antenna and to the electrodes which contact the pt layer .
the latter are connected to a dc current source through a bias - tee . to perform ishe - detected fmr measurements ,
a microwave synthesizer is connected to the microwave antenna , and the output power is turned on and off at a modulation frequency of 9 khz . the voltage across pt
is measured by a lock - in after a low - noise preamplifier ( gain 100 ) . for the detection of auto - oscillations ,
high - frequency low - noise amplifiers are used ( gain 33 db to 39 db , depending on the frequency range ) .
two spectrum analysers simultaneously monitor in the frequency domain the voltages @xmath37 and @xmath8 across the pt layer and in the microwave antenna , respectively ( fig .
the resolution bandwidth employed in the measurements is set to 1 mhz .
we acknowledge e. jacquet , r. lebourgeois and a. h. molpeceres for their contribution to sample growth , and m. viret and a. fert for fruitful discussion .
this research was partially supported by the anr grant trinidad ( astrid 2012 program ) .
v. v. n. acknowledges support from the program cmirapro of the region rhne - alpes .
s. o. d. and v. v. n. acknowledge respectively support from the russian programs megagrant no .
2013 - 220 - 04 - 329 and competitive growth of kfu .
32ifxundefined [ 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 * * , ( ) link:\doibase 10.1038/nature10309 [ * * , ( ) ] link:\doibase 10.1038/ncomms3944 [ * * , ( ) ] link:\doibase 10.1038/nature13534 [ * * , ( ) ] link:\doibase 10.1038/nature08876 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevlett.113.197203 [ * * , ( ) ] \doibase doi:10.1016/0304 - 8853(96)00062 - 5 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevlett.83.1834 [ * * , ( ) ] link:\doibase 10.1038/nmat3459 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevlett.107.066604 [ * * , ( ) ] link:\doibase 10.1103/physrevb.87.174417 [ * * , ( ) ] link:\doibase 10.1103/physrevb.89.140406 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.112.197201 [ * * , ( ) ] link:\doibase 10.1063/1.4759039 [ * * , ( ) ] link:\doibase 10.1063/1.4819157 [ * * , ( ) ] link:\doibase 10.1109/lmag.2014.2350958 [ * * , ( ) ] link:\doibase 10.1103/physrevb.60.7395 [ * * , ( ) ] link:\doibase 10.1109/tmag.2003.821564 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.111.217204 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1063/1.4871516 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.112.106602 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.4813315 [ * * , ( ) ] link:\doibase 10.1103/physrevb.87.144411 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.100.017207 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.107.107204 [ * * , ( ) ] link:\doibase 10.1038/ncomms6616 [ * * , ( ) ] | * spin - orbit effects @xcite have the potential of radically changing the field of spintronics by allowing transfer of spin angular momentum to a whole new class of materials . in a seminal letter to nature @xcite , kajiwara _ et al . _ showed that by depositing platinum ( pt , a normal metal ) on top of a 1.3 @xmath0 m thick yttrium iron garnet ( yig , a magnetic insulator ) , one could effectively transfer spin angular momentum through the interface between these two different materials .
the outstanding feature was the detection of auto - oscillation of the yig when enough dc current was passed in the pt .
this finding has created a great excitement in the community for two reasons : first , one could control electronically the damping of insulators , which can offer improved properties compared to metals , and here yig has the lowest damping known in nature ; second , the damping compensation could be achieved on very large objects , a particularly relevant point for the field of magnonics @xcite whose aim is to use spin - waves as carriers of information .
however , the degree of coherence of the observed auto - oscillations has not been addressed in ref .
@xcite . in this work
, we emphasize the key role of quasi - degenerate spin - wave modes , which increase the threshold current .
this requires to reduce both the thickness and lateral size in order to reach full damping compensation @xcite , and we show clear evidence of coherent spin - orbit torque induced auto - oscillation in micron - sized yig discs of thickness 20 nm . * when spin transfer effects were first introduced by slonczweski and berger in 1996 @xcite , the authors immediately recognized that the striking signature of the transfer process would be the emission of microwave radiation when the system is pumped out of equilibrium by a dc current . since the spin transfer torque on the magnetisation is collinear to the damping torque , there is an instability threshold when the natural damping is fully compensated by the external flow of angular momentum , leading to spin - wave amplification through stimulated emission . using analogy to light ,
the effect was called swaser @xcite , where sw stands for spin - wave . until 2010 ,
all swaser devices required a charge current perpendicular to the plane to transfer angular momentum between different magnetic layers @xcite .
this implied that the effect was restricted to conducting materials .
the situation has radically changed since spin - orbit effects such as the spin hall effect ( she ) @xcite are used to produce spin currents in normal metals . here
a right hand side rule links the deflected direction of the electron and the orientation of its spin .
this allows the creation of a pure spin current transversely to the charge current , with an efficiency given by the spin hall angle @xmath1 . using a metal with large @xmath1 , such as pt ,
a charge current flowing in plane generates a pure spin current flowing perpendicular to the plane , which can eventually be transferred through an interface with ferromagnetic metals , resulting in the coherent emission of spin - waves @xcite , but also with non - metals such as yig @xcite .
the microscopic mechanisms of transfer of angular momentum between a normal metal and a ferromagnetic layer are quite different depending on the latter being metallic or not . in the first case ,
electrons in each layer have the possibility to penetrate the other one , whereas in the second case the transfer takes place exactly and solely at the interface .
it is thus much more sensitive to the imperfection of the interface .
still , a direct experimental evidence that spin current can indeed cross such an hybrid interface is through the so - called spin pumping effect @xcite : adding a normal metal on top of yig increases its ferromagnetic resonance ( fmr ) linewidth @xcite , which is due to the new relaxation channel at the interface through which angular momentum can escape and get absorbed in the metal .
this effect being interfacial , the broadening scales as @xmath2 , where @xmath3 is the thickness of yig . even for yig ,
whose natural linewidth is only a few oersted at 10 ghz , it is hardly observable if @xmath3 exceeds a couple hundreds of nanometers . for these thick films
though , the spin pumping can still be detected through inverse spin hall effect ( ishe ) . in a normal metal with strong spin - orbit interaction ,
the pumped spin current is converted into a transverse charge current .
this generates a voltage proportional to the length of the sample across the metal , which can easily reach several tens of microvolts in millimeter - sized samples . since the first experiment by kajiwara _
et al . _
@xcite , many studies reported the ishe detection of fmr using different metals on yig layers @xcite , hereby providing clear evidence of at least partial transparency of the hybrid yig@xmath4metal interfaces to spin currents .
due to onsager relations , these results made the community confident that a spin current could thus be injected from metals to yig and lead to the swaser effect . from the beginning
it was anticipated that the key to observe auto - oscillations in non - metals was to reduce the threshold current .
the first venue is of course to choose a material whose natural damping is very low . in this respect yig is the optimal choice .
the second thing is to reduce the thickness since the spin - orbit torque ( sot ) is an interfacial effect .
this triggered an effort in the fabrication of ultra - thin films of yig of very high dynamical quality @xcite . for 20
nm thick yig films with damping constant as low as @xmath5 , a striking result was that there were no evidence of auto - oscillations in millimeter - sized samples at the highest dc current possible in the top pt layer @xcite , _
i.e. _ , before it evaporates .
it is worth mentioning at this point that reducing further the thickness or the damping parameter of such ultra - thin yig films @xcite does not help anymore in decreasing the threshold current , as the relevant value of the damping is that of the yig@xmath4pt hybrid , which ends up to be completely dominated by the spin - pumping contribution .
but most notably , none of these high - quality ultra - thin yig films display a purely homogeneous fmr line .
the reason for that is well known . in such extended films ,
there are many degenerate modes with the main , uniform fmr mode , which through the process of two - magnon scattering broaden the linewidth @xcite . a striking evidence of these degenerate modes can be obtained by parametrically pumping the sw modes .
it reveals an uncountable number of modes which are at the same energy as the fmr mode @xcite .
any threshold instability will be affected by the presence of those modes , as learnt from lasers where mode competition is known to have a strong influence on the emission threshold @xcite .
thus , the next natural step was to reduce as well the lateral size in order to lift the degeneracy between modes through confinement .
the first microstructures of yig appeared revealing that the patterning indeed narrowed the linewidth through a decrease of the inhomogeneous part @xcite .
the effect is clear in the perpendicular geometry , where magnon - magnon processes are suppressed owing to the fact that the fmr mode lies at the bottom of the sw dispersion relation .
however , this is not the case in the parallel geometry where the fmr mode is not the lowest energy sw mode .
even then , we showed that the linewidth in a micron - sized yig@xmath4pt disc could be still tuned thanks to sot @xcite . in the following ,
we describe the direct electrical detection of auto - oscillations in similar samples and show that the threshold current is increased by the presence of quasi - degenerate sw modes .
pt microdisc . * * a * , sketch of the sample and measurement configuration .
the bias field @xmath6 is oriented transversely to the dc current @xmath7 flowing in pt .
the inductive voltage @xmath8 produced in the antenna by the precession of the yig magnetisation is amplified and monitored by a spectrum analyser . *
b e * , power spectral density maps measured at fixed @xmath9 koe and variable @xmath7 .
the four quadrants correspond to different possible polarities of @xmath6 and @xmath7 .
, width=283 ] we study magnetic microdiscs with diameter 2 @xmath0 m and 4 @xmath0 m which are fabricated based on a hybrid yig(20 nm)@xmath4pt(8 nm ) bilayer .
the 20 nm thick yig layer is grown by pulsed laser deposition @xcite and the 8 nm thick pt layer is sputtered on top of it @xcite .
their physical parameters are summarized in table i. we stress that the extended yig film is characterized by a low gilbert damping parameter @xmath10 and a remarkably small inhomogeneous contribution to the linewidth , @xmath11 oe .
each microdisc is connected to electrodes enabling the injection of a dc current @xmath7 in the pt layer , and a microwave antenna is defined around it to obtain an inductive coupling with the yig magnetisation , as shown schematically in fig .
1a . [ cols="^,^,^,^,^",options="header " , ] [ tab : mat ] first , we monitor with a spectrum analyser the voltage produced in the antenna by potential auto - oscillations of the 4 @xmath0 m yig disc as a function of the dc current @xmath7 injected in pt . the in - plane magnetic field @xmath12 koe is applied in a transverse direction with respect to @xmath7 , as shown in fig .
1a .
this is the most favorable configuration to compensate the damping and obtain auto - oscillations in yig , as spins accumulated at the yig@xmath4pt interface due to she in pt will be collinear to its magnetisation .
color plots of the inductive signal measured as a function of the relative polarities of @xmath6 and @xmath7 are presented in fig .
1b e . at @xmath13
, we observe in the power spectral density ( psd ) a peak which starts at around 2.95 ghz and @xmath14 ma and then shifts towards lower frequency as @xmath7 is increased ( fig .
1b ) , a clear signature that spin transfer occurs through the yig@xmath4pt interface .
the linewidth of the emission peak , lying in the 1020 mhz range for @xmath15 ma , also proves the coherent nature of the detected signal .
an identical behaviour is observed at @xmath16 and @xmath17 ( fig .
1d ) .
in contrast , the psd remains flat in the two other cases ( fig .
1c
d ) .
therefore , an auto - oscillation signal is detected only if @xmath18 , in agreement with the expected symmetry of she . in order to characterize the flow of angular momentum across the yig@xmath4pt interface
, we now perform ishe - detected fmr spectroscopy on our microdiscs .
the configuration of this experiment is similar to the previous case , but now the antenna generates a uniform microwave field @xmath19 to excite the fmr of yig while the dc voltage across pt is monitored at zero current ( see fig .
2a ) .
in other words , we perform the reciprocal experiment of the one presented in fig .
1 . as described in the introduction , a voltage @xmath20 develops across pt when the fmr conditions are met in yig .
this voltage changes sign as the field is reversed , which is expected from the symmetry of ishe , and shown in fig .
2b , where the fmr spectra of the 4 @xmath0 m and 2 @xmath0 m microdiscs are respectively detected at 1 ghz and 4 ghz .
we also note that for a given field polarity , the product between @xmath20 and @xmath7 must be negative to compensate the damping @xcite , which enables to observe auto - oscillations in fig
. 1 . from these ishe measurements , the dispersion relation and frequency dependence of the full linewidth at half maximum of the main fmr mode can be determined , as shown in figs . 2c and 2d , respectively .
the dispersion relation follows the expected kittel law .
the damping parameters of the 4 @xmath0 m and 2 @xmath0 m microdiscs , extracted from linear fits to the data , @xmath21 ( continuous lines in fig .
2d , @xmath22 is the pulsation frequency and @xmath23 the gyromagnetic ratio ) , are found to be similar with an average value of @xmath24 .
the small inhomogeneous contribution to the linewidth observed in both microdiscs , @xmath25 oe and @xmath26 oe , respectively , decreases with the diameter and is attributed to the presence of several unresolved modes within the resonance line @xcite . in order to emphasize the increase of damping due to pt
, we have reported in fig .
2d the broadening produced by the homogeneous contribution of the bare yig using a dashed line . the observed increase of damping is due to spin pumping @xcite , @xmath27 where @xmath28 is the reduced planck constant , @xmath29 the saturation magnetisation and @xmath30 the spin - mixing conductance of the yig@xmath4pt interface .
this allows us to extract @xmath31
m@xmath32 , which lies in the same window as previously reported values @xcite . from the spin - mixing conductance @xmath30 , the spin diffusion length @xmath33 and the resistivity @xmath34 of the pt layer
, we can also calculate the transparency of the yig@xmath4pt interface to spin current @xcite , @xmath35 .
the physical parameters extracted for the yig@xmath4pt hybrid bilayer are summarized in the last raw of table i. is oriented at an angle @xmath36 with the dc current @xmath7 in the pt .
the precession of the yig magnetisation induces voltages @xmath37 in the antenna and @xmath8 across pt , which are amplified and monitored by spectrum analysers . *
b d * , @xmath37 and @xmath8 at @xmath12 koe for three different angles @xmath36 . *
e * , dependence of the normalized signals in both circuits and * f * , of the threshold current for auto - oscillations on @xmath36 .
dashed lines show the expected angular dependences.,width=321 ] to gain further insight about the origin of the auto - oscillation signal , we now monitor how the auto - oscillations of the 4 @xmath0 m disc evolve as the angle @xmath36 between the in - plane bias field fixed to @xmath12 koe and the dc current @xmath7 is varied from @xmath38 to @xmath39 .
the results are summarized in fig .
3 .
pannels b d show the auto - oscillation voltages detected in the antenna ( @xmath8 ) and across the pt electrode ( @xmath37 ) . at @xmath40 ,
the auto - oscillation signal is only visible in the @xmath8 channel . at @xmath41 , both @xmath37 and @xmath8 channels
exhibit the auto - oscillation peak . at @xmath42
, it almost vanishes in @xmath8 , while it slightly increases in @xmath37 .
the normalized signals as a function of @xmath36 are plotted in fig .
3e .
the pt electrode and antenna loop being oriented perpendicularly to each other ( see fig .
3a ) , the ac flux due to the precession of magnetisation picked up by each of them respectively varies as @xmath43 and @xmath44 ( dashed lines in fig .
3e ) .
more importantly , this study of angle dependence also allows us to extract the threshold current for auto - oscillations as a function of @xmath36 .
as @xmath36 deviates from the optimal orientation @xmath45 , the absolute value of the threshold current rapidly increases , see fig .
3f . in fact , the sot acting on the oscillating part @xmath46 of the magnetisation scales as @xmath47 , where @xmath48 is the spin polarisation of the dc spin current produced by she in pt at the yig@xmath4pt interface . therefore ,
the expected threshold current scales as @xmath49 , which is plotted as a dashed line in fig .
3f , in very good agreement with the data . in summary ,
the results reported in figs . 1 and 3
unambiguously demonstrate that the auto - oscillations observed in our hybrid yig@xmath4pt discs result from the action of sot produced by @xmath7 .
we have also shown that they correspond to the reverse effect of the spin pumping mechanism illustrated in fig .
2d and its detection through ishe in fig . 2b . in the last part of this letter , we analyse quantitatively the main features of auto - oscillations , which allows us to determine their nature and to understand the role of quasi - degenerate sw modes in the sot driven dynamics . for this ,
we compare the auto - oscillations observed in the 4 @xmath0 m and 2 @xmath0 m microdiscs . figs . 4a and
4b respectively present the inductive signal @xmath8 detected in the antenna coupled to these two discs as a function of @xmath7 .
the configuration is the same as depicted in fig .
1a , with a slightly larger bias field set to @xmath50 koe .
one can clearly see a peak appearing in the psd close to 3.6 ghz in both cases , at a threshold current of about @xmath51 ma in the 4 @xmath0 m disc and @xmath52 ma in the 2 @xmath0 m disc .
these two values correspond to a similar threshold current density in both samples of @xmath53 , in agreement with our previous study @xcite . as the dc current
is varied towards more negative values , the peaks shift towards lower frequency ( fig .
4c ) , at a rate which is twice faster in the smaller disc .
this frequency shift is mainly due to linear and quadratic contributions in @xmath7 of oersted field and joule heating , respectively @xcite ( from the pt resistance , the maximal temperature increase in both samples is estimated to be @xmath54c ) . at the same time , the signal first rapidly increases in amplitude , reaches a maximum , and then , more surprisingly , drops until it can not be detected anymore , as seen in fig .
4e , which plots the integrated power _ vs. _
@xmath7 .
the maximum of power measured in the 4 @xmath0 m disc ( 2.9 fw ) is four times larger than the one measured in the 2 @xmath0 m disc ( 0.7 fw ) , which is due to the inductive origin of @xmath8 .
the latter can be estimated from geometrical considerations , @xmath55 . here
@xmath56 is the magnetic constant , @xmath57 the diameter of the disc and @xmath58 the angle of uniform precession ( the prefactor @xmath59 accounts for microwave losses and impedance mismatch in the measured frequency range with our microwave circuit ) .
for the same @xmath58 , the inductive voltage produced by the 4 @xmath0 m disc is thus twice larger than by one produced by the 2 @xmath0 m disc , hence the ratio four in power .
moreover , the maximal angle of precession reached by auto - oscillations is found to be about @xmath60 in both microdiscs @xcite . finally , the disappearance of the signal as @xmath7 gets more negative is accompanied by a continuous broadening of the linewidth , which increases from a few mhz to several tens of mhz ( fig .
4d ) .
this rather large auto - oscillation linewidth is also consistent with a small precession angle , _
i.e. _ , a small stored energy in the yig oscillator @xcite . by repeating the same analysis as a function of @xmath6
, we can determine the bias field dependence of the auto - oscillations in both microdiscs .
the onset frequency and threshold current at which auto - oscillations start are plotted in figs .
4f and 4 g , respectively .
the onset frequency in the 4 @xmath0 m and 2 @xmath0 m microdiscs is identical and closely follows the dispersion relation of the main fmr mode plotted as a continuous line .
the small redshift towards lower frequency , which increases with the applied field , is ascribed to the joule heating and oersted field induced by @xmath7 ( the kittel law in figs . 2c and 4f
is obtained at @xmath61 ma ) .
we also note that the main fmr mode is the one which couples the most efficiently to our inductive electrical detection , because it is the most uniform .
hence , we conclude that the detected auto - oscillations are due to the destabilisation of this mode by sot . in order to reach auto - oscillations ,
the additional damping term due to sot has to compensate the natural relaxation rate @xmath62 in yig .
given the transparency @xmath63 of the yig@xmath4pt interface and the spin hall angle @xmath1 in pt , this condition writes : @xmath64 where @xmath65 is the section of the pt layer . the homogeneous contribution to @xmath62
is given by the gilbert damping rate , which for the in - plane geometry is : @xmath66 we remind that this expression is obtained by converting the field linewidth to frequency linewidth through @xmath67 . if only the homogeneous contribution to the linewidth is taken into account , the threshold current @xmath68 is thus expected to depend linearly on @xmath6 , as shown by the dashed lines plotted in fig .
4 g using eqs .
[ threshold ] and [ gilbert ] , and the parameters listed in table i ( the only adjustment made is for the 4 @xmath0 m disc , where the calculated @xmath68 has been reduced by 20% in order to reproduce asymptotically the experimental slope of @xmath68 _ vs. _ @xmath6 ) .
it qualitatively explains the dependence of @xmath68 at large bias field in both microdiscs , but underestimates its value and fails to reproduce the optimum observed at low bias field . to understand this behaviour , the finite inhomogeneous contribution to the linewidth
@xmath69 measured in fig .
2d should be considered as well .
in fact , this contribution dominates the full linewidth at low bias field . in that case
, the expression of the relaxation rate writes : @xmath70 the form of the last term in eq .
[ inhom ] is due to the kittel dispersion relation and is responsible for the existence of the optimum in @xmath68 at @xmath71 .
using the value of @xmath72 oe extracted in fig .
2d for the 2 @xmath0 m disc in eq .
[ inhom ] in combination with eq .
[ threshold ] , the continuous blue line of fig .
4 g is calculated , in very good agreement with the experimental data . to get such an agreement for the 4 @xmath0 m disc
, @xmath69 has to be increased by 25% compared to the value determined in fig .
2d . in this case ,
both the position of the optimum ( observed at @xmath73 koe ) and the exact value of @xmath68 are also well reproduced for the 4 @xmath0 m disc , as shown by the continuous red line in fig .
4 g .
hence , it turns out that quasi - degenerate sw modes , which are responsible for the inhomogeneous contribution to the linewidth , strongly affect the exact value and detailed dependence _ vs. _
@xmath6 of @xmath68 .
in fact , it is the total linewidth that truly quantifies the losses of a magnetic device regardless of the nature and number of microscopic mechanisms involved . even in structures with micron - sized lateral dimensions ,
there still exist a few quasi - degenerate sw modes as evidenced by the finite @xmath69 observed in fig .
2d . due to magnon - magnon scattering
, they are linearly coupled to the main fmr mode , which as a result has its effective damping increased , along with the threshold current .
the presence of these sw modes is also known to play a crucial role in sot driven dynamics .
the strongly non - equilibrium distribution of sws promoted by sot in combination with nonlinear interactions between modes can lead to mode competition , which might even prevent auto - oscillations to start @xcite .
we believe that the observed behaviours of the integrated power ( fig .
4d ) and linewidth ( fig .
4e ) _ vs. _ @xmath7 are reminiscent of the presence of these quasi - degenerate sw modes .
a meaningful interpretation of these experimental results is that as the fmr mode starts to auto - oscillate and to grow in amplitude as the dc current is increased above the threshold , its coupling to other sw modes whose amplitudes also grow due to sot becomes larger , which makes the flow of energy out of the fmr mode more efficient .
this reduces the inductive signal , as non - uniform sw modes are poorly coupled to our inductive detection scheme . at the same time , it enhances the auto - oscillation linewidth , which reflects this additional nonlinear relaxation channel . the smaller inhomogeneous linewidth in the 2 @xmath0 m disc ( fig .
2d ) results in a field dependence of the threshold current closer to the one expected for the purely homogeneous case ( fig .
4 g ) .
this indicates that reducing further the lateral size of the microstructure will allow to completly lift the quasi - degeneracy between spin - wave modes @xcite , as predicted by micromagnetic simulations , which show that this is obtained for lateral sizes smaller than @xmath74 m .
this could extend the stability of the auto - oscillation for the fmr mode , and experimental techniques capable of detecting sws in nanostructures @xcite should be used to probe this transition . very importantly for the field of magnonics , it was recently shown that this constraint on confinement could be relaxed in one dimension such as to produce a propagation stripe @xcite .
other strategies might consist in using specific non - uniform sw modes or to engineer the sw spectrum using topological singularities such as vortices , or bubbles , which could be most relevant to design active magnonics computational circuits . |
dark matter is supposed to be the principal driver of structure formation in the universe , and a whole industry has developed over the last few decades searching for the still elusive dark matter particle .
particle physics has provided some well - motivated candidates . among them , the neutralino , a weakly interacting massive particle ( wimp ) associated with a supersymmetric extension of the standard model of particle physics , is currently favoured ( see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* for reviews ) .
although this particle interacts with standard model particles only weakly , it may nevertheless be detectable .
recently the dark matter community has been stimulated by a variety of observed `` anomalies '' in cosmic - ray signals . among these
are : ( i ) an excess in the positron fraction recently ( re)measured by the pamela experiment @xcite ; ( ii ) an excess in the total flux of electrons and positrons measured by the fermi satellite @xcite ; ( iii ) an excess in diffuse microwave radiation in the general direction of the galactic centre ( the wmap haze
although these results may well find their explanation in ordinary astrophysical processes ( see , for example , @xcite for a pulsar explanation of the pamela results ) , it is alluring to relate them all to dark matter annihilation in the galactic halo .
in addition to these indirect signals , the dama / libra dark matter detection experiment has , for about a decade @xcite seen a significant amplitude modulation in its detector count rate .
although apparently incompatible with results from other experiments ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , the dama / libra collaboration interprets this as an annual modulation of the flux of dark matter particles through their detectors .
the small - scale structure of cold dark matter ( cdm ) haloes can , in principle , substantially influence the signal both in indirect and in direct dark matter detection experiments .
annihilation rates and detector count rates depend strongly on the density and energy distributions of particles local to other particles and to detector nuclei , respectively , and so are sensitive to small - scale fluctuations in these distributions .
for example , for cross - sections consistent with relic abundance constraints , the pamela data can be explained through annihilation only if rates are 100 to 1000 times those predicted for a locally smooth dark matter distribution , thus requiring substantial `` boost factors '' ( e.g. * ? ? ?
such boosts could come from non - standard particle physics , or they might reflect substantial inhomogeneities in the dark matter distribution on small scale . a population of abundant , self - bound subhaloes of very low mass but very high internal density has often been invoked in this context @xcite , but recent high - resolution simulations suggest that such subhaloes are neither dense enough nor abundant enough in the inner regions of @xmath0cdm haloes to have more than a minor effect on the observable signatures of annihilation @xcite . another possible boost mechanism is related to the fine - grained structure of dark matter haloes . before the onset of nonlinear structure formation , dark matter was almost uniformly distributed , with weak density and velocity perturbations and with very small thermal motions .
the particles thus occupied a thin , space - filling and almost three - dimensional sheet in the full six - dimensional phase - space .
subsequent collisionless evolution under gravity stretched and folded this sheet , but did not tear it .
thus , the dark matter distribution at a typical point in a present - day halo is predicted to be a superposition of many fine - grained streams , each of which has a very small velocity dispersion , has a density and mean velocity which vary smoothly with position , and corresponds to material from the vicinity of a different point in the linear initial conditions .
folds in the fine - grained phase - sheet give rise to projective catastrophes known as caustics where the spatial density and hence the annihilation rate is locally very high , limited only by the small but nonzero thickness of the phase - sheet ( e.g. * ? ? ?
* ; * ? ? ?
although the extraordinary improvement of n - body simulations in recent years has allowed many aspects of the dark matter distribution at the solar position to be predicted in considerable detail ( see , for example , * ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , such simulations are still very far from resolving the fine - grained phase - space structure which gives rise to caustics .
they are thus unable to provide a realistic estimate of how much caustics boost annihilation in the inner galactic halo .
fine - grained structure might also play a crucial role in the interpretation and modelling of dark matter signals in laboratory detectors like dama / libra .
it is unclear , for example , whether the maxwellian usually assumed describes the dark matter velocity distribution at the solar position accurately .
indeed , recent simulation work has shown that a multivariate gaussian is likely a better description , and that the assembly history of the milky way may be reflected in broad features in the particle energy distribution @xcite .
one may also wonder whether individual fine - grained streams might eventually be visible in detector signals as spikes at specific velocities .
axion detectors like admx , in particular , have very high energy resolution , and could in principle disentangle hundreds of thousands of streams .
again this is far beyond the resolution limit of current analysis techniques applied to even the highest resolution n - body simulations of halo formation . in recent work
we have developed an entirely new approach capable of following the evolution of fine - grained streams and their associated caustics in fully general simulations of halo formation .
this approach is based on integrating the geodesic deviation equation ( gde ) in tandem with the n - body equations of motion .
for every simulation particle and at every timestep it returns the spatial density and the velocity dispersion tensor of the fine - grained stream in which the particle is embedded .
this in turn allows all passages of the particle through a caustic to be identified and recorded .
in @xcite we presented this gde scheme and tested it on orbits in fixed potentials and on n - body simulations of static equilibrium haloes .
in @xcite we then applied the scheme to a simplified model of cdm halo formation : collapse from spherically symmetric , self - similar and linear initial conditions .
this showed that earlier work based on spherically symmetric similarity solutions ( e.g. * ? ? ?
* ; * ? ? ?
* ) was unrealistic because these solutions are violently unstable to nonradial perturbations .
these substantially alter the stream and caustic structure . finally in @xcite
we showed how the gde scheme allows the annihilation enhancement due to caustics and other fine - grained structure to be calculated explicitly by time - integration along particle orbits rather than by spatial integration over the particle distribution .
the current paper is the culmination of this programme and analyses for the first time the fine - grained structure of haloes forming from fully general @xmath0cdm initial conditions .
we resimulate haloes from the aquarius project , which studied six milky way mass objects at various numerical resolutions .
the coarse - grained structure of these haloes has already been studied in considerable detail in previous papers , e.g. the subhalo population in @xcite and @xcite , the dark matter distribution in the inner regions in @xcite and various radial profiles in @xcite .
the plan of our paper is as follows . in section 2
we briefly describe the initial conditions and the numerical approach we use to resolve the fine - grained phase - space structure .
section 3 begins by presenting results on one of the aquarius haloes at a variety of resolutions .
we demonstrate that , for fixed gravitational softening , convergent results can be obtained , free of significant discreteness or two - body relaxation effects .
we then analyse the implications for dark matter detection of the structure predicted for fine - grained streams and caustics .
finally , we use our full halo sample to tackle the question of how much scatter in fine - grained properties is expected .
we make some concluding remarks in section 4 .
an appendix shows that although discreteness effects are negligible in our high resolution simulations , their predictions for fine - grained stream densities ( but not for caustic structure ) are influenced by gravitational softening because this affects the tidal forces on particles passing close to halo centre .
we resimulate the six milky way - mass haloes of the aquarius project @xcite using our gde technique to follow the fine - grained phase - space evolution in detail .
the cosmological parameters assumed for these @xmath0cdm simulations are @xmath5 and @xmath6 , where all quantities have their standard definitions .
the haloes were selected at @xmath7 based on their mass , and were required to have no close massive companion at that time ; they are named aq - a to aq - f .
each halo was resimulated at a variety of numerical resolutions , indicated by a number @xmath8 in its full name , e.g. aq - a-3 .
the resolution levels differ in the particle mass and softening length employed , with 1 designating the highest resolution and 5 the lowest . in the following
we will use the same naming convention and mass resolution as in the original papers but different softening lengths .
this is because accurate integration of the geodesic deviation equation ( gde ) requires a larger softening length to achieve stable results than does integration of the particle trajectories themselves .
unless otherwise stated , we use a constant comoving plummer - equivalent softening of @xmath9 kpc in all our simulations .
the appendix examines the effect of softening on our results in some detail .
all other simulation parameters are the same as in @xcite , to which we refer the reader for further technical information . our experiments are carried out with the p - gadget-3 code @xcite with the gde modifications described in @xcite and @xcite , where we focused on a description of the relevant equations in physical coordinates and applied them to static potentials , to isolated equilibrium haloes and to halo formation from self - similar , spherically symmetric initial conditions .
the simulations of this paper take into account the full @xmath0cdm framework , using the cosmological parameters given above , and are carried out in comoving coordinates .
we therefore begin by describing how the gde and the associated tensors can be transformed from physical to comoving coordinates , and how this is implemented in our simulation code . for completeness
, we also describe how the physical stream density is calculated and how caustic passages can be identified .
the time - dependent transformation to the comoving frame used by our simulation code is given by vector and two of them denote a @xmath10 matrix .
an overline denotes a @xmath11 vector and two of them denote a @xmath12 matrix .
see also @xcite and @xcite . ]
@xmath13 where @xmath14 denotes the hubble parameter , @xmath15 the scale factor , and comoving coordinates are primed . in physical
coordinates the distortion tensor is defined as @xmath16 where @xmath17 and @xmath18 denote the initial position and velocity ( in physical coordinates ) , respectively .
it describes how a local phase - space element in physical coordinates is deformed along the trajectory of the particle while conserving its volume in order to obey liouville s theorem . in comoving coordinates the distortion tensor is accordingly defined as @xmath19 where all phase - space coordinates are now expressed in comoving coordinates .
the relation between physical and comoving distortion tensors can be derived from the differential relations , yielding @xmath20 where we have defined the two transformation tensors @xmath21 and @xmath22 we note that one of these transformation tensors is evaluated at @xmath23 whereas the other is evaluated at the initial time @xmath24 .
the reason for this is the time - dependence of the coordinate transformation in eq .
1 , where both the scale factor and the hubble parameter change with time . liouville s theorem guarantees that @xmath25 , so the comoving distortion tensor has the same conserved determinant as the physical one .
this means that local phase - space elements in the comoving frame conserve both volume and orientation as the system evolves . to integrate the comoving distortion tensor
, we need to derive its equations of motion .
we start with the physical gde equation @xmath26 and introduce the peculiar potential field @xmath27 which is related to the density field via poisson s equation @xmath28 where @xmath29 denotes the comoving mean background density and the laplacian is taken in comoving space .
we note that the force field driving the motion of simulation particles is also derived from this peculiar potential .
since the gde is directly related to the equations of motion of the particles themselves , it is natural to introduce a peculiar configuration - space tidal tensor @xmath30 with components @xmath31 @xmath32 we can use this tidal tensor to write the equations of motion for the comoving distortion tensor in the following form @xmath33 thus , the equations of motion for the comoving distortion tensor have exactly the same form as those for the physical one .
the only difference is the appearance of the scale factor @xmath15 in the comoving equations .
the initial conditions for these equations follow from those for the physical distortion tensor , taking into account @xmath34 @xmath35 based on these equations we can construct kick- and drift - operators for the leapfrog time integrator , similar to those for the position and velocity in the comoving frame . let us now discuss how to derive stream densities from these comoving quantities . each stream density is associated with a particular simulation particle , and describes the local density of the fine - grained stream it is embedded in . our goal is to derive an equation that yields the physical stream density from the comoving distortion tensor discussed above . to begin , we note that the configuration - space distortion tensor can be derived from the phase - space distortion tensor by applying two projection operators as follows @xmath36 where @xmath37 is the spatial gradient of @xmath38 , the mean initial sheet velocity as a function of initial position ( see * ? ? ?
* for a more formal definition ) .
caustic passages can be identified by sign changes in the determinant of this tensor .
we note that this is true both in physical and in comoving space .
the physical stream density can be calculated from the volume stretch factor implied by this linear transformation in physical configuration - space @xmath39 as we described in @xcite .
using @xmath40 and defining the peculiar velocity gradient @xmath41 we can write the physical stream density as : @xmath42\right|}.\ ] ] at early times when the dark matter is nearly uniform , all velocities and thus all velocity gradients are small and the second term in the determinant in this equation is small compared to the first .
thus we may approximate @xmath43 this equation is sufficiently accurate for our purposes and we will use it below to calculate the stream densities associated with each particle in our simulations . for consistency we will also neglect the ( small ) initial variations in density and will assume the cosmological mean value everywhere , i.e. @xmath44
the results in the next few subsections are based on resimulations of the aq - a halo at resolution levels 3 , 4 and 5 .
we note again that our simulations use a significantly larger softening length than the original aquarius simulations in order to mitigate the numerical noise sensitivity of the gde .
unless otherwise stated , we use a plummer - equivalent comoving softening length of @xmath45 . note that we do not change softening between the different resolution levels that we are going to discuss . we will demonstrate that we reach convergence at level 4 with this set - up .
additional experiments show that this is the smallest softening length for which we could achieve convergence at this resolution level ; smaller values require a larger particle number to converge . in the appendix
we show that our results for fine - grained stream densities ( but not for caustic counts ) are sensitive to the assumed softening because of its effect on the central structure of halo and subhalo potentials . where needed
, we also assume a dark matter particle with a velocity dispersion of @xmath46 today .
this corresponds to a neutralino of mass @xmath47 that decoupled kinetically at a temperature around @xmath48 mev .
to get started we show in fig .
[ fig : mean_density ] the spherically averaged density profile of the aq - a halo at redshift @xmath7 for all three resolution levels .
note that @xmath49 kpc for this halo .
convergence is excellent over the full radial range plotted , as is more obvious in the lower panel where we plot the difference between our profiles and that given by @xcite for the highest resolution simulation aq - a-1 .
softening clearly effects all three of our simulations similarly and only in the innermost regions ; the deviation is at the percent level at the radius corresponding to the sun s position within the milky way , but reaches 10% by 4 kpc . @xcite
made similar convergence tests for the radial profiles of a variety of quantities in the original simulations .
our focus in this paper is not on the coarse - grained but on the fine - grained structure of our haloes .
we begin by looking at the distribution of the number of caustics passed by particles in aq - a .
caustics are identified through changes in the sign of the determinant of the comoving configuration - space distortion tensor and are counted along the trajectory of each particle . in fig .
[ fig : phasespace_caustics ] ( top panel ) we show the distribution of this caustic count in a phase - space diagram at @xmath7 for aq - a-3 .
colour encodes the number of caustics passed by each particle , as indicated by the colour bar .
the particles were sorted by their caustic count before plotting , so that the particle with the highest caustic count is plotted in each occupied pixel .
this phase - space diagram can be compared directly to the one in @xcite which shows an isolated halo which grew from a spherically symmetric and radially self - similar perturbation of an einstein - de sitter universe . in both cases
the number of caustic passages increases towards the centre of the halo , due to the shorter dynamical timescales in the inner regions : caustic count is roughly proportional to the number of orbits executed by each particle , because caustics occur near orbital turning points .
the most important difference between the isolated and @xmath0cdm haloes is in the subhaloes and associated tidal streams seen in the @xmath0cdm case .
subhaloes stand out clearly in fig .
[ fig : phasespace_caustics ] since their particles have short orbital periods and undergo many caustic passages . in the bottom panel of fig .
[ fig : phasespace_caustics ] we eliminate self - bound subhaloes to show only particles belonging to the main halo . here
tidal streams corresponding to disrupted subhaloes can be traced very well .
material from such disrupted subhaloes has a different caustic count distribution than other main halo material , and so stands out in these phase - space plots . in fig .
[ fig : phasespace_caustics_zoom ] we show two zooms into the phase - space structure around the most prominent subhalo of fig . [ fig : phasespace_caustics ] .
these show the structure of the tidal streams in more detail .
we note that these tidal streams are _ not _ equivalent to the fine - grained streams we will discuss below . as we will see , the latter are not visible in such phase - space plots because of their large number and their low densities .
caustic counts can be used to study how dynamical mixing is related to particle location .
to this end , we filter the particle distribution by number of caustic passages and plot the result in three orthogonal projections in fig . [
fig : caustic_nums ] .
the top row of this figure shows that particles that have passed no caustic are almost homogeneously distributed , with no clear structure .
particles that have passed exactly one caustic already delineate the large - scale structure of the density field .
the main filament passing through the primary halo is visible , as is the halo itself and additional smaller filaments pointing towards it .
the halo is even more dominant in the distribution of particles that have passed two or ten caustics .
the bottom row of fig .
[ fig : caustic_nums ] shows particles that have passed at least @xmath50 caustics .
these are found only near the centres of the main halo and of the more massive subhaloes .
these are the densest regions with the shortest orbital times .
typical values of the caustic count in various regions thus indicate their level of dynamically mixing .
large values correspond to well - mixed regions whereas small numbers correspond to dynamically `` young '' regions .
particles that have passed no caustic are still in the quasilinear phase of structure growth
. we can compress the caustic count information in a radial profile .
this is shown in fig .
[ fig : caustic_passages ]
. profiles of the median caustic count for aq - a-3,4 and 5 are plotted as solid curves .
as already demonstrated in @xcite , the numerical identification of caustics is very robust , and as a result the number of caustic passages is little affected by numerical noise .
this is why the median profile is almost independent of resolution in fig .
[ fig : caustic_passages ] , with remarkably good agreement in the inner halo . in the outer regions subhaloes
show up as peaks in the caustic count and small shifts in their position between the different simulations show up as apparent noise . for the highest resolution simulation , aq - a-3
, we also plot the upper and lower 1 , 5 and 25% points of the count distribution at each radius . these parallel the distribution of the median count and span an order of magnitude at each radius .
the median profile can be compared to that given by @xcite for an isolated halo growing from self - similar initial conditions . at a given fraction of the virial radius ,
the typical number of caustic passages is only a few times larger in the more complex @xmath0cdm case .
the increase in caustic count towards halo centre can also be seen in fig .
[ fig : caustic_hist_rad ] ( top panel ) , where we present count histograms for particles in a set of nested spherical shells .
the shift of the distributions towards lower counts with increasing radius is very obvious , and within about 30 kpc this is accompanied by a suppression of the high - count tail . at larger radii , however , the distributions always extends out to about 200 counts .
this tail is contributed by particles from present or recently disrupted subhaloes ( see below ) . in the outer halo ,
the majority of particles have nevertheless passed relatively few caustics . as noted above ,
simulations using the gde technique require significantly more gravitational softening than standard n - body simulations in order to limit discreteness noise in the tidal field which otherwise leads to an unphysical , nearly exponential decay in stream density .
caustic identification is , however , relatively stable against such effects , so when studying caustic counts it is possible to reduce the softening .
we show the effects of this in the bottom panel of fig .
[ fig : caustic_hist_rad ] which compares the caustic count distribution within 0.16 mpc in our standard resimulation with that in two additional resimulations with two and four times smaller softening .
smaller softening results in better resolution of the innermost regions both of the main halo and of subhaloes , and thus to shorter dynamical times and larger caustic counts for particles orbiting in these regions .
this is evident as an extension of the high - count tail with decreasing softening .
notice , however that the region affected is more than two orders of magnitude below the peak of the distribution .
the shift towards higher counts only affects a few percent of particles that pass close to halo centre , and the distribution is essentially unaffected below a count of about 400 .
in the standard cdm paradigm , nonlinear evolution in the dark matter distribution can be viewed as the continual stretching , folding and wrapping of an almost 3-dimensional `` phase - sheet '' which initially fills configuration space almost uniformly and is confined near the origin in velocity space ( see , for example * ? ? ? * ) .
caustics are one generic prediction of such evolution .
another is that the phase - space structure near a typical point within a dark matter halo should consist of a superposition of streams , each of which has extremely small velocity dispersion and a spatial density and mean velocity which vary smoothly with position .
if the number of streams at some point , for example at the position of the sun , is relatively low , the resulting discreteness in the velocity distribution could give rise to measurable effects in an energy - sensitive detector of the kind used in many dark matter experiments . to assess this possibility
, we need to estimate the number and density distributions of streams at each radius .
our gde formalism makes this possible by providing a value for the density of the fine - grained stream associated with each simulation particle .
the set of stream densities corresponding to particles within some spherical shell is thus a ( mass - weighted ) monte carlo sampling of the stream densities at that radius . in fig .
[ fig : stream_density ] we show the distribution of fine - grained stream density as a function of radius in halo aq - a .
continuous curves give radial profiles at @xmath7 for the median stream density of particles at three numerical resolutions , while broken curves compare the tails of the distributions for the two highest resolutions .
agreement is excellent at all radii and throughout the distribution for aq - a-3 and aq - a-4 , but the lower resolution simulation aq - a-5 gives lower stream densities at all radii ( and at all points in the distribution , although we do not show this in order to avoid confusing the plot ) .
this is a consequence of discreteness noise in the tidal tensor which affects integration of the gde .
we note that fig .
[ fig : caustic_passages ] and fig .
[ fig : stream_density ] demonstrate that the caustic count and stream density distributions are independent of particle number at our two highest resolutions .
thus , discreteness effects , in particular 2-body relaxation , have no significant influence on the evolution . in the appendix we show that while the caustic count distribution is , in addition , independent of the assumed gravitational softening , the same is not true for the stream density distribution .
this is because softening influences the tidal forces on particles which pass within one or two softening lengths of the centre of a halo or subhalo , and so influences the evolution of their stream density . as we discuss in the appendix , this affects a surprisingly large fraction of halo particles at some point during their evolution , skewing the stream density distribution out to quite large radii . for our standard softening length of 3.4 kpc , the effects are similar in amplitude to those expected from our neglect of the baryonic component , so we do not attempt to correct for them . a remarkable result from fig .
[ fig : stream_density ] is that the median stream density depends very little on radius , and is within an order of magnitude of the cosmic mean density at all radii .
we found a very similar result in @xcite for collapse from self - similar and spherically symmetric initial conditions , so it is likely to apply quite generally to collisionless , nonlinear collapse from a smooth and near - uniform early state .
it implies that the dark matter distribution from the immediate neighborhood of a randomly chosen point in the early universe is almost equally likely to be compressed or diluted relative to the cosmic mean by subsequent nonlinear evolution , and furthermore that whether it is compressed or diluted is almost independent of its final distance from halo centre . the thin dashed and dotted lines in fig .
[ fig : stream_density ] show the 0.5 , 2.5 , 10 , 90 , 97.5 and 99.5% points of the stream - density distribution as a function of distance from the centre of our aq - a-3 and aq - a-4 simulations .
this distribution is very broad , spanning almost 12 orders of magnitude near halo centre . despite this , within @xmath51
even the upper 0.5% tail of stream densities lies below the mean density of the halo which is shown as a solid black curve for comparison . at 8 kpc ,
the equivalent of the solar radius , fewer than 1% of all dark matter particles are part of a fine - grained stream with density exceeding 1% of the local mean halo density .
this is a first indication that fine - grained streams are unlikely to influence direct detection experiments strongly .
the total number of streams at a typical point in a radial shell can be estimated from the stream density distribution as the ratio of the mean halo density in the shell to the harmonic mean of the stream densities of the particles it contains .
this number is dominated by the extended low - density tail of the stream - density distribution ; a very large number of very low - density streams is predicted in the inner regions .
a more relevant measure for dark matter detection can be obtained from the median stream density of the particles .
half of all events in a dark matter detector will come from streams with densities exceeding this value , and so will come from a number of streams somewhat less than half the ratio of the mean halo density to this median value . in fig .
[ fig : stream_number ] we show the results of such calculations for our resimulations of aq - a at various resolutions .
solid lines show the total number of streams at each radius estimated from the harmonic mean stream density , whereas dashed lines show the number of `` massive '' streams estimated from the median individual stream density .
the two estimates differ dramatically , particularly in the inner halo , reflecting the very broad distribution of stream densities at each radius , and in particular the presence of simulation particles with very low stream densities . as in fig.[fig : stream_density ] , convergence is somewhat poorer than for the caustic count statistics we looked at above , but the results for aq - a-3 and aq - a-4 nevertheless agree remarkably well .
this is particularly notable for the solid lines , given the sensitivity of the harmonic mean to the low - density tail of the distribution . from fig .
[ fig : stream_number ] we estimate the total number of streams at halocentric radii near that of the sun to be around @xmath1 and the number of `` massive '' streams to be about @xmath52 .
clearly cdm haloes are predicted to be very well mixed in their inner regions , and the velocity distribution near the sun should appear very smooth .
the very large number of streams predicted near the sun s position can be understood from the analysis in @xcite , which shows that stream density should fall inversely as the number of caustic passages in a one - dimensional system , but as the cube of this number for regular orbits in a three - dimensional system , and even faster for chaotic orbits .
given that typical particles near the sun have passed a few hundred caustics , typical stream densities are predicted to be @xmath53 .
the local mass overdensity at the solar position is @xmath54 , so one predicts @xmath55 fine - grained streams passing through our vicinity .
additional scatter is introduced by the fact that a significant fraction of the particles are `` pre - processed '' in smaller mass systems before they fall into the main halo .
we show the distribution of stream densities in a different form in fig .
[ fig : stream_hist_rad ] . for a series of spherical shells with mean radii increasing by factors of two , we have made histograms of the fine - grained stream densities of aq - a-3 particles , normalising them to unity so that their shapes can be compared . beyond 30
kpc these histograms are all quite similar , and resemble slightly skewed log - normal distributions . at the bottom of our plot , three orders of magnitude below peak , they span 14 orders of magnitude in stream density . at smaller radii ,
the low - stream - density tail becomes more extended , the peak of the distribution shifts very little , and the shape of the high - stream - density tail is unchanged .
the latter is determined by the behaviour near caustics . as we will see below the _ maximum _ densities at caustics
are predicted to be in the range explored by this high - mass tail .
it is notable that within 20 kpc these tails do not extend up to the local mean density of the halo .
the implications of these distributions for direct detection experiments on earth are most easily drawn from fig .
[ fig : f_rho ] , which shows the cumulative distribution of fine - grained stream density for aq - a-3 particles with @xmath56 at @xmath7 .
specifically , we plot the fraction of particles with stream density exceeding @xmath57 against @xmath58 , where @xmath59 is the mean halo density at 10 kpc .
the fraction of particles with @xmath60 is about @xmath61 , so the probability that a single stream dominates the signal in a direct detection experiment ( i.e. that the earth lies sufficiently close to a sufficiently strong caustic ) is about @xmath61 .
the fraction of particles with @xmath62 is about @xmath63 , hence the probability to see a single stream containing 10% of the signal is about 0.002 .
similarly , the fraction of particles with @xmath64 is about @xmath65 , so a single stream containing 1% of the signal will be seen with probability 20% .
finally , the fraction of particles with @xmath66 is about @xmath67 ; this means that at a typical `` earth '' location there will be a few streams which individually contribute more than 0.1% of the local dark matter density .
thus an experiment which registers 1000 true dark matter `` events '' should get a few duplicates , i.e. events coming from the same fine - grained stream and so having ( very nearly ) the same velocity .
if the dark matter is made of axions then a resonant detector should find an energy spectrum where a few tenths of a percent of the total energy density is concentrated in a few very narrow `` spectral lines ''
. given the very broad range of stream densities and caustic counts present at each point within our haloes , it is interesting to investigate how these properties are related to the dynamical history of individual particles .
we begin by studying stream density . as we saw above , the `` typical '' stream density of particles ( specifically , the local median value ) is almost independent of local mean halo density , thus of local dynamical time .
lower stream density corresponds to greater stretching of the initial phase - sheet and so , one might think , to more effective dynamical mixing .
an anticorrelation of fine - grained phase density with local orbital time ( i.e. with radius ) is , however , at most marginally evident in the stream density distributions of fig .
[ fig : stream_density ] and then only in their low - stream - density tails . to a first approximation
there is no correlation between stream density distribution and radius .
the overall mean density profile of the halo is built up by having more streams of every density at smaller radii .
thus the relation of stream density to dynamical mixing is far from evident .
we explore this issue in fig .
[ fig : evolve_decomp_s ] .
red and blue points here highlight the positions at a series of redshifts of particles which at @xmath7 lie within @xmath68 of halo centre .
we divided this region into a set of thin nested spherical shells and then for each shell we identified particles in the upper and lower 1% tails of the stream density distribution .
the particles with the highest @xmath7 stream densities ( which are typically close to caustics ) are plotted blue in each panel while those with the lowest stream densities ( which are typically in collapsed regions and far from caustics ) are plotted red .
thus at @xmath7 there are the same number of red and blue points in the plot , and both populations are distributed in halocentric radius exactly as the mass distribution as a whole .
the spatial distributions of the two populations differ markedly , however , both at @xmath7 and at all earlier times .
most of the red particles are part of clumps already at early times , while most of the blue particles either appear diffuse or are distributed relatively smoothly through the main halo at all times .
there is thus indeed a correlation between stream density and dynamical history , despite the lack of correlation between stream density and halocentric distance in fig .
[ fig : stream_density ] .
the number of caustic passages is a clearer measure of dynamical mixing , because it is closely related to the number of orbits a particle has executed over its entire dynamical history .
we examine this using fig .
[ fig : evolve_decomp_c ] which is constructed in exactly the same way as fig .
[ fig : evolve_decomp_s ] except that the particles in each @xmath7 spherical shell were ranked by caustic count rather than by stream density .
qualitatively the two series of plots show some similarities , but the differences between the red and blue populations are substantially more marked in the caustic count case .
the red points are almost all very close to the centres of collapsed clumps at all times , whereas the blue population is always diffuse except at low redshift when some of it is accreted onto the main halo . the evolution of the blue particles in fig .
[ fig : evolve_decomp_c ] shows an interesting shell - like structure which can be understood with reference to a spherical infall model .
many of the blue particles have yet to pass a caustic at @xmath7 , and so are on their first passage through the halo .
such particles come from a narrow range of lagrangian radii in the initial conditions , hence the apparent shell .
the fact that the caustic count is a much better indicator of the overall level of dynamical mixing than the stream density is easily understood ; the stream density varies strongly along the trajectory of each individual particle reaching high values each time it passes a caustic , whereas the caustic count increases monotonically with the number of orbits completed .
the number of fine - grained streams passing through the solar system is of interest for the search for dark matter using laboratory devices .
caustics , on the other hand , could be of interest for indirect dark matter searches that focus on the annihilation products of dark matter particles .
annihilation explanations for many of the apparent `` anomalies '' in indirect detection signals , for example the rise in the positron fraction detected by pamela above the expectation from secondary production , require annihilation rates to be substantially `` boosted '' above predictions based on `` standard '' particle physics assumptions and haloes with a locally smooth dark matter density field .
the annihilation rate per unit volume is proportional to the square of the local dark matter density and so can reach very high values in caustics .
caustics could thus , at least in principle , provide the necessary boost . to quantify this with our simulations ,
it is important not only to identify caustics , but also to estimate their peak densities , since emission around the peak dominates caustic luminosity ( see * ? ? ?
* ) . in fig .
[ fig : max_caustic_density ] we plot median peak caustic density as a function of radius for aq - a . to make this plot we recorded the peak caustic density and the radial position at which it was achieved for all particles that pass through a caustic in a short time interval around @xmath7 .
we bin these caustic passages into a set of radial shells , and then find the median value of peak caustic density for each shell .
the top panel shows results at three different resolutions , aq-3 , aq - a-4 and aq - a-5 .
again , it is clear that the lower resolution of aq - a-5 has significantly affected the results , while aq - a-3 and aq - a-4 agree reasonably well .
the peak caustic density calculation , like that of the stream density , depends on the full phase - space distortion tensor , so it is not surprising that the convergence properties of the two measures are similar . in the bottom panel
we compare the spherically averaged mass density profile of aq - a-3 with the radial distribution of peak caustic density .
the thick red line repeats the median profile from the top panel , while thin blue lines show the upper and lower quartiles of peak caustic density as a function of radius .
notice that at each radius the distribution of peak caustic densities is very broad , mirroring the range of fine - grained stream densities seen in fig .
[ fig : stream_density ] .
the peak densities of almost all caustics in the inner part of the halo are substantially below the local mean halo density .
a quarter of all caustics have peak densities exceeding the local mean at about @xmath69 , but only outside @xmath70 do more than half of the caustics have peak densities higher than the local mean density .
the low peak densities predicted for caustics in the inner halo , suggest that any annihilation boost will be small .
this was already the case for the isolated , smoothly growing halo studied in @xcite . in a @xmath0cdm halo , additional small - scale structure should decrease stream densities , and thus peak caustic densities , even further . to investigate this
, we calculate the ratio of intra - stream annihilation ( both particles belonging to the same fine - grained stream ) to inter - stream annihilation ( the two particles belonging to different fine - grained streams ) .
the former can be calculated for each simulation particle from its gde - estimated stream density and can be integrated correctly through caustics using the formalism of @xcite .
the latter can be calculated from a sph kernel estimate of the local smooth dark matter density . for the boost to be substantial
the ratio of these two annihilation rates should be large . in fig .
[ fig : boostprofile ] we plot this ratio for aq - a as a function of radius , after averaging the rates over thin spherical shells .
again , results for aq - a-3 and aq - a-4 agree well , while those for aq - a-5 are clearly affected by its low resolution .
this reflects the similar effects seen in earlier plots of stream and caustic densities .
[ fig : boostprofile ] shows that the contribution of caustics to the overall annihilation luminosity is very small boosting is completely negligible in the inner halo , and is still small ( about a factor of 1.1 ) near the virial radius .
this is even lower than the already small effects seen in the smooth halo collapse simulation presented by @xcite . since an annihilation interpretation of the pamela results requires a boost factor of 100 to 1000 , caustics are clearly far too weak to provide an explanation .
since bound subhaloes are also unable to produce local boosts much above unity near the sun @xcite , non - standard particle physics such as sommerfeld enhancement ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) must be invoked to explain the pamela data through annihilation . in the last few subsections we discussed the fine - grained structure of aq - a in considerable detail and studied how it is affected by numerical resolution .
it is not , of course , clear how representative these results are , since they are based on a single milky way - like halo . to quantify the scatter expected as a result of variations in formation history , environment , etc . , it is important to check some of these properties for other haloes .
the aquarius project simulated six milky way - mass haloes at very high resolution , and so provides an opportunity to address this issue .
the question of how representative these particular haloes are of the full population of similar mass objects is discussed by @xcite . here , we concentrate on the caustic count profile which we showed in fig.[fig : caustic_passages ] to be very well converged in aq - a-3 , aq - a-4 and aq - a-5 , at least within @xmath71 where the `` noise '' due to substructure is small . given this robust result , we decided for the following comparison to rerun the other five aquarius haloes at resolution level 4 in order to save computation time .
all resimulations were carried out with our standard softening length of @xmath9 kpc . in the top panel of fig .
[ fig : caustic_com ] we show profiles of median caustic count as a function of radius for all six aquarius haloes .
we plot results within @xmath71 only , since at larger radii the profiles are subject to large stochastic fluctuations at the positions of massive subhaloes ( see fig.[fig : caustic_passages ] ) .
the variations between the six haloes are relatively small and are systematic with radius as can be seen in the lower panel of fig .
[ fig : caustic_com ] where we plot the ratio of each of the individual profiles to their mean .
differences are largest in the central regions but still scatter by less than 20% around the mean .
there is some correlation of the caustic count profile with the radial mass density profile .
haloes a and c have the most concentrated mass density profiles ( see @xcite ) and also have the highest caustic counts in the inner regions , as might naively be expected .
halo e , on the other hand , is the least concentrated of all the haloes and yet also has a relatively high caustic count near the centre .
clearly , the details of halo assembly history do affect the median number of caustic passages significantly .
this is not determined purely by the local dynamical time of the final halo . in fig .
[ fig : stream_number_com ] we do the same exercise for the number of streams . for each of our resimulated aquarius haloes
we use the median stream density of the particles at each radius to estimate a characteristic stream number , as in the dashed curves of fig .
[ fig : stream_number ] which demonstrate that good convergence is achieved at resolution level 4 .
we plot the profiles out to @xmath72 in this case ; at larger radii stochastic effects due to massive subhaloes again dominate the variations .
the top panel shows the profiles themselves , while the lower panel plots the ratio of each individual profile to their ( geometric ) mean .
halo - to - halo differences here are larger than for the caustic counts , but still relatively modest given the large dynamic range of the radial variation .
it is interesting that the ranking of haloes here is similar but not identical to that in the caustic count profiles of fig .
[ fig : caustic_com ] ; the most concentrated haloes a and c not only have the largest caustic counts in their inner regions , but also the smallest numbers of streams .
this is unexpected since a naive argument might have led one to expect that a larger number of caustics would correspond to a _ larger _ number of streams .
we have demonstrated how n - body simulation techniques can be extended to follow fine - grained structure and the associated caustics during the formation of dark matter haloes from fully general @xmath0cdm initial conditions .
we have shown that , for the large particle numbers of our standard experiments , our integrations of the geodesic deviation equation ( gde ) produce distributions of fine - grained stream density and caustic structure which are independent of particle number , hence insensitive to discreteness effects such as two - body relaxation .
this requires somewhat larger gravitational softenings than are used in traditional n - body simulations , and we show in the appendix that while caustic count distributions are insensitive to the assumed softening , the same is not true for the stream density distributions .
we have resimulated the six milky way - mass haloes of the aquarius project , allowing us to analyse the scatter in fine - grained properties among haloes of similar mass . by identifying caustic passages along its trajectory , we are able to assign a cumulative caustic count to each simulation particle which is robust against numerical noise and serves to indicate the extent of dynamical mixing in its phase - space neighborhood .
this caustic count provides an excellent means to highlight subhaloes and tidal streams in @xmath73@xmath74 phase - space plots , because particles which became part of condensed structures at early times complete more orbits , and so pass more caustics than particles which remain diffuse until accreted onto the main halo .
filtering the particle distribution by caustic count decomposes it into interpenetrating components which have experienced different levels of dynamical mixing .
particles that have passed no caustic form an almost uniform subcomponent , while particles with large caustic count ( e.g. @xmath75 ) are found only in the dense inner regions of haloes .
particles with intermediate count outline the skeleton of the cosmic web .
direct dark matter detection experiments are , in principle , sensitive to the fine - grained structure of the milky way s halo at the position of the sun .
it is especially important to know if a significant fraction of the local dark matter density could be contributed a one or a few fine - grained streams , since each of these would be made of particles of a single velocity with negligible dispersion .
if many streams contribute to the local density and none is dominant , then a smooth velocity distribution can be assumed .
our simulations show the unexpected result that the distribution of fine - grained stream density is almost independent of radius within the virialised region of dark matter haloes .
only the low - density tail of the distribution appears to extend downwards at small radii .
because of the extreme breadth of the distribution , a very large number of streams ( @xmath76 ) is predicted at the solar position , but about half of the total local dark matter density is contributed by the @xmath52 most massive streams .
the most massive individual stream is expected to contribute about 0.1% of the local dark matter density .
this is potentially important for axion detection experiments which have extremely high energy resolution and may be able to detect 0.1% of the total axion energy in a single `` spectral line '' .
apart , from this possibility , dark matter detection experiments may safely assume the velocity distribution of the dark matter particles to be smooth .
it has been suggested that the very high local densities associated with caustics might significantly enhance dark matter annihilation rates in dark matter haloes , and thus be of considerable significance for indirect detection experiments .
indeed , for completely cold dark matter it can be shown that the contribution from caustics is logarithmically divergent and hence dominant @xcite .
this divergence is tamed by the small but finite initial velocity dispersions expected for realistic cdm candidates , and calculations for idealised , spherical self - similar models indicate quite modest enhancements for standard wimps @xcite .
our methods allow us to identify all caustics during halo formation from fully general @xmath0cdm initial conditions .
we find that they are predicted to make a substantially smaller contribution to the local annihilation rate than in the spherical model .
this is because the more complex orbital structure of realistic haloes results in lower densities for typical fine - grained streams and thus to lower caustic densities when these streams are folded .
the enhancement due to caustics near the sun is predicted to be well below @xmath77 and so to be completely negligible . only in the outermost halo
does the enhancement reach 10% .
the standard n - body technique of estimating annihilation rates from sph estimates of local dm densities ( see , for example , * ? ? ?
* ) should therefore be realistic , and such estimates will not be significantly enhanced by caustics .
similarly , caustics can not be invoked to provide the large boost factors required by annihilation interpretations of ` anomalies ' in recently measured cosmic ray spectra from the pamela or atic experiments .
given that unresolved small subhaloes also appear insufficient to provide a substantial boost @xcite , an annihilation interpretation of these signals is only tenable with nonstandard annihilation cross - sections .
our comparison of results for the six different haloes of the aquarius project showed that the halo - to - halo variation in the fine - grained properties we have studied is relatively small , and thus does not have significant impact on the applicability of our principal conclusions to the particular case of our own milky way .
thus our final conclusion must be that the fine - grained structure of @xmath0cdm haloes has almost no influence on the likelihood of success of direct or indirect dark matter detection experiments .
although we demonstrated that two - body relaxation and other discreteness effects have no significant influence on our conclusions , we found that our predicted stream density distributions are strongly affected by gravitational softening , which modifies the tidal forces on particles passing within a softening length or two of halo centre .
this numerical effect remains a significant uncertainty in our results .
the analysis in the appendix shows that reducing the softening below our standard value tends to reduce the density of streams , thus to suppress further the importance of fine - grained structure for detection experiments .
however , the change in orbit structure induced by our standard softening is similar in scale and amplitude to that caused by the neglected baryonic components of the galaxies , and even the sign of the effect of the latter is uncertain .
this is clearly an area where further work is required to reduce the remaining uncertainty in our quantitative conclusions .
the simulations for this paper were carried out at the computing centre of the max - planck - society in garching .
mv thanks stephane colombi , roya mohayaee and volker springel for helpful discussions .
a. a. , et al .
m. , 2009 , physical review letters , 102 , 181101 o. , barbarino g. c. , bazilevskaya e. a. , 2009 , , 458 , 607 n. , finkbeiner d. p. , slatyer t. r. , weiner n. , 2009 , , 79 , 015014 l. , 2009 , new journal of physics , 11 , 105006 l. , bringmann t. , edsj j. , 2008 , , 78 , 103520 r. , et al . , 2008 , european physical journal c , pp 167+ r. , et al . , 2010 , european physical journal c , pp 39+ g. , hooper d. , silk j. , 2005 , , 405 , 279 m. , springel v. , white s. d. m. , jenkins a. , 2010 , , 406 , 896 m. , strumia a. , tamburini m. , 2007 , nuclear physics b , 787 , 152 j. , kuhlen m. , madau p. , zemp m. , moore b. , potter d. , stadel j. , 2008 , , 454 , 735 j. , moore b. , stadel j. , 2005 , , 433 , 389 d. p. , lin t. , weiner n. , 2009 , , 80 , 115008 g. b. , 2006 , journal of physics conference series , 39 , 166 p. , gelmini g. , 2005 , , 71 , 123520 a. m. , hofmann s. , schwarz d. j. , 2005 , journal of cosmology and astro - particle physics , 8 , 3 j. , matsumoto s. , nojiri m. m. , 2004 , physical review letters , 92 , 031303 j. , matsumoto s. , nojiri m. m. , saito o. , 2005 , , 71 , 063528 c. j. , 2001 , , 64 , 063515 d. , finkbeiner d. p. , dobler g. , 2007 , , 76 , 083012 g. , kamionkowski m. , griest k. , 1996 , , 267 , 195 m. , silk j. , 2009 , , 79 , 083523 d. , cholis i. , gelfand j. , 2009 , , 80 , 063005 r. , shandarin s. , silk j. , 2007 , journal of cosmology and astro - particle physics , 5 , 15 r. , shandarin s. f. , 2006 , , 366 , 1217 b. , ghigna s. , governato f. , lake g. , quinn t. , stadel j. , tozzi p. , 1999
, , 524 , l19 a. , sikivie p. , 2006
, , 73 , 023510 a. , sikivie p. , 2008
, , 77 , 043531 j. f. , et al .
a. , 2010 , , 402 , 21 c. , gondolo p. , freese k. , 2004 , , 70 , 123513 a. , 1931 , annalen der physik , 403 , 257 v. , 2005 , , 364 , 1105 v. , et al .
j. , 2008a , , 391 , 1685 v. , white s. d. m. , frenk c. s. , navarro j. f. , jenkins a. , vogelsberger m. , wang j. , ludlow a. , helmi a. , 2008b , , 456 , 73 j. , potter d. , moore b. , diemand j. , madau p. , zemp m. , kuhlen m. , quilis v. , 2009 , , 398 , l21
m. , white s. d. m. , helmi a. , springel v. , 2008 , , 385 , 236 m. , et al .
, 2009a , , 395 , 797 m. , white s. d. m. , mohayaee r. , springel v. , 2009b , , 400 , 2174 s. d. m. , vogelsberger m. , 2009 , , 392 , 281
according to figs . [
fig : caustic_passages ] and [ fig : stream_density ] the caustic count and stream density distributions are independent of particle number at our two highest resolutions .
this demonstrates that discreteness effects , in particular two - body relaxation , are not significantly affecting the phase - space quantities we derive from our gde integrations . in this appendix
we investigate whether these distributions are sensitive to our assumed gravitational force softening . to facilitate this
, we keep the _ particle number _ fixed and change the _ softening length_. specifically , we resimulated aq - a-3 using softening lengths half and and twice our fiducial value , @xmath78 , but with all other parameters held to their original values . the top panel of fig .
[ fig : profile_check ] shows that the caustic count distribution is independent of softening in the main halo , but that its high count tail is affected by softening in subhaloes .
subhaloes are effectively small @xmath79 systems , and their centres are significantly better resolved with a smaller softening length .
this results in larger caustic counts at the centre of subhaloes , which are evident as enhanced `` noise '' in the high - count tail at larger radii for the simulation with the smallest softening .
this effect is almost absent in the main halo .
we conclude that our main results for caustic counts are robust to changes in softening length , a conclusion supported by the results of @xcite . the bottom panel of fig .
[ fig : profile_check ] shows a similar plot but now for the stream density distribution .
here we see a very different result .
the stream density distribution depends quite strongly on softening , with smaller softenings resulting in a shift of the entire stream density distribution towards lower values .
the effect is largest at small radii but is evident throughout the main halo . in the inner halo
the median stream density drops by more than three orders of magnitude when the softening length is reduced by a factor of 4 .
the difference is even larger if we focus on the low - density tail of the distribution and is still quite substantial in the high - density tail .
to understand the origin of this effect , we repeated the aq - a-4 simulation with half the fiducial softening and stored the particle data at all @xmath80 time - steps between @xmath81 and @xmath7 .
this allows us to follow the orbit and the phase - space density evolution of each particle in detail .
we find the @xmath82 most bound particles at @xmath7 and use their centre - of mass position to define the halo centre at all earlier times .
( we checked that they are all indeed still very close to halo centre at @xmath81 . ) for every halo particle we then construct trajectories of halocentric distance and fine - grained stream density , and define a minimum physical halocentric distance over any chosen scale factor interval ( @xmath83 ) , and a minimum physical stream density achieved between @xmath81 and any later epoch ( @xmath84 ) .
[ fig : hist_peri ] shows the @xmath7 distribution of all particles in the @xmath85 versus @xmath86 plane .
this histogram shows the expected trend that particles at small @xmath7 radii tend to have passed closer to halo centre in their past than particles at large @xmath7 radii .
interestingly , the typical minimum radius is more than an order of magnitude smaller than the final radius , and the scatter in the distribution is quite large .
thus at a third the virial radius ( around 80 kpc ) the distribution peaks at a closest approach distance of about 3 kpc , thus within our fiducial softening length . at a tenth the virial radius , most particles have passed within 1.0 kpc of halo centre .
thus , everywhere within the halo a significant fraction of particles have passed through a region where tidal forces are substantially affected by the softenings we are using . the tidal field sources evolution in the gde ,
so this affects our stream density estimates .
specifically , particles feel stronger shearing at pericentre if the softening length is smaller , so we expect larger stream density decrements for smaller softenings .
to investigate this further , we select a subset of particles according to the following criteria : halocentric distance at @xmath7 between @xmath87 and @xmath71 ; caustic count increment from @xmath88 to @xmath89 between @xmath90 and @xmath91 ; part of the main halo ( i.e. not part of any resolved subhalo ) at @xmath88 .
this selects a set of particles with similar orbital periods which have all been part of the main body of the system since @xmath92 .
the top panel of fig .
[ fig : hist_count ] shows the distribution of these particles in the @xmath93 versus @xmath94 plane .
the ratio @xmath93 indicates the decrease in minimal physical stream density between @xmath92 and @xmath7 , which , as expected , is strongly correlated with minimum halocentric distance .
particles with stream density changes substantially smaller than the median have typically never been closer to the centre than 2 or 3 kpc , while particles with substantially larger than average changes in stream density have almost all been within 3 kpc of halo centre , many within 1 kpc .
the bottom panel of fig .
[ fig : hist_count ] shows the same particle set in the @xmath93 versus @xmath95 plane , i.e. the drop in physical stream density from @xmath92 to @xmath7 versus the initial stream density at @xmath92 .
there is no correlation between these quantities , showing that the change in stream density after @xmath92 is independent of what happened to the particle at earlier times .
in addition , the distribution of stream density changes is broader than the initial distribution , showing that the main features of the final distribution were established over the time interval considered .
together these plots show that , for this particle subset , low stream densities at @xmath7 are associated almost exclusively with particles that pass within our fiducial softening length of the centre some time after @xmath92 . even in the main part of the stream density distribution about
half of the particles pass within 3 kpc of halo centre .
as we will see below , the time - average halocentric distance of these particles is about 40 kpc .
this explains why changing our softening between 1.7 and 6.8 kpc has effects out to large radii , as seen in the stream density distributions of fig .
[ fig : profile_check ] , and why the effects are stronger near halo centre and in the low - density tail of the distribution .
the stored data for our test simulation allow us to trace the detailed evolution of stream density for individual particles .
we illustrate this for a few particles in the extreme regions of the distribution labelled a to d in the bottom panel of fig .
[ fig : hist_count ] .
we pick @xmath96 particles at random from each region and plot in fig .
[ fig : time ] their physical stream density and physical halocentric distance as a function of scale factor .
these plots demonstrate that particles in fully general cdm halos show quite similar stream density behaviour to those in simpler halo models : caustic `` spikes '' occur several times per orbit at turning points of the fundamental oscillations , and secular evolution is evident in the steady decrease in the minimum stream densities achieved between caustic passages .
this resembles the isolated halo results in @xcite , showing that our gde scheme works correctly in the cosmological setup of this paper .
the selection criteria for our particle subset are evident in the similar apocentric distances and similar caustic and pericentric passage numbers of all orbits , despite their very different initial and final stream densities .
the stream density decrease is different in the four regions , reflecting their location in the @xmath93 versus @xmath97 plane .
our gde approach finds stream densities spanning almost @xmath98 orders of magnitude for orbits of very similar size and period .
the orbits in regions a and c look similar both in their stream density and in their radial orbit behaviour . the only difference is an offset of about 6 orders of magnitude in stream density which is preserved throughout the evolution .
this emphasises that evolution from @xmath92 to @xmath7 is independent of what happened before @xmath92 if one restricts oneself to main halo particles with similar orbital periods .
the particles in boxes b and d also have similar orbital periods , but they now have similar initial stream densities and very different stream density evolution .
those in box d barely change their stream density over the period shown , while the stream densities of those in box b drop by about 11 orders of magnitude . here
there are clear qualitative differences in orbital structure .
the particles in box d avoid the halo centre and in some cases are almost circular , and their caustics are relatively uniformly spaced in time and show rather little variation . in contrast , the particles in box b make much larger radial excursions , often passing close to the centre , and their caustic structure is more variable with caustic passages often occurring in groups .
this suggest that further work on the structure of the stream density evolution curves might allow classification of the orbits into tubes , boxes and other families . | we present the first and so far the only simulations to follow the fine - grained phase - space structure of galaxy haloes formed from generic @xmath0cdm initial conditions .
we integrate the geodesic deviation equation in tandem with the n - body equations of motion , demonstrating that this can produce numerically converged results for the properties of fine - grained phase - space streams and their associated caustics , even in the inner regions of haloes .
our effective resolution for such structures is many orders of magnitude better than achieved by conventional techniques on even the largest simulations .
we apply these methods to the six milky way - mass haloes of the aquarius project . at 8 kpc from halo centre a typical point intersects about @xmath1 streams with a very broad range of individual densities ; the @xmath2 most massive streams contribute about half of the local dark matter density . as a result
, the velocity distribution of dark matter particles should be very smooth with the most massive fine - grained stream contributing about 0.1% of the total signal .
dark matter particles at this radius have typically passed 200 caustics since the big bang , with a 5 to 95% range of 50 to 500 .
such caustic counts are a measure of the total amount of dynamical mixing and are very robustly determined by our technique .
the peak densities on present - day caustics in the inner halo almost all lie well below the mean local dark matter density . as a result caustics provide a negligible boost ( @xmath3 ) to the predicted local dark matter annihilation rate . the effective boost is larger in the outer halo but never exceeds about @xmath4 .
thus fine - grained streams and their associated caustics have no effect on the detectability of dark matter , either directly in earth - bound laboratories , or indirectly through annihilation radiation , with the exception that resonant cavity experiments searching for axions may see the most massive local fine - grained streams because of their extreme localisation in energy / momentum space . [ firstpage ]
cosmology : dark matter methods : numerical |
the announcement by the super - kamiokande collaboration@xcite of evidence for neutrino oscillation ( and hence nonzero neutrino mass ) in their atmospheric neutrino data is a major milestone in the search for new physics beyond the standard model .
an outstanding feature of these oscillations is the maximal mixing between the @xmath0 and @xmath1 ( @xmath3 ) in sharp contrast with the mixing pattern in the quark sector . also the inferred @xmath4 ev@xmath5 is lower than most `` see - saw motivated '' extrapolations from @xmath6 values in the small or large angle msw solutions to the solar neutrino problem@xcite : @xmath7ev@xmath5 with @xmath8 and @xmath9 ev@xmath5 with @xmath10 .
the solar neutrino problem provided the first indication for neutrino oscillation and this evidence keeps building up .
it can also be resolved by the maximal @xmath11 vacuum oscillation with fine tuned small mass difference @xmath12 ev@xmath5 .
maximal mixing with larger @xmath13 values yield an energy independent suppression of all solar neutrinos ( except when it is in the large angle msw range mentioned above ) . while this does not resolve the solar neutrino problem at present , it does considerably ameliorate it .
all the above suggests considering maximal ( @xmath11 ) mixing alongside maximal ( @xmath14 ) mixing@xcite .
three specific `` bimaximal mixing '' patterns@xcite having particularly simple forms are : _ case ( a)@xcite : _ @xmath15 where @xmath16 ; we will call this the symmetric mixing pattern .
_ case ( b)@xcite : _
@xmath17 this has been called in the literature as bimaximal mixing@xcite . _ case ( c)@xcite : _ @xmath18
we call this democratic mixing . in the above equations ,
we have defined @xmath19 as follows : @xmath20 with @xmath21 being the weak eigenstates and @xmath22 , the mass eigenstates .
our goal is to explore possible extensions of the standard model that may provide a theoretical understanding of the maximal mixing patterns .
attempts to understand the pattern ( a ) , made in our previous paper@xcite were largely unsuccessful . also the chooz upper bound@xcite on @xmath23 oscillation with @xmath24 ev@xmath5 , tends together with super - kamiokande data , to disfavor this possibility .
there has been several recent attempts to derive the second pattern ( b)@xcite . here we will show by using an extension of the standard model , that it is possible to generate the pattern ( c ) in a consistent and natural way .
our motivation is quite clear : if nature presents us with such a neutrino mixing pattern , we must seek an extension of the standard model that can naturally lead to it .
hopefully a theory that naturally provides this pattern will have other testable predictions . in ref.@xcite , permutation symmetry was imposed on the charged lepton mass matrix and not on the neutrino mass matrices in order to motivate the pattern ( c ) .
no underlying theoretical justification was discussed for such a hypothesis . in the framework of gauge theories ,
such an assumption is hard to understand apriori since the charged leptons and the neutrinos are members of the same isodoublet of the standard model gauge group @xmath25 and therefore the permutation symmetry could lead to a similar mass matrix for both the charged lepton sector as well as the neutrino sector . if that happens , the neutrino mixing matrix which is given by @xmath26 could substantially differ from ( c ) .
it is therefore important to investigate whether the above mixing pattern arises in a complete theory .
also the putative mass pattern @xmath27 should be provided by the model rather than arbitrarily fixed .
it is considerations such as these which motivate us to add this brief note to the exploding literature on neutrino models .
we find that by combining the permutation symmetry @xmath28 with a @xmath29 symmetry in the left - right symmetric extension of the standard model , we can obtain the maximal mixing pattern ( c ) in a technically natural manner ( i.e. without setting any parameters to zero by hand ) . in the limit of exact permutation symmetry , all the neutrinos are degenerate as are the elctron and the muon . as a result
, the mixing angles can be rotated away . however , once one admits permutation breaking terms to accomodate the electron muon mass difference , the neutrino degeneracy is removed and the democratic form ( pattern c ) of the maximal mixing pattern remains .
in fact , the masses of @xmath2 and @xmath0 get related to the elctron and muon masses arising completely from radiative corrections . to avoid arbitrary deviations from the maximal pattern , we assume that the permutation symmetry ( but not the @xmath29 ) is softly broken .
this adds only small , finite , corrections to the mixing pattern and one obtains a complete and realistic gauge theoretic derivation of the maximal mixing pattern c.
we consider a left - right symmetric extension of the standard model with the usual fermionic field content@xcite .
we omit the discussion of the quark sector for now . denoting the leptons by @xmath30 ,
the @xmath31 transform as the @xmath32 doublets respectively under the left - right gauge group @xmath33 .
the subscript @xmath34 stands for the generations .
we choose the following set of higgs bosons to achieve the symmetry breaking : three sets of left and right doublets dnoted by @xmath35 ( @xmath36 ) ; the @xmath37 vev will break the @xmath38 gauge symmetry down to the standard model @xmath39 group .
we choose three bidoublets @xmath40 to break the electroweak @xmath41 symmetry and give mass to the quarks and charged leptons as well as the dirac mass for the neutrinos . in order to implement the double seesaw@xcite mechanism for neutrino masses ,
we introduce three gauge singlet fermion fields , @xmath42 . in order to get the desired pattern for lepton masses , we demand the theory to respect the symmetry @xmath43 for all dimension four terms .
we assume that all interactions of dimension four are invariant under permutation of the three indices @xmath36 for the fields that carry the subscript @xmath34@xcite .
this symmetry will be softly broken by terms of dimension @xmath44 .
we assume that under left - right symmetry @xmath45 and @xmath46 the transformation of the various fields under symmetry @xmath47 is given in table i. the quark fields are assumed to be singlets under the above groups .
the yukawa couplings invariant under the above symmetries are : @xmath48 it is then clear that after the @xmath40 acquire vev s , they will give dirac mass to the charged leptons and the neutrinos . to get the desired pattern of charged lepton masses and the dirac mass for the neutrinos , we choose the vev pattern for the @xmath49 s as follows : @xmath50 . on the other hand , for the fields @xmath51 , we choose @xmath52 . due to left - right symmetry
, one can assure that @xmath53 .
it is crucial that the the vev pattern for @xmath51 is stable since this is what distinguishes the neutrino sector from the charged lepton sector and leads to the maximal mixing pattern ( c ) of democratic type for the neutrinos .
it is important for this that there be no tadpole terms involving the @xmath54 components of the @xmath51 fields .
this is verified by making the observation that all the @xmath55 have same @xmath56 quantum number ; as a result terms like @xmath57 which could generate the tadpoles are not present in the potential .
the above vev pattern has the consequence that all elements of the charged lepton mass matrix are nonzero whereas the dirac mass matrix for the neutrinos is diagonal . to see the resulting mixing matrix ,
let us write the charged lepton mass matrix : @xmath58 where @xmath59 and @xmath60 .
three eigen vectors of this mass matrix can be written as : @xmath61 where the particles in the above equation with superscript zero denote that they are the weak eigenstates prior to the diagonalization of the mass matrix .
note that this matrix is precisely the matrix in eq .
the corresponding eigenvalues are : @xmath62 it is then easy to see that if the majorana mass matrix for the neutrinos is diagonal , the consequent neutrino mixing matrix determined by the diagonalization of the charged lepton matrix above and is precisely of type ( c ) described above .
note however that the muon and the electron masses are equal .
in order for them to be much less than the @xmath63 mass as observed , we must have @xmath64 .
it is interesting to note that if instead of @xmath28 symmetry , one assumes @xmath65 then indeed one ends up with @xmath66 as has been noted already@xcite .
as alluded to before , we must have permutation symmetry breaking terms to split their masses . before proceeding to that discussion ,
let us turn to the neutrino sector to make sure that no mixing angles emerge in the neutrino sector that could vitiate the maximal pattern .
equation ( 3 ) and the @xmath67 vev pattern imply that the dirac mass matrix for the three neutrinos is diagonal with all @xmath68 given by @xmath69 . in order to understand the small neutrino masses
we must implement a seesaw mechanism .
it turns out that in this case the appropriate one for us is the double seesaw mechanism discussed in @xcite .
from the terms involving the gauge singlet fermions @xmath42 s in the lagrangian : @xmath70 the @xmath71 mass matrix comes out to be @xmath72 with @xmath73 providing the largest mass scale in the problem .
each of the entries in eq .
( 7 ) except the @xmath74 is a @xmath75 unit matrix . in the limit of exact permutation symmetry
, @xmath76 would also be a unit matrix .
the light neutrino eigenvalues are given by : @xmath77 it is important to emphasize that there is no mixing in the purely neutrino sector so that in the basis where the charged leptons are diagonal , we have the desired maximal mixing pattern .
this in our opinion is the big model building challenge that we have solved in this article .
clearly , if permutation symmetry had not been broken by the different @xmath42 masses , the mixing matrix would have been arbitrary .
to get a feeling for the scale of new physics @xmath78 , we note that @xmath79 gev . therefore ,
assuming @xmath80 , we get @xmath81 gev ; and for @xmath82 gev and @xmath83 , we get @xmath84 ev . if we choose @xmath85 gev and @xmath86 , we get @xmath87 ev , which is in the range required to solve the atmospheric neutrino puzzle . at this stage
it might appear that the muon- and electron - neutrino masses can be chosen at will by adjusting the @xmath88 .
but this is not so since the muon and electron masses which are tiny at the tree level ( if we choose @xmath66 ) must also arise out of the mass splitting among the @xmath42 s at the one loop level .
the radiative contributions to the muon and electron masses arise from the diagram of type shown in fig.1 and we can estimate this contribution to be : @xmath89 where @xmath90 is the typical heavy higgs boson mass that appears in the loop .
we have also used the fact the vevs of @xmath91 satisfy the relation @xmath92 .
= 8.5 cm = 8.5 cm choosing @xmath93 and @xmath94 , @xmath95 and @xmath96 , we estimate @xmath97 note that since we need to get the entire masses for the muon and the electron from the one loop correction , we must choose @xmath98 gev and @xmath99 gev .
this then implies that @xmath100 ev and @xmath101 ev .
we thus see that @xmath102 relevant for solving the solar neutrino problem is @xmath103 ev@xmath5 .
this is comfortably in the right range for solving the solar neutrino problem using the large angle msw solution .
let us now discuss the vev pattern assumed in the preceding analysis .
two points need to be discussed are ( i ) the specific vev pattern for the field @xmath51 that differentiates the neutrino dirac mass from the charged lepton mass matrix and ( ii ) the induced @xmath104 vev .
note that due to the nontrivial transformation of the @xmath105 field under the @xmath106 symmetry , the only allowed terms involving it in the potential are @xmath107 , @xmath108 , @xmath109 .
similar thing happens for @xmath110 .
note further that the @xmath56 symmetry forbids terms like @xmath111 .
the absence of terms of the form @xmath112 guarantees that once we choose the vev of the form @xmath113 , there are no tadpole like terms that can destabilize that vacuum . finally the fact that under left - right symmetry @xmath114 guarantees there is a discrete symmetry between these two fields leading to a stable minimum with @xmath53 .
turning now to the second point , note that the potential involving the @xmath115 fields has the form @xmath116 where sum over @xmath34 has been omitted for simplicity . minimizing this
we get that @xmath117 .
a few comments about the model are in order .
\(i ) it is worth pointing out that in presence of the permutation symmetry breaking terms in the singlet fermion sector , there will be small deviations from the equality of the @xmath118 s and consequently of the scalar doublet masses .
but these effects are small and they do not alter any of our conclusions .
\(ii ) the quark fields are assumed to be singlets under @xmath119 .
therefore , their masses arise from the @xmath120 couplings only and thus are not constrained by the patterns in the lepton sector .
\(iii ) the lightest of the singlet fermions @xmath121 which couples to electrons can be produced at lep energies but has a cross section of order @xmath122 which at the highest lep energies is about @xmath123 cm@xmath5 and is thus practically invisible .
in conclusion , we have succeeded in constructing a natural gauge model for the democratic maximal mixing for neutrinos suggested by the present neutrino data if lsnd results are not included .
the models also predicts a small mass difference between the @xmath2 and @xmath0 as needed for the large angle msw solution to the solar neutrino problem . to the best of our knowledge ,
this is the first time that a gauge model for understanding the democratic lepton mass matrix in an extension of the standard model has been constructed .
the model is essentially an electroweak scale model with low scale for the right handed @xmath125 s and uses the double seesaw mechanism to generate small neutrino masses .
some of the new fermions of the model are light in the sense of collider physics .
but their couplings to known particles are weak and thus there is no conflict with existing data . | the recently announced super - kamiokande data on atmospheric neutrino oscillation seems to require a maximal mixing between the @xmath0 and @xmath1 within the conventional three neutrino picture .
it is then tempting to suggest as has been done in literature , that the solar neutrino deficit be also understood as resulting from a maximal mixing between @xmath2 and @xmath0 . in this letter
, we propose a left - right symmetric extension of the standard model where permutation symmetry leads to one of the maximal mixing patterns in a technically natural manner .
the double seesaw mechanism gives small majorana masses for neutrinos needed to understand the atmospheric as well as the solar neutrino puzzles .
umd - pp-98 - 132 |
for normal galaxies those not deriving their radio luminosity from a nuclear monster " the radio emission is a probe of recent star - formation @xcite . in particular ,
the emission below 30 ghz is thought to be largely dominated by synchrotron radiation from ultrarelativistic electrons accelerated in supernova remnants . as most of the remnants responsible for this non - thermal emission result from massive , short - lived stars , the resulting radio emission tracks recent star - formation .
further evidence in favor of this claim comes from the existence of a tight radio / far - infrared ( fir ) correlation , which has been found to hold over several orders of magnitude in radio luminosity @xcite .
the complex astrophysics connecting the integrated radio emission to supernova explosions has led to a historical preference for other methods .
normal galaxies , with quite low radio flux densities , are also usually below the sensitivity of typical wide - area radio surveys .
alternative indicators of star - formation rate ( sfr ) include continuum emission in both the near - ultraviolet and the far - infrared @xcite .
another common diagnostic , referred to later in this paper , is the use of integrated emission - line fluxes @xcite .
( for additional indicators , see the excellent review by @xcite . )
however , star - formation is not the only process that produces radio emission ; active galactic nuclei ( agn ) can also be strong emitters in the radio .
this ambiguity has led to much debate , particularly for the less well - studied sub - mjy population . on the one hand
, there is a widely held belief that the population of sub - mjy radio sources is dominated by star - forming galaxies @xcite .
for example , using an infrared - selected sample , @xcite found the radio / fir correlation persisted down to @xmath1jy flux densities , and they conclude that the radio emission of these sources is dominated by star - formation ( with redshift still unknown ) .
however , the hypothesis that there are a significant number of agn at these low flux densities has recently received some attention as well .
@xcite , in a theoretical paper , suggested that the flattening in the observed counts of the normalized local luminosity function below s @xmath0 1 mjy is due to a radio - quiet " agn population . in a follow - up observational study , @xcite concluded that radio - quiet agn comprise a significant population below @xmath0 300 @xmath1jy , and that they may , in fact , dominate .
studying the general properties of the low - luminosity radio population , however , has proven to be difficult ; the perennial trade - off between coverage area and depth discourages surveys that are both deep and wide . in this paper , we take advantage of the overlapping coverage of two surveys , first ( faint images of the radio sky at twenty centimeters ) and the sdss ( sloan digital sky survey ) , to study the faint radio emission of a large sample of optically - normal ( quiescent " ) galaxies . by selecting galaxies with no strong optical emission lines from star - formation or agn activity , we construct a sample that is likely to be inherently radio - quiet in nature . any radio emission we do find , then , must result from either extremely low - level sf and/or agn activity , or activity that is obscured in the optical . by observing how this emission behaves as a function of other sfr indicators
, we will attempt to discern the origin ( sf or agn ) of the radio flux , and , in the absence of agn activity , the radio emission will give us a handle on the amount of star - formation occurring in these galaxies .
@xcite found that galaxies with no h@xmath6 detectable in their sdss spectra were mostly normal early - type galaxies with predominantly red optical colors , though some have relatively blue uv to optical colors ( see also : rich et al .
they calculated that some of these galaxies had uv - derived sfrs as high as 1 m@xmath4 yr@xmath5 , and a significant portion of those showed structure , like traces of faint spiral arms , or disturbed light profiles .
they identified an evolutionary sequence from star - formation to quiescence via nuclear activity . using a sample of nearby ,
early - type galaxies in the moses project ( morphologically selected ellipticals in sdss ) , @xcite reached a similar conclusion .
@xcite looked at optically - quiescent galaxies with red rest - frame colors and found that they were virtually indistinguishable in the color - magnitude - concentration space from galaxies classified as low - ionization nuclear emission - line regions ( liners ) .
still , very little has been done on the radio properties of optically - quiescent galaxies , and we intend to use this relatively unexplored wavelength regime to add to our knowledge of such galaxies . for the analysis of such faint radio sources , we have utilized the stacking technique presented in @xcite , hereafter referred to as paper i. this technique significantly reduces the noise in the combined images , allowing the investigation of median radio properties for sources significantly below the first detection threshold without requiring countless more hours of telescope time .
in addition , stacking all of the fields guarantees that the radio data will not be restricted by survey limits . in de vries et al .
( 2007 ; hereafter , paper ii ) we applied this method to the analysis of low - luminosity agn , concluding that radio emission from star - formation can dominate over that due to the agn emission in composite systems .
we begin by describing the radio data and the details of the sample selection ( [ data ] ) .
we then go into the details of the stacking method ( [ stack ] ) . in [ results ]
we present our analysis and a detailed discussion of the results ; [ conclusions ] contains our concluding remarks . throughout this paper
we use the standard frw cosmology of h@xmath7 = 70 km s@xmath5 mpc@xmath5 , @xmath8 = 0.7 , and @xmath9 = 0.3 .
the 1.4 ghz radio data we present are taken from the vla first survey @xcite , which includes 9,030 @xmath10 sky coverage of the north galactic cap and equatorial strip , making it ideal for comparison with targets from the sloan digital sky survey ( sdss ) . with only @xmath0 90 sources @xmath11 at the @xmath12 detection limit , 99.98% of the 5 billion beam areas are ` blank ' ( paper i ) .
the data , obtained from 180 sec vla snapshots , have a typical rms of 0.15 mjy beam@xmath5 . by stacking 3000 such snapshots
, we can reduce the noise in the resultant image to approximately 3 @xmath1jy beam@xmath5 and achieve the sensitivity equivalent of 150 observing hours . to convert to rest - frame 1.4 ghz luminosity
, we perform a k - correction assuming an average radio spectral index @xmath6 = -0.5 .
this index provides a compromise between steep - spectrum radio - lobe emission and flatter spectrum emission from the agn cores or star - forming regions , and it is the same index adopted throughout paper ii .
the sample of quiescent galaxies was drawn from the sfr catalog of the mpa / jhu vacs @xcite .
the sfr catalog is a super - set of the galaxies discussed in @xcite and contains all galaxies in the sdss dr4 spectroscopic survey .
dr4 of the sdss has 6851 @xmath10 of imaging and 4681 @xmath10 of spectroscopic coverage .
( see @xcite for more information on dr4 ) .
our final sample of quiescent galaxies is complete to a stellar mass lower limit of 10@xmath13 m@xmath4 . to accurately classify the objects , brinchmann et al .
reanalyzed the 1-d sdss spectra using their own optimized pipeline as outlined in @xcite .
a galaxy was classified as either star - forming ( sf ) , low s /
n sf , agn , low s / n agn , composite ( c ) , or unclassifiable based on the line ratios of the four emission lines described in baldwin , phillips , & terlevich ( 1981 , hereafter bpt ) .
composite galaxies have line ratios intermediate between sf and agn . for the sf and low
s / n sf classes , the fiber sfr was obtained from detailed modeling of the emission lines , while for the agn , c , and unclassifiable galaxies , the fiber sfr was estimated by calibrating the relation between the specific sfr ( calculated for the star - forming galaxies ) and the measured d4000 value as defined in @xcite .
we will refer to the latter as d4000 " sfrs to differentiate them from sfrs obtained via emission lines . for both types ,
total sfrs were calculated using an empirically based aperture correction scheme based on the likelihood distribution of the specific star formation rate for a given set of colors @xcite .
to single out only quiescent " galaxies , we kept only those galaxies catalogued as unclassifiable .
the majority of this class are galaxies with no or very weak emission lines @xcite .
as the sdss uses fiber - fed spectrographs , this is a ` nuclear ' classification .
the resultant sample contains @xmath0 185,000 quiescent galaxies .
we obtain additional parameters of interest for each galaxy from the nyu value - added galaxy catalog ( nyu - vagc ) . as presented by @xcite ,
the nyu - vagc is a collection of galaxy catalogs derived from the sdss .
the latest version of the catalog accompanies the sdss dr4 .
we also make use of the derived ugriz absolute petrosian magnitudes available in the associated catalog of k - corrections @xcite .
as these data were originally computed using a different value for the hubble constant ( h@xmath7 = 100 ) , we adjusted the derived absolute magnitudes to agree with the standard cosmology used elsewhere in this paper .
the star - forming sample , first introduced in paper ii , consists of all galaxies classified as star - forming ( sf ) in the dr2 sfr catalog .
this sample has s / n @xmath14 3 in all four emission lines of the bpt diagram and falls below the conservative threshold for agn rejection .
star - formation rates quoted here were derived by @xcite via careful modeling of the emission lines , and they take into account dust attenuation and fiber - corrections .
note that these sfrs assume a kroupa imf @xcite .
for absolute magnitudes and derived parameters , we consulted the stellar mass catalog as presented by @xcite .
the agn sample was taken directly from paper ii and consists of objects from the @xcite agn compilation .
it should be noted that they use a more liberal agn cut than @xcite , and this sample therefore consists of objects classified as agn as well as those classified as composite in binchmann et al .
this is a simple naming - convention issue and does not affect our results .
it is important to remember , however , that objects in the agn sample are affected by star - formation to different degrees , with those closest to the cut having emission - line ratios more commonly associated with hii regions .
we use the d4000 sfrs ( as we do with the quiescent galaxies ) from @xcite .
0.5 , ( larger than normal , so the reader may observe the obvious sidelobes ) , and each pixel is 1.8 " on a side .
the cutouts comprising this image correspond to the galaxies making up the right - most circle in figure [ fig : sfr4 ] , prior to being converted to radio luminosity .
the image has a peak flux density of 90 @xmath1jy , with an rms of 6 @xmath1jy beam@xmath5 .
this image is virtually indistinguishable from any of the other high s / n images generated for this study , independent of the type of galaxy . ]
using the sample selection criteria described above , we compiled our optically - selected quiescent galaxy sample .
we then obtained 1.4 ghz data centered on each target position from the first survey database .
these square cutouts " of the radio sky were 0.2 on a side and had the typical first rms of @xmath15 = 0.15 mjy beam@xmath5 ; these cutouts constitute our 3-d image block `` .
the size of the images was chosen pragmatically based on limitations of idl memory .
since all the analysis is based on the value of the central pixel , the image size is not important . in any case , all stacked images have a fwhm @xmath16 8 '' and hence easily fit within the image format .
these blocks can be sorted by a variety of parameters ( e.g. , redshift , color , sfr , etc . ) and stacked to yield median radio flux density / luminosity as a function of the parameter of interest , with a bin size ( @xmath17 / bin ) determined by the radio survey depth required ( @xmath18 @xmath0 @xmath15 / @xmath19 ) . bins typically contained either 1500 or 3000 images each .
the median of the parameter of interest was determined for each bin . to obtain the associated radio luminosity ,
the cutouts were processed through an idl procedure which converted each individual image to units of radio luminosity using the target galaxy s measured redshift ( paper i ) and then output the median image and @xmath20 upper / lower bounds for each pixel . the basic algorithm for determining the error bars involved computing the probability that the true median lay between elements ( k , k+1 ) in the sorted image stack using binomial probabilities as described in @xcite . as these sources are all quite faint in the radio , we will simply quote the radio luminosity derived from the central pixel in each co - added image as done in paper i. the median was used as opposed to the mean because it is more statistically robust in the presence of outliers and/or contamination .
for example , @xcite note that @xmath0 2% of the unclassifiable galaxies have line emission , some quite significant .
the median will be little changed by a 2% contaminant .
the mean , on the other hand , is sensitive not only to outliers , but also to the threshold values imposed to try to weed out those outliers ( paper i )
. when we used means , we found the radio luminosities all increased by 1 - 2 orders of magnitude .
that being said , the interpretation of the median is slightly more complicated .
the median shifts toward the ` local ' mean in the case of low s / n data .
( see paper i for a more detailed discussion ) . once the radio data were stacked , we corrected for the so - called ` snapshot bias ' .
this phenomenon is somewhat analogous to the ` clean bias ' reported for the first and nvss surveys , where the result of the non - linear clean algorithm is that a small amount of flux from the above - threshold sources is redistributed over the field @xcite .
a similar bias for sub - threshold sources ( that have therefore never been cleaned ) was discovered and calibrated in paper i. for sources fainter than 0.75 mjy , only 71% of the true mean flux density is recovered per bin , while sources above 0.75 mjy are well - described by the addition of a constant 0.25 mjy clean bias .
since the majority of the target galaxies in this study are sub - threshold , we accounted for this snapshot bias via the constant multiplicative factor of 1.40 .
an example of a median - stacked image is shown in figure [ fig : fitsimages ] .
the image is 0.5 on a side and resulted from stacking 3000 quiescent lrgs .
the cutouts that went into this image were the flux density maps of the galaxies that were later converted to units of luminosity and stacked to give the right - most quiescent lrg bin ( circle ) in figure [ fig : sfr4 ] .
the image peaks at 90 @xmath1jy and has an rms of 6 @xmath1jy beam@xmath5 .
. the clear selection bias toward stronger radio sources is evident below @xmath21 @xmath0 17 ; this is a result of the sdss target selection algorithm ( see text ) . ]
in order to explore the radio emission of the quiescent galaxy sample , we used the median - stacking technique detailed in [ stack ] .
typical radio flux density levels attainable via this method are shown in the plot of flux density vs. apparent magnitude ( z - band ) in figure [ fig : flux ] .
each point is the result of median - stacking 3000 quiescent galaxies .
as figure [ fig : flux ] demonstrates , a pronounced selection effect becomes apparent when the data are represented in this manner .
this effect is the result of the sdss target selection algorithm which includes different samples ; the main galaxy sample , the luminous red galaxy ( lrg ) sample , and the quasar , star , or serendipity samples @xcite .
galaxy targets are selected for inclusion in the sdss main galaxy sample if they satisfy a number of criteria outlined in detail in @xcite .
approximately 55% of our sample of quiescent galaxies were targeted as part of the main galaxy sample .
the remaining 45% of our quiescent sample , save a small percentage of miscellaneous objects which we discarded , were targeted as part of the lrg sample . as described by @xcite , the lrg sample extends to higher z and to fainter magnitudes than the main galaxy sample and consists of objects that are luminous ( l @xmath14 3l@xmath22 ) and intrinsically very red . objects may also have more than one target flag set at a time . of particular concern to us
are those objects targeted for both the main and lrg samples .
since the lrg group is generally more restrictive , and for reasons that will become apparent in our [ results ] , we have chosen to keep these objects with the lrgs rather than the main galaxy sample .
as advised in @xcite , we then keep only those objects with the galaxy_red flag set and redshifts z @xmath14 0.2 .
these further restrictions account for the breakdown of the luminosity cut at lower redshifts and ensure the purity of the lrg sample .
if the data are first segregated into main galaxies and lrgs , a rather different picture emerges .
figure [ fig : flux_select ] again shows flux density vs. z - band apparent magnitude , but this time the sources targeted as part of the main galaxy sample are plotted as diamonds , while those targeted as part of the lrg sample are plotted as crosses .
the main galaxy sample show typical radio flux densities in the 10s of @xmath1jy , with an increasing number of southern equatorial strip sources comprising the low - magnitude tail . of these galaxies ,
fewer than 0.1% were found to be coincident with cataloged first sources .
the lrgs have significantly higher median radio flux density levels ( though still far below the first threshold for detection ) and exhibit a trend toward lower flux densities with increasing magnitude .
out of the lrg sample , 0.9% are found to be within 2 " of a source in the first catalog , a detection rate nearly ten times that of the main galaxy sample .
as mentioned in the introduction , radio emission can be used as a tracer of star - formation for galaxies with no contaminating nuclear activity .
we use the star - forming galaxy sample , described in section [ data ] , to obtain the relationship between star - formation rate and radio luminosity .
the sf galaxies have measured sfrs based on h@xmath6 and other optical emission lines , and we sorted the sample by this
h@xmath6-derived " sfr and binned / stacked the data using the method outlined in section [ stack ] .
the result , published in paper ii and repeated in figure [ fig : lumvssf ] , shows a tight correlation between radio luminosity and sfr for the sf galaxies which extends over more than two orders of magnitude in sfr .
the least - squares fit to the data gives log(l@xmath23 ) = ( 1.37 @xmath24 0.02 ) log(sfr ) + ( 27.67 @xmath24 0.01 ) .
there have been a number of studies deriving sfr from 1.4 ghz emission , many of which use the radio - sfr relation presented by @xcite .
this relation , and those like it @xcite assume a linear relationship between sfr and radio luminosity .
indeed , the tightness of the radio - fir correlation was first interpreted as evidence for a linear relationship .
however , several authors @xcite have shown that the synchrotron component , which generates @xmath0 90% of the emission at 1.4 ghz , depends non - linearly on the sfr .
@xcite found that , by decomposing the radio emission in the radio - fir relation into thermal bremsstrahlung and non - thermal synchrotron components , the non - thermal radio luminosity scaled as @xmath25 , even while the thermal emission was directly proportional to sfr .
@xcite argued that the radio - fir correlation is a conspiracy resulting from the non - linearity of both sfr indicators .
they find that @xmath26 goes as @xmath27 , a result which is very close to our derived power - law dependance of 1.37 .
figure [ fig : lumvssf ] shows our results for the quiescent galaxies overplotted on the sf galaxies and agn of paper ii .
the first thing to notice is the large difference between the behavior of the main and lrg - targeted samples . keeping in mind that the lrg bins do have significantly higher median redshifts than the main galaxy bins
, the samples nevertheless exhibit very different trends .
the lrg sample maintains relatively constant median radio luminosity even with optically derived sfrs ( from d4000 ) varying from 0.1 to 10 m@xmath4 yr@xmath5 , suggesting that the radio emission from these optically - normal lrgs is dominated by agn activity rather than star - formation .
i.e. , if some of the radio luminosity came from star - formation , there would be a dependance on sfr .
indeed , for lrgs with sfr @xmath28 1 m@xmath4 yr@xmath5 , less than 1% of the radio flux should originate from sf activity .
it has been pointed out to us that perhaps the sfrs in the mpa / jhu catalog were never intended to be appropriate for the lrgs in the catalog , seeing as they derived from star - forming galaxy colors and reach quite spectacular rates for galaxies believed to be red and dead .
this does not change our result .
whether the lrg sfrs are in the ballpark or complete nonsense , the median radio luminosity of the sample is simply too high to be attributable to sf alone .
one may ask how lrgs actually classified as agn compare .
we therefore compiled a sample of galaxies classified as agn or composite from the mpa / jhu sfr catalog that were also targeted as lrgs .
we found only @xmath0 600 such galaxies , not even enough for a full bin and not nearly enough to examine trends in the stacked behavior .
the partial bin we were able to make lies basically right on top of the quiescent lrgs , with a median sfr of @xmath0 4 m@xmath4 yr@xmath5 and an l@xmath29 of 10@xmath30ergs s@xmath5 hz@xmath5 .
figure [ fig : lumvssf ] therefore provides evidence that quiescent lrgs host agn , a result that has some interesting implications .
most current models of cosmic structure formation include agn feedback , and both radiative and mechanical heating of the radio mode " agn are now recognized as important in galaxy formation scenarios ( for e.g. , croton et al .
, 2006 , and granato et al . , 2004 ) . as radio emission from black holes often indicates the presence of jets , these data lend observational support to the existence of a mechanism for low - level feedback in these galaxies .
in addition , this result also gives a quantitative feel for the feedback subsequently implied .
the constant median value of l@xmath31 @xmath0 4 @xmath32 10@xmath33 ergs s@xmath5 hz@xmath5 radiated by these quiescent lrgs , while still technically low - luminosity , is over two orders of magnitude larger than that attributed to the ground state " of the optically - detected low - luminosity agn in paper ii . as to why the lrgs are so much more radio - luminous , perhaps this is evidence that more massive agns are more radio - luminous .
but speculation aside , it is true that here we are dealing with median statistics and therefore have no knowledge of the underlying luminosity distribution ( i.e. , whether it is bimodal , etc . ) .
we also can not say whether the spread in the distribution is due to time variability or static differences between galaxies .
however , by treating this value of radio luminosity as a time - average ( rather than ensemble - average ) , we get an estimation of the average luminosity of radio jets which may result in agn feedback in lrgs . , while the bottom panel ( b ) shows that of the bin with the highest sfr .
the insets show the spectral features associated with oii , @xmath34 , and nii.,title="fig : " ] , while the bottom panel ( b ) shows that of the bin with the highest sfr .
the insets show the spectral features associated with oii , @xmath34 , and nii.,title="fig : " ] we now turn our attention to the main sample galaxies in figure [ fig : lumvssf ] .
these galaxies show a trend in radio luminosity with sfr which intersects the sf sample around 1 m@xmath4 yr@xmath5 . the agreement is worse for higher values of sfr , where the main sample galaxies deviate systematically from the sf galaxies by a little over a factor of 3
. there are two competing theories for the origin of the radio emission from these galaxies : that it is due to sf , or that it is due to agn activity . the fact that l@xmath31 shows a dependence on d4000 sfr , and that it disagrees with the sf sample by no more than a factor of 3 over the whole range of sfrs
, suggests that it may arise in part from star - formation .
alternatively , one may argue we are seeing the results of a correlation between @xmath35 and mass that is due to agn activity , a valid argument that we elaborate on below .
however , since the @xmath35 is near the level of the sf galaxies , we think it only fair to entertain the notion that it is due to sf . if we assume this to be the case , the disagreement between the main sample and sf galaxies may be an artifact of the different methods ( d4000 vs. emission lines ) used by @xcite to measure sfrs for the different samples . although @xcite proved the reliability of these d4000 values for galaxies in the star - forming regions of the bpt diagram , that does nt guarantee that the calibration yields spot - on values for galaxies with suppressed emission lines .
@xcite argue that the d4000 - specific sfr calibration and the color - based aperture corrections are only completely accurate for the sf galaxies , since both methods are dependent on time - scale and other galaxy types have different star - formation histories . as calculated from d4000 , while the bottom panel ( b ) shows that of the bin with the highest sfr .
the insets show the spectral features associated with oii , @xmath34 , and nii.,title="fig : " ] as calculated from d4000 , while the bottom panel ( b ) shows that of the bin with the highest sfr .
the insets show the spectral features associated with oii , @xmath34 , and nii.,title="fig : " ] luminosity vs. sfr for the quiescent main and star - forming galaxy samples .
values for [ oii ] luminosity are rough estimates using the conversion from equivalent width to luminosity in @xcite with median g-band luminosity . ]
the wary critic might ask whether d4000 is measuring star - formation at all in this case , or whether agn emission is contaminating the d4000 break as well , leading to overestimation of the d4000 sfrs for the main sample .
we have considerable reason to believe that this , at least , is not the case .
for one thing , the absence of observable line emission from the agn makes it unlikely that the agn would contribute significantly to the continuum near the 4000 break . and
furthermore , even for optically - detected agn , @xcite showed that the agn contribution to the blue continuum is ignorable . to get a better feel for these objects , we take advantage of our large sample size and examine trends in median - stacked spectra .
we have created median composite sdss spectra for both the main galaxy sample and the lrgs as a function of sfr bin as shown in [ fig : lumvssf ] . these spectra , normalized such that the continuum value at 4000 is unity ,
are shown in figures [ fig : spectra ] and [ fig : lrgspectra ] , respectively . in each case , the top panel shows the spectrum for the lowest sfr bin and the bottom shows that of the highest sfr bin .
we should note that we also tried binning spectra by sdss z - band absolute magnitude , and the resulting composite spectra had line emission no more obvious than those spectra binned by sfr that we show here .
these high s / n composite spectra clearly show very weak emission features .
they do , however , show strong balmer line absorption ; strong enough to plausibly swallow most of the h@xmath6 emission .
this is supported by some [ nii ] emission even in the absence of h@xmath6 .
the inset of figure [ fig : spectra]b shows a zoomed - in view of this region for the high - sfr galaxies in the main sample .
h@xmath6 emission is just visible over the accompanying deep absorption feature .
some may argue that the [ nii]/h@xmath6 ratio is more easily explained by a liner ( low ionization emission - line region ) spectrum , and this may well be the case . however , there is also an obvious turn - up below @xmath0 3300 in the spectrum of the high - sfr main galaxies , and no sign whatsoever of a similar trend in the low - sfr main galaxies or lrgs . for both samples , the spectra with higher median
sfrs show greater [ oii ] emission , a useful proxy for star - formation in the absence of h@xmath6 @xcite .
( we should note , however , that @xcite found [ oii ] emission in red and post - starburst galaxies can be as strong as in star - forming galaxies . )
we have attempted to quantify this result for the main sample galaxies in figure [ fig : oii_vs_sfr ] .
the [ oii ] luminosities on the vertical axis were calculated by determining the equivalent widths and using the conversion to luminosity given in @xcite .
they are only rough estimates , however , as we used magnitude in the @xmath36-band rather than the b - band , and we used the median magnitude for the conversion of the composite equivalent width . for comparison
, we also plot the results for the star - forming sample .
the [ oii ] luminosities for the main sample do increase with sfr , but only weakly relative to the star - forming sample . if we assume the [ oii ] in the quiescent main sample traces sf just as reliably as in the sf sample , and if we use the sf galaxies to calibrate an [ oii]-sfr relation , the implied sfrs are all @xmath16 1 m@xmath4 yr@xmath5 .
there also appears to be a kink around 1 m@xmath4 yr@xmath5 , the same point at which the radio luminosities diverge , prompting speculation of a change in population at that point .
regardless , the weakness of the [ oii ] dependence on d4000 sfr may indicate that both sf and agn activity are contributing to the line emission , with the agn component dominating .
of course , why these galaxies have such weak line emission has been the question all along , and it pertains to both agn and sf .
are shown in the legend .
we fit a line to the quiescent galaxy data ; the least - squares fit gives log(@xmath35 ) = ( -0.92 @xmath24 0.02)@xmath37 + ( 7.5 @xmath24 0.4 ) .
additional error bars are overplotted on the star - forming galaxy points .
these larger error bars correspond to 10% / 90% of the sfr dispersion for the galaxies within each bin . ]
the real question here is which component ( agn or sf ) dominates the radio emission , and we do not believe that these composite optical spectra alone can answer that question . in a recent paper , @xcite use a sample of optically obscured , 24-@xmath1m - selected galaxies observed by spitzer in the first look survey ( fls ) to conclude that the 1.4 ghz luminosity is dominated by star - formation activity despite the presence of an agn component .
irac photometry sets the majority ( 66% ) in a redshift interval of z = 1 - 3 .
the same conclusion was drawn in paper ii for optically classified agn / composites in the local universe with sfr @xmath14 1 m@xmath4 yr@xmath5 . if agn activity is the source of the radio emission from the main galaxy sample , then one possible explanation for the weak line emission is dilution .
it has been shown that stellar continuum light of a host galaxy may dilute the nuclear emission lines of an agn and hinder its detection @xcite . @xcite
present evidence that this effect leads to an increasing number of misclassified weak - lined agn with increasing redshift in the sdss and other redshift surveys .
we test for this effect in figure [ fig : vollim ] , where we take a volume - limited subsample in a narrow optical luminosity range and assess the effect of redshift on radio luminosity versus u - r color .
note that the magnitude cut of -22 @xmath28 @xmath37 @xmath28 -23 has preferentially selected the more luminous galaxies ( in the radio as well as the optical ) .
if increasing redshift bins show larger luminosities , we might conclude that agn are contributing significantly to the radio emission . while there seems to be a slight trend ,
the effect is quite small and the error bars are large .
the null result is therefore inconclusive . , but with uv - derived sfrs from galex overplotted for each point as open squares .
though the uv - derived sfrs and the d4000 sfrs are discrepant by 1 - 2 orders of magnitude for the main galaxies , we find them to be tightly correlated , arguing for a common origin . ] if the radio emission of the main galaxy sample stems from star - formation , dilution might still be the culprit . to test this theory , we plot radio luminosity versus z - band absolute magnitude for the quiescent and star - forming galaxy samples in figure [ fig : sfr2 ] . the sfr axis on the right - hand side of this figure ( and those that follow )
utilizes the radio - sfr calibration equation obtained in the previous section using the star - forming galaxy sample .
if we suppose for the moment that the radio emission of the main sample quiescent galaxies originates with star - formation , then we can apply it to these galaxies as well ( it obviously does not apply to the lrgs given our above results ) .
we derived the mass ( upper ) axis from a linear least - squares fit to the plot of log(stellar mass ) versus @xmath37 , where the masses came from the stellar mass catalog @xcite .
the star - forming galaxy sample was first sorted by sfr and split into 4 subgroups to aid in the interpretation of the quiescent galaxy sfrs .
the median sfrs and ranges in median sfr for the bins in a particular subgroup are shown in the legend .
the stacks were reduced to 1500 objects per bin to allow for an adequate number of data points in each subgroup , and the resultant subgroups are plotted as different symbols for easy differentiation .
the data for the lrg and main samples were binned and stacked following our standard procedure , with the error bars representing @xmath38 in median radio luminosity .
we find that the quiescent main sample follow a power - law fit of log(l@xmath31 ) = ( -0.92 @xmath24 0.02)m@xmath39 + ( 7.6 @xmath24 0.5 ) . while the h@xmath6-based method does yield a large overall range of star - formation rates , the sf sample display a narrow range of sfrs for a given absolute magnitude of the host galaxy .
this result was previously confirmed by @xcite .
if we assume no agn contamination for the moment , then using stacked radio luminosity as an indicator yields sfrs for the quiescent main galaxies that are consistently 3 - 5 times lower than the sf galaxies .
keep in mind that we are not suggesting that the quiescent main sample galaxies are intrinsically the same as the line - emitting sf galaxies , which we already demonstrated to populate a tight sequence in sfr - m@xmath40 space .
but if we are to believe the trend demonstrated in figure [ fig : lumvssf ] , particularly for the low - l@xmath29 end , then one interpretation is that the strength of h@xmath6 needed for detection ( and for subsequent sfr - estimation ) is proportional to absolute magnitude .
this interpretation implies a sort of h@xmath6-dilution . to illustrate this point further
, we have taken the star - forming galaxies in figure [ fig : sfr2 ] and overplotted larger error bars representing the 10% / 90% range in sfr dispersion for the galaxies comprising each data point .
figure [ fig : sfr4 ] uses the same axes to show how the radio - derived star - formation rates for the quiescent galaxies compare to those available from @xcite in a different way than figure [ fig : lumvssf ] . the 8-point stars still represent the quiescent galaxies targeted in the main galaxy sample , while the objects in the lrg sample are shown as open circles in the upper left - hand region of the plot .
we have then overplotted ( as crosses and squares , respectively ) the median optically - derived sfr for each bin using the sfr axis on the right - hand side of the plot . for the quiescent lrg sample ,
the d4000 sfrs are roughly two orders of magnitude lower than those inferred from the radio data ; i.e. , there is excess radio emission over that which is attributable to star - formation .
this is again consistent with the hypothesis that the radio emission from these lrgs is dominated by agn activity .
the data for the main sample quiescent galaxies show somewhat better agreement .
the d4000 sfrs for the main sample overlap the radio - derived sfrs at a few m@xmath4 yr@xmath5 and an absolute magnitude of m@xmath39 = -22.5 , though the two estimates show different slopes with absolute magnitude .
if the quiescent main sample are star - forming , this discrepancy could simply be explained by the different methods used to calculate sfr between sf and unclassifiable " galaxies .
if the emission is dominated by an agn contribution , alternatively , it must scale with galaxy stellar mass to explain the observed correlations .
this is not surprising for star formation in a homogenous sample of galaxies , but one might argue that this is expected for agn as well , given the well - known scaling relation between black hole mass and bulge luminosity @xcite . our own results for the quiescent lrgs may even support this claim .
as we have discussed above , radio emission is an indirect indicator of star - formation .
the uv continuum , on the other hand , is thought to be directly proportional to the photospheric emission of young stars @xcite . still , it is not free from problems of its own , most notably significant dust obscuration . nevertheless , as another comparison
, we overplot uv - derived sfrs from galex @xcite in figure [ fig : sfr3 ] , corrected to our imf .
as our samples do not completely overlap , we have simply taken the median of the uv sfrs available for each bin . for our main sample
quiescent galaxies , there were generally @xmath0 300 galaxies ( per 3000-galaxy bin ) with uv sfrs , and we find these galaxies to have fairly representative radio luminosities ( i.e. , there is no selection bias between our samples ) .
none of the quiescent galaxies targeted as lrgs had uv - sfrs available from this catalog .
these uv - based sfrs were obtained by @xcite via sed - fitting of emission from the uv to the z - band with various models from the @xcite population synthesis code .
because the uv luminosity imposes the tightest constraint on the sfr , the authors refer to these sfrs as uv - derived " .
their quoted rates are averaged over a span of 100 myr and constitute the first test of the two - component dust attenuation model proposed by @xcite .
@xcite compare these rates with the optically - derived rates for the @xcite star - forming sample and report good overall agreement , with some systematic deviation due to differences in modeling dust attenuation .
it is therefore not surprising that we also find good agreement in the star - forming sample , as they are derived from the same catalog and served as our radio - sfr calibrators .
the interpretation of the uv and radio - derived sfrs for the main sample quiescent galaxies is , once again , less straightforward .
the uv - derived sfrs are between 2 - 3 orders of magnitude lower than the sfrs derived from radio - stacking . given the rough agreement of the radio and d4000 sfrs , it is then no surprise that @xcite find a @xmath0 2 order - of - magnitude discrepancy for the uv versus optically - derived sfrs of their no h@xmath6 " sample , which comprise roughly @xmath0 10% of our total main quiescent sample . they conclude that , since these galaxies fall largely on the red sequence , the uv sfrs are more realistic .
another of their arguments is that such high sfrs as @xcite estimate would result in a large fraction of their no h@xmath6 " galaxies having detectable h@xmath6 . on the contrary
, we have shown in figure [ fig : sfr2 ] that this result could also be explained if the strength of h@xmath6 required to constitute a detection " is simply a function of optical luminosity , with the levels predicted for these quiescent galaxies falling below the detection limit for a given absolute magnitude .
the strongest argument for the uv - sfrs is that the 2 - 3 orders of magnitude discrepancy between the radio and uv would require an underestimation of dust extinction of 5 - 8 magnitudes .
this is , admittedly , quite large , but perhaps the disagreement between sfr estimates could be explained by the location of the star - formation ; if the star - formation is dominated by nuclear star - formation , it might be easier to hide both the line - emission and the uv flux from observation .
this is speculative , but we would like to emphasize again that the radio emission and d4000 index are two independent checks on the presence of star - formation , and they are consistent with one another to within a factor of @xmath0 3 .
while they both can be higher than the true sfr for completely different reasons , we feel this is a coincidence that requires further explanation before the possibility of sf in these galaxies can be thrown out altogether .
we will , however , freely acknowledge that if the galex sfrs are right , then for sure the radio - derived sfrs are due to agn activity .
overall , whether this discrepancy is ultimately the result of an agn component to the radio emission or a poor dust - attenuation model is not a question we feel can be definitively answered here .
we have investigated the radio emission from a sample of @xmath0 185,000 optically - selected quiescent galaxies from the sloan digital sky survey . by stacking first cutouts centered on the galaxy positions ,
we reach median radio flux densities down to the 10s of @xmath1jy .
we calibrate the relation between 1.4 ghz radio emission and optically - derived sfr using a sample of star - forming galaxies , allowing us to look for star - formation in their quiescent counterparts .
the main findings of this study are summarized below . * we find a tight correlation between radio luminosity and sfr for the star - forming galaxy sample , with l@xmath41 going as ( sfr)@xmath3 .
* the quiescent galaxy sample consists of two subgroups : those targeted as part of the main galaxy sample , and those targeted as part of the luminous red galaxy ( lrg ) sample .
the former show partial overlap with the star - forming sample on a plot of l@xmath31 vs. optically - derived sfr , while the latter show a relatively constant , high value of median radio luminosity that does not vary with sfr .
* for the lrgs , the correspondingly high rates of radio - derived star - formation conflict with those estimated by @xcite via calibration of the 4000- break by two orders of magnitude on a plot of derived - sfr vs. m@xmath40 .
this result , combined with the relative constancy of l@xmath31 with d4000 sfr , suggests that the radio emission of these quiescent lrgs derives not from star - formation but from an agn component .
this implies the possible presence of radio jets and a mechanism for low - level feedback in these galaxies . *
the main sample galaxies , on the other hand , exhibit a l@xmath29-sfr trend which intersects the sf sample around 1 m@xmath4 yr@xmath5 and deviates by no more than a factor of 3 over the whole range of sfrs .
this agreement as well as the dependence of l@xmath31 with sfr suggests that the l@xmath31 might arise in part from star - formation ( particularly in so far as the agn of paper ii show no such dependence . )
if we assume the radio emission of these galaxies originates from star - formation , the deviation at high / low values of sfr might then be due to the different diagnostic measures used to estimate sfr for the two samples ( i.e. , 4000- break vs. emission - lines ) .
however , we can not rule out the possibility that the radio emission may have a large or even dominant agn component , with a l@xmath29-mass relation causing the observed correlation .
* composite spectra of the main sample show balmer absorption strong enough to plausibly swallow any h@xmath6 emission present .
this is supported by some [ nii ] emission even in the absence of h@xmath6 .
the [ oii ] emission , however , increases only slightly with d4000 sfr , giving reason to doubt the validity of d4000 ( and thus radio emission ) as a proxy of star - formation in these galaxies until we understand why the line emission is so highly - suppressed . *
the main quiescent sample shows a correlation between radio luminosity and z - band absolute magnitude of the form : log(@xmath26 ) = ( -0.92 @xmath24 0.02)@xmath37 + ( 7.6 @xmath24 0.4 ) .
the sfrs implied by radio emission reach above 10 m@xmath4 yr@xmath5 , yet lie just below the objects classified as star - forming " for a given absolute magnitude .
if we assume the radio emission stems from sf , one possible interpretation is that the strength of the optical emission lines needed to constitute a detection " varies with galaxy luminosity , and using our radio - stacking method allows us to probe low levels of star - formation despite sf dilution in the optical .
if we assume the radio emission stems from agn activity , then both the d4000 and radio - derived sfrs are too high ( for separate reasons ) , and the observed trend implies a relation between agn radio luminosity and galaxy mass . * uv - derived sfrs from galex are 2 - 3 orders of magnitude lower than the sfrs predicted by radio - stacking for the main quiescent galaxy sample .
this indicates possible agn contamination in the radio emission of the quiescent galaxies , incomplete modeling of the dust attenuation for the uv emission , or both .
the different processes at play obviously need to be better understood before we can choose the most useful sfr diagnostic tool for these optically - quiescent galaxies .
in short , we use a novel stacking technique to provide strong evidence that quiescent lrgs harbor agn .
this result has clear implications for galaxy formation scenarios , providing observational support to the existence of a mechanism for low - level feedback in these galaxies .
the results for the quiescent main sample galaxies are more nebulous , with the discrepancies between the different sf indicators clearly highlighting how much we still have to learn about these persistently unclassifiable galaxies .
overall , we believe the technique utilized here shows great promise as a tool to investigate further the radio emission of various sub - populations in the vast community of radio - quiet galaxies .
the authors wish to thank david helfand for sparking such lively discourse and being a champion of the championless ( views , that is ) .
they also wish to thank samir salim for kindly providing the uv data and taking the time to craft such a careful critique of the paper .
jah acknowledges support of the gaann fellowship from the u.s .
department of education , as well as a uc davis graduate block grant fellowship , and grant hst - go-10412.03-a from the space telescope .
rhb acknowledge the support of the national science foundation under grant ast 00 - 98355 .
rlw acknowledges the support of the space telescope science institute , which is operated by the association of universities for research in astronomy under nasa contract nas5 - 26555 .
the work by wdv and rhb was partly performed under the auspices of the u.s .
department of energy , national nuclear security administration by the university of california , lawrence livermore national laboratory under contract no . w-7405-eng-48 .
funding for the sloan digital sky survey ( sdss ) has been provided by the alfred p. sloan foundation , the participating institutions , the national aeronautics and space administration , the national science foundation , the u.s .
department of energy , the japanese monbukagakusho , and the max planck society .
the sdss web site is http://www.sdss.org/. the sdss is managed by the astrophysical research consortium ( arc ) for the participating institutions .
the participating institutions are the university of chicago , fermilab , the institute for advanced study , the japan participation group , the johns hopkins university , los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , university of pittsburgh , princeton university , the united states naval observatory , and the university of washington . | we investigate the radio emission of @xmath0 185,000 quiescent ( optically unclassifiable ) galaxies selected from the sloan digital sky survey ( sdss ) . by median - stacking first cutouts centered on the optically - selected sources , we are able to reach flux densities down to the 10s of @xmath1jy .
the quiescent galaxy sample is composed of two subgroups inhabiting vastly different regimes : those targeted for the sdss main galaxy sample ( @xmath055% ) , and those targeted for the luminous red galaxy ( lrg ) sample ( @xmath045% ) . to investigate the star - formation of these quiescent galaxies , we calibrate a radio - sfr conversion using a third sample of star - forming galaxies .
we confirm a tight power - law dependence for the star - forming ( sf ) sample , where l@xmath2 @xmath0 ( sfr)@xmath3 .
comparing this sfr - indicator with indicators in the optical and uv , we derive conflicting sfr estimates for the main sample quiescent galaxies .
these radio - derived sfrs intersect those calculated using the 4000- break ( d4000 ) around an sfr of 1 m@xmath4 yr@xmath5 and agree to within a factor of 3 over the range of sfrs .
however , we find that the radio - derived sfrs are too high relative to the sfrs estimated for similar populations of galaxies using analysis of uv emission , implying either contamination of the radio by active galactic nuclei ( agn ) or incomplete dust modeling . if agn activity is dominant in these galaxies , then a relation between agn radio luminosity and galaxy mass is required to explain the observed trends . for the lrgs , on the other hand , we find the radio luminosity to be relatively high ( compared to the sf galaxies ) and independent of sfr as derived from d4000 , indicating an agn component dominates their radio emission .
agn - based radio emission often implies the existence of radio jets , providing evidence of a mechanism for low - level feedback in these quiescent lrgs . |
for the first time the @xmath0 spectroscopy study was performed in novosibirsk in 60th at the vepp-2 collider .
the shape of the @xmath3 resonance have been measured @xcite . since a lot of different experiments for spectroscopy
have been done @xcite and as a rule the most precise data on vector meson parameters were obtained in @xmath0 production . the current status of the vector meson spectroscopy is following : * all main states of @xmath4 systems are established . *
charmonium and bottomonium families are well known .
* excitation states of @xmath4 system for @xmath5 quarks are not well established .
* there are evidences for existence of @xmath6 or 4-quarks states in the vector meson decays @xcite .
* there are evidences for existence of @xmath7 or 6-quarks states @xcite .
the main problems of the light vector meson spectroscopy connect with the fact that in the mass region @xmath8 total integrated luminosity @xmath9 was collected at dci and adone .
this statistic is incompatible with that collected in the energy regions of the charmonium and bottomonium families . at the contrary , in the low energy region from the hadron production threshold
to @xmath10 the systematic studies have been performed in novosibirsk at the @xmath0 collider vepp-2 m .
it was in operation from 1974 to 2000 and the total integrated luminosity @xmath11 was collected .
important measurements were done by olya @xcite and nd @xcite experiments , but the main part of integrated luminosity were taken by the cmd-2 @xcite and snd @xcite experiments .
now the experimental program is finished , and the final data analysis is in a progress .
main advantages of the experiments on vector mesons production in @xmath0 annihilation are following : clean initial state with the well defined quantum numbers , high mass resolution , good conditions for an exclusive reactions study .
the main problem of the @xmath0 data analysis connect with uncertainties in the interference between several resonances that often introduces model dependence into final results ( for example @xcite ) .
there are also model dependences of the data analysis @xcite , which can be resolved only after significant increasing of experimental statistic .
the precise measurement of the two pion production cross section have been performed for many years @xcite .
the systematic uncertainty of 0.6 % is achieved in the last cmd-2 experiment @xcite in the energy range below @xmath1 . for higher energies the results are not so precise , but dm2 data @xcite strongly emphasise the signal of @xmath13 ( fig.[pi2 ] ) .
there is some wide enhancement in the the cross section around @xmath14 which may be taken as an evidence for the @xmath15 resonance , but at the same time another models are discussed @xcite .
cross section.,width=302 ] the main contributions to three pion production cross section at low energy come from the @xmath17 and @xmath18 resonances .
it is well known that the interference between these resonances are destructive @xcite . for many years in the energy region above @xmath1 the experimental data was not so precise @xcite .
the last snd measurement @xcite shows that there are visible peak in the cross section at @xmath14 ( fig.[pi3v ] ) . after applying the radiative corrections and
the detection efficiencies the total cross section was obtained in which the clear resonance signal is seen . taking into account the data below @xmath1 and the dm2 data @xcite the set of fits
were performed @xcite .
the best fit ( fig.[pi3 ] ) requires contributions of @xmath19 , @xmath20 , @xmath21 , and @xmath22 with the relative phases ( + ) , ( - ) , ( - ) , ( + ) .
.,width=283 ] cross section.,width=283 ] the four charged pion production was studied by many groups @xcite .
the most detailed investigation have been reported by cmd-2 @xcite . in this work
the pwa analysis have been performed and it was shown that the @xmath24 intermediate state dominates in the energy region below @xmath10 .
the snd results @xcite confirm the cmd-2 data ( fig.[pi4c ] ) .
cross section with the recent vepp-2 m data @xcite.,width=302 ] using the pwa analysis of the reaction @xmath25 cmd-2 obtained @xcite that @xmath24 and @xmath26 intermediate states dominate in the reactions mechanism ( figs.[pi4o],[pi4a1 ] ) .
the recent snd data @xcite are in agreement with the cmd-2 results within the systematic uncertainty of the experiments ( fig.[pi4n ] ) . in the reaction @xmath25 @xcite.,width=302 ] @xcite.,width=302 ] cross section with the recent vepp-2 m data @xcite.,width=302 ] the five pions production cross section have been studied by cmd-2 @xcite and dm2 @xcite .
it was shown that tree diagrams ( figs.[d_om],[d_eta ] ) dominate in these reactions . in the @xmath28
cross section the clear peak of the @xmath22 is seen and probably some contribution of the @xmath21 exists . in the @xmath29 reaction
the clear peak of @xmath30 determines the cross section shape but some contribution of @xmath13 is not excluded .
main diagram.,width=264 ] main diagram.,width=264 ] cross section.,width=302 ] cross section.,width=302 ] the main reaction channel @xmath32 is seen in the four pion final state but less systematic uncertainty in the cross section measurement was achieved by snd using the @xmath33 reaction @xcite . combining snd data with the data of dm2 @xcite and cleoii @xcite
the fit of the cross section was performed ( fig.[omegapi0 ] ) . note ,
that there is a systematic bias between the dm2 and cleoii data , which can be connected with a normalisation problem or with the bias in the energy scale .
cross section.,width=264 ] first indication of a radiative decay of radial excitations of light vector mesons was found out by cmd-2 @xcite .
two events of the reaction @xmath34 were identified .
the estimated production cross section is in agreement with the data of cmd-2 @xcite and dm2 @xcite for the reaction @xmath35 .
the preliminary snd results on the cross section @xmath37 @xcite together with the dm1 data @xcite can be successfully fitted if the contributions of the @xmath38 , @xmath19 , @xmath20 , and @xmath39 resonances are taken into account ( fig.[kskl ] ) .
the data on the reaction @xmath40 @xcite are in agreement with a such model .
cross section.,width=302 ] the pwa analysis of the @xmath41 reaction have been performed by dm2 @xcite .
it was shown that isoscalar process @xmath42 dominates .
the cross section @xmath43 is small .
the @xmath44 vector state observed in the hadron production @xcite is not confirmed in the @xmath0 production at vepp-2 m @xcite .
.the classification of vector mesons by pdg00 . [ cols="^,^,^,^",options="header " , ] [ t2 ] the classification ( table [ t1 ] ) of the light vector mesons proposed by pdg @xcite can not be accepted without a serious discussion .
some of resonances included in the table are no well established . in the contrary ,
the data on @xmath15 and @xmath21 are ignored .
the difficulty of the existing data analysis connect with the low statistical accuracy of the data above @xmath10 .
moreover the model uncertainty of resonances mass and width may exceed @xmath45 @xcite . the quality of the experimental data is demonstrated in the table [ t2 ] and the following conclusions can be done after review of the current data : * @xmath15 is ignored by pdg but as pointed out by d. peaslee @xcite there are several old and new experiments ( omega @xcite , lass @xcite , obelix @xcite ) in which some evidences for the @xmath15 were obtained .
* @xmath21 is identified by @xmath46 cross section @xcite .
* @xmath30 is identified by @xmath47 and @xmath29 production in @xmath0 and in @xmath48 experiments .
* @xmath49 has no solid ground .
* @xmath13 is seen in the @xmath0 production in @xmath50 , @xmath51 , and @xmath52 final states .
it is identified in the gamma production @xcite and in the @xmath48 production @xcite .
* @xmath22 is identified by @xmath28 cross section @xcite .
* @xmath39 is identified by the @xmath53 cross section @xcite .
* @xmath54 is identified in the hadron production of @xmath26 by gams @xcite .
there are a set of questions which must be answered to clear the situation with excited states of the light vector mesons : * do @xmath15 exist ?
what is the nature of this object ?
it is @xmath55 @xmath4 state or it is lowest 4-quark vector state ? * do @xmath21 is @xmath55 @xmath4 state or it is lowest 4-quark vector state ?
* do @xmath49 exist ? * @xmath13 , @xmath22 , and @xmath39 have practically the same mass .
they have to have common decay channels , so its real inputs are hidden in cross section shapes because of the interference .
are there three resonances @xmath13 , @xmath22 , and @xmath39 or there are only two ?
* do other light quarks states exist ?
the adequate @xmath0 collider for a such study is pep - n .
the experiment at pep - n is able to provide a good efficiency and particle identification for hadron and radiative transitions between different states in the energy region @xmath56 .
the other methods using existing facilities are not able to solve all problems of the spectroscopy of the light vector mesons because of following problems : * below @xmath57 the luminosity of existing @xmath0 colliders fall down .
the designed maximum energy of the vepp-2000 @xcite is @xmath58 . *
using the hadronic @xmath59 decays is not possible to establish the spectrum of vector mesons above @xmath60 because of kinematics . *
the initial state radiation method @xcite is a very effective method to demonstrate the cross section shape and to tag the most interesting phenomena , but the precision of this method is not known and some theoretical and experimental limits can be foreseen . * the previous experience shows us that experiments for the hadron production , @xmath61-production , and @xmath48 production can not substitute precise @xmath0 experiments for the vector meson spectroscopy .
* the knowledge of the vector meson spectroscopy is incomplete . *
the heavy quarkonium spectra are known much better than the spectrum of the light vector mesons . *
it is required to measure a complete set of hadron production cross sections in the energy region @xmath56 with the integrated luminosity about @xmath62 .
* this luminosity investment will provide an opportunity to study as traditional and exotic states , hadronic and radiative transitions .
* the two new @xmath0 projects vepp-2000 @xcite at novosibirsk and pep - n at slac are intend to bring a light on the light vector meson spectroscopy .
this two projects are complimentary in many aspects , so the realisation of both is very well required .
auslander et al . ,
. fiz . 9 : 114 - 119 , 1969 .
d.e . groom et al .
( particle data group ) , eur .
c15 : 1 , 2000 .
achasov et al . , phys .
b440 : 442 - 448 , 1998 .
e - print archive : hep - ex/9807016 .
cmd-2 collaboration ( r.r .
akhmetshin et al . ) , phys .
b462 : 380 , 1999 .
e - print archive : hep - ex/9907006 . m.n .
achasov et al . , phys .
b479 : 53 - 58 , 2000 .
e - print archive : hep - ex/0003031 . m.n .
achasov et al . , phys .
b485 : 349 - 356 , 2000 .
e - print archive : hep - ex/0005017 . a. zallo , talk at this workshop .
p.m. ivanov et al .
b107 : 297 , 1981 . l.m .
barkov et al . , nucl .
b256 : 365 - 384 , 1985 .
dolinsky et al . , phys .
rept . 202 : 99 - 170 , 1991 .
anashkin et .
al . , icfa instr .
bulletin 5 : 18 , 1988 .
achasov et al . , nucl .
instrum . meth .
a449 : 125 - 139 , 2000 .
e - print archive : hep - ex/9909015 .
achasov et al . , phys .
d63 : 072002 , 2001 .
e - print archive : hep - ex/0009036 . g.j .
gounaris and j.j .
sakurai , phys .
21 : 244 , 1968 .
m. bando et al . , phys . rev .
54 : 1215 , 1985 .
achasov et al . , sov .
54 : 664 - 671 , 1991 ; int . j. mod
a7 : 3187 - 3202 , 1992
. m. benayoun , s.i .
eidelman , v.n .
ivanchenko , z. phys .
c72 : 221 - 230 , 1996 .
achasov and a.a .
kozhevnikov , phys .
d62 : 117503 , 2000 .
cmd-2 collaboration ( r.r .
akhmetshin et al . ) .
budkerinp-99 - 10 , apr 1999 .
e - print archive : hep - ex/9904027 . d. bisello et al . , phys .
b220 : 321 , 1989 . m. benayoun et al . ,
j. c2 : 269 , 1998 .
a. antonelli et al .
, z. phys .
c56 : 15 - 19 , 1992 .
achasov et al . , phys .
b462 : 365 - 370 , 1999 .
e - print archive : hep - ex/9910001 .
cmd2 collaboration ( r.r .
akhmetshin et al . ) , phys .
b466 : 392 - 402 , 1999 .
e - print archive : hep - ex/9904024 .
cmd-2 collaboration ( r.r .
akhmetshin et al . ) , phys .
b489 : 125 - 130 , 2000 .
e - print archive : hep - ex/0009013 . m.n .
achasov et al . , phys .
b486 : 29 - 34 , 2000 .
e - print archive : hep - ex/0005032 . d. bisello et al . , nucl . phys .
21 : 111 , 1991 .
d. bisello et al . , z. phys .
c39 : 13 - 19 , 1988 .
d. bisello et al . , z. phys .
c52 : 227 - 230 , 1988 .
bityukov et al . , phys .
b188 : 383 , 1987 .
aulchenko et al . , jetp lett . 45 : 145 - 147 , 1987 ; pisma zh .
fiz . 45 : 118 - 120 , 1987 .
peaslee , private communication .
obelix collaboration ( a. bertin et al . ) , phys .
b408 : 476 , 1997 .
obelix collaboration ( a. bertin et al . ) , phys .
b414 : 220 - 228 , 1997 .
a. abele , phys .
b391 : 191 , 1997 .
a. abele , phys .
b468 : 178 , 1999 .
alde et al . , z. phys .
c66 : 379 , 1995 .
koop , talk at this workshop . | the experimental status of light vector meson spectroscopy is discussed .
the last results of @xmath0 experiments obtained at the vepp-2 m collider in novosibirsk are described and the comparison with the old data in the mass region from @xmath1 to @xmath2 is performed . |
it is well known that the practical application of superconductors crucially depends on the high electricity transport without dissipation , so the searching for new superconductors with a high transition temperature has attracted considerable attention in material science .
recently , the novel superconductivity in iron - based layered superconductors has been reported , providing a new path to the high temperature superconductivity .
the @xmath2 under doping with @xmath3 ions at the @xmath4 sites was first discovered to exhibit superconductivity with @xmath5 by kamihara et al @xcite .
then the holes were introduced in @xmath2 by partially substituting @xmath6 with @xmath7 , the superconductivity was observed with @xmath8 k @xcite . shortly after these studies , @xmath9 was pushed up surprisingly to above 40 k when @xmath6 in @xmath10 was substituted by other rare earth elements @xcite . in later experiments , the maximum @xmath9 at about 55
k was achieved in @xmath11 @xcite and @xmath12 @xcite .
though these superconductors have a high critical temperature , the mechanism and the possible difference compared with copper - based high-@xmath9 superconductors are still very controversial . obviously , more experimental and theoretical work are needed to understand the type - ii superconductivity as well as the microscopic mechanism in these new iron - based superconductors .
since the vortex - lattice solid ( glass ) state without linear resistivity is crucial for the application of high-@xmath9 superconductors , the melting of the flux - lattice in bulk type - ii superconductors is of fundamental and practical significance @xcite .
the elastic theory combined with the phenomenological lindemann criterion has been used to study the flux - lattice melting in layered superconductors@xcite , which is based on the instability of the crystal .
however , the regular flux - lattice structure loses simultaneously in the melting , the information for vortex liquid is then lacking , so the lindemann criterion has drawback in the study of the vortex melting .
the density functional theory has been successfully used in the flux - lattice melting of layered superconductors@xcite . due to the large anisotropy in these superconductors , weak josephson interaction between layers
can be neglected , both the interplane and intraplane interaction between the pancake vortices are electromagnetic in origin .
the density functional method is usually well appropriate for layered superconductors with large anisotropy .
recently , a rather large anisotropy @xmath13 @xcite was found in @xmath0 and @xmath1 ( 1111-type ) by a bond structure calculation .
the strong anisotropy in these systems was also confirmed recently in experiments@xcite . in this work , we study the flux - lattice melting in these two samples in the framework of the density functional theory and the mean - field substrate model proposed recently @xcite . in our first - principles calculations ,
the input contains just a few parameters . the two - dimensional ( 2d ) one - component plasma ( ocp ) pair distribution functions ( pdf ) , 2d ocp static structure factors ( ssf ) and 2d ocp direct correlation functions ( dcf ) are calculated in the celebrated hypernetted - chain ( hnc ) closure .
the validity of the hansen - verlet freezing criterion is analyzed .
the melting lines for @xmath0 and @xmath1 are plotted by using the mean - field substrate approach .
the paper is organized as follows .
section ii describes the calculation of 2d ocp pdf . in section iii
, we analyze the hansen - verlet freezing criterion and present the calculation of flux - lattice melting lines .
a short summary is given in the last section .
the electromagnetic vortex - vortex interaction is given by a three - dimensional ( 3d ) anisotropic pair potential . in fourier space
, it reads @xcite @xmath14},\ ] ] where @xmath15 and @xmath16 , @xmath17 and @xmath18 denote the component of @xmath19 perpendicular and parallel to the @xmath20-axis respectively , @xmath21^{1/2}$ ] is the planar penetration depth , @xmath22 is the flux quantum and @xmath23 is the layer spacing between feas layers . in our calculation , @xmath23 is twice lattice parameter @xmath20 , i. e. @xmath24 17.5 @xcite . according to eq .
( 1 ) , the intralayer potential is repulsive while the interlayer potential is attractive .
some parameters measured in a muon spin relaxation study @xcite are used in the present work , e. g. @xmath25 , @xmath26 k for @xmath0 and @xmath27 , @xmath28 k for @xmath1 . for convenience ,
the mean inter - particle spacing is taken as the length unit in the following . in the classical liquid theory ,
pdf is defined as @xcite @xmath29 where @xmath30 represents the probability to find a particle at @xmath31 , @xmath32 is the probability to find two particles at @xmath31 and @xmath33 respectively , and the pair correlation function @xmath34 is defined by @xmath35 . in the uniform and isotropic liquid , the ornstein - zernike relation can be written as @xcite @xmath36 while in the fourier space , it becomes @xmath37 ( a ) 2d ocp pdf ( b ) 2d ocp ssf and ( c ) 2d ocp dcf for @xmath10 ( x=0.1,0.075 ) at different temperatures . in all figures , from top to bottom at the first peaks : 10k for @xmath0 , 14k for @xmath0 , 10k for @xmath1 , 20k for @xmath0 , 14k for @xmath1 , and 20k for @xmath1.,width=283 ] the hnc closure is widely used in the vortex system of type - ii superconductors @xcite . in this closure , the bridge function
is set to zero and the dcf can be written as @xmath38-y(r)-1,\ ] ] where @xmath39 . following the method proposed in ref .
@xcite , we extract the correlation functions by splitting @xmath40 and @xmath41 into a short- and a long - ranged part as @xmath42 @xmath43 here @xmath44 , @xmath45 and @xmath46 , @xmath47 denotes the in - plane coordinate and @xmath48 is the layer index @xcite . in this way
, the hnc equation in the short range part can be written as @xmath49-y^{0}_{s}(\rho)-1.\ ] ] the in - plane pdf is then given by @xmath50 using an efficient algorithm proposed previously @xcite , we are able to calculate the short range part of 2d dcf and @xmath51 successfully and then obtain the in - plane pdf .
these functions are tabulated in a discrete form at about @xmath52 points .
since the pdf decreases rapidly in the order of the mean inter - particle spacing , they can be cut off in the real space to simplify the calculation .
the criterion for this cutoff is that the converged functions are not influenced . in our calculation , all of the converged functions are stable and the relative difference is less than @xmath53 .
actually , the value of pdf saturates at high field , consistent with the well - known fact that the vortex system decouples to a 2d system at high field .
these pdf can be regarded as the 2d ocp one , which is shown in fig .
1(a ) for @xmath10 ( [email protected] , 0.075 ) .
it displays an oscillating behavior and becomes very small at large @xmath47 .
dcf based on the mean - field substrate model for @xmath10 ( x=0.1,0.075 ) at t=14k , b=200gs.,width=264 ]
in the classical liquid theory , the ssf is defined as @xmath55 to provide a visual representation of the vortex structure , we calculate the 2d ocp ssf according to eq .
the 2d ocp dcf are then easily achieved with the relationship @xmath56 .
both 2d ocp ssf and 2d ocp dcf are shown in figs .
1(b ) and ( c ) .
in fact , the maximum values of these functions are found to decrease as temperature increases , shedding an insight of the 2d flux - lattice system in the melting process .
furthermore , all these functions for @xmath0 exhibit a larger oscillation than those for @xmath1 at the same temperatures , indicating that the 2d lattice structure in @xmath0 is more clear than that in @xmath1 . in the density functional approach first suggested by ramakrishnan and yussouff @xcite , the free energy function is given by @xmath57 \nonumber \\
& -&\frac{1 } { 2 } \int dr \int dr^ { ' } c(|\mathbf{r}-\mathbf{r}^{'}| ) \nonumber \\ & \times & [ \rho(\mathbf{r})-\overline{\rho}^{3d } ] [ \rho ( \mathbf{r}^{'})-\overline{\rho}^{3d}],\end{aligned}\ ] ] here @xmath58 is the density at @xmath59 , @xmath60 is the three - dimensional homogenous liquid density , and @xmath61 is the two - body fluid phase dcf . the two - body interactions between the particles are expressed in the second integral on the right hand side . flux - lattice melting lines for @xmath10 .
inset : flux - lattice melting line for @xmath62 calculated in ref .
, width=302 ] recently , a mean - field substrate model was proposed to investigate the thermodynamic properties in layered superconductors @xcite . here
we describe it briefly for completeness .
since the interlayer interaction between vortices is much weaker than the intralayer one in the strongly anisotropic superconductors , the vortices can be regarded as a collection of pancake vortices . in this way
, the interplane and out - of - plane interaction could be separated and the dcf can be written as @xcite @xmath63 where @xmath64 is the out - of - plane interaction potential . in fourier space , the effective dcf of the mean - field substrate model takes the following form @xmath65 we employ the above equation to extract the dcf based on the substrate model at different temperatures and magnetic fields .
for instance , we present the dcf at t=14k and b=200gs in fig .
2 . as proposed in ref .
@xcite , the single order parameter @xmath66 ( @xmath67 ) can be used to determine the flux - lattice melting lines , which is equivalent to the nontrivial solution for @xmath68 . on the other hand ,
the hansen - verlet freezing criterion is successfully used in the classical liquid theory @xcite .
this criterion states that the liquid freezes when the first peak of the structure factor reaches a critical value .
the density functional theory in the vortex system of type - ii superconductors is also based on the classical liquid theory , thus it is necessary to analyze the validity of the hansen - verlet freezing criterion in this issue . actually , the single order parameter @xmath66 is just relative to the 2d hansen - verlet freezing criterion through the equation @xmath56 . for the vortex system in layered superconductors with a large anisotropy ,
the liquid freezes when the structure factor at @xmath69 reaches the value @xmath70 .
interestingly , such a value is well consistent with the value of s(k ) at low - field regime along the melting line obtained in previous work @xcite .
finally , we employ the freezing criterion described above to locate the melting lines .
critical fields for @xmath10 @xmath71 at different temperatures are calculated , the melting lines are shown in fig .
( 3 ) . for comparison , the melting line for @xmath62 obtained in ref .
@xcite is also given in the inset .
one can observe that the monotonic meting lines for @xmath2 are qualitatively alike but much lower than that for @xmath62 .
recently , a quite large vortex liquid area in @xmath0 compounds is observed in an experimental study based on the measurement of nernst signal @xcite , consistent with the melting lines predicted in the present work .
interestingly , we find that in @xmath2 samples the melting line is changed substantially with the variation of @xmath3 content , the melting line for @xmath0 is much higher than that for @xmath1 .
the very recently discovered ternary iron - arsenides @xmath72 ( 122-type)@xcite were suggested to be nearly isotropic@xcite . the present method to locate the melting lines of course can not be applicable to these superconductors with low anisotropy .
the methodological difference between the present work and the previous study in ref .
@xcite lies in the determination of 2d ocp direct correlation functions . in ref .
@xcite these functions were obtained from monte - carlo simulations , while they are achieved @xmath73 @xmath74 from the microscopic vortex interaction in the present study .
therefore , the present approach is more intuitive , and the results are obtained without statistical errors like in monte - carlo simulations .
the method adopted in the present work is also different from that in ref .
@xcite where the density functional theory was first extended to the flux - lattice melting . a consistency to the clausius - clapeyron relation @xcite
can be reached with the use of the mean - field substrate model in the present work , but can not be obtained in ref .
@xcite .
we use the density functional theory and the mean - field substrate model to investigate the flux - lattice melting in iron - based superconductors @xmath75 .
the 2d ocp pdf , 2d ocp ssf and 2d ocp dcf for @xmath0 and @xmath1 are calculated at different temperatures via the hnc closure and an efficient algorithm .
it is shown that the melting lines are quite low , well consistent with an experimental study on the nernst signals @xcite .
furthermore , the melting line is observed to change considerably with the modulation of @xmath3 content in @xmath76-doped @xmath2 compounds , indicating that the interaction between vortices crucially depends on the @xmath3 content .
more experimental studies are needed to test the present theoretical predications .
this work was supported by national natural science foundation of china under grant nos .
10574107 and 10774128 , pcsirt ( grant no .
irt0754 ) in university in china , national basic research program of china ( grant nos . 2006cb601003 and 2009cb929104 ) .
s. segupta , c. dasgupta , h. r. krishnamurthy , g. i. menon and t. v. ramakrishman , phys .
lett . * 67 * , 3444 ( 1991 ) ; g.i.menon , c. dasgupta , h. r. krishnamurthy , t. v. ramakrishman and s. segupta , phys .
b * 54 * , 16192 ( 1996 ) . a. houghton , r. a. pelcovits and a. sudbo , phys .
b * 40 * , 6763 ( 1989 ) ; x. hu and q. h. chen , phys .
lett . * 92 * , 209701 ( 2004 ) ; q. m. nie , m. b. luo , q. h. chen and x. hu , europhys .
lett . * 71 * , 445 ( 2005 ) . | we report the theoretical study of the flux - lattice melting in the novel iron - based superconductor @xmath0 and @xmath1 . using the hypernetted - chain closure and an efficient algorithm
, we calculate the two - dimensional one - component plasma pair distribution functions , static structure factors and direct correlation functions at various temperatures .
the hansen - verlet freezing criterion is shown to be valid for vortex - liquid freezing in type - ii superconductors .
flux - lattice meting lines for @xmath0 and @xmath1 are predicted through the combination of the density functional theory and the mean - field substrate approach . |
rapid periodic oscillations are observed in the optical flux of high accretion rate ( `` high-@xmath8 '' ) cataclysmic variables ( cvs ) ( nova - like variables and dwarf novae in outburst ) @xcite .
these oscillations have high coherence ( @xmath9@xmath10 ) , short periods ( @xmath1140 s ) , low amplitudes ( @xmath12% ) , and are sinusoidal to within the limits of measurement .
they are referred to as `` dwarf nova oscillations '' ( dnos ) to distinguish them from the longer period , low coherence ( @xmath1310 ) quasi - periodic oscillations ( qpos ) of high-@xmath8 cvs , and the longer period , high coherence ( @xmath14@xmath15 ) oscillations of dq her stars .
dnos appear on the rising branch of the light curve of dwarf nova outbursts , typically persist through maximum , and disappear on the declining branch of the light curve .
the period of the oscillation decreases on the rising branch and increases on the declining branch , but because the period reaches minimum about one day after maximum optical flux , dwarf novae describe loops in plots of oscillation period versus optical flux .
rapid periodic oscillations have been detected in the flux of numerous high-@xmath8 cvs , but the dwarf nova ss cyg in outburst is one of the best studied .
oscillations in the optical flux have been detected with periods ranging from 7.3 s to 11 s @xcite , and oscillations in the extreme - ultraviolet ( euv ) and soft x - ray flux have been detected with periods ranging from 2.8 s to 11 s in data from _ heao 1 _ , _ exosat _ , the _ extreme ultraviolet explorer _ ( _ euve _ ) , and _ rosat _ @xcite .
@xcite showed that the euv oscillation period is a single - valued function of the euv flux , explained the loops observed in plots of oscillation period versus optical flux as the result of the delay between the rise of the optical and euv flux at the beginning of dwarf nova outbursts
@xcite , and so provided strong evidence for the proposal , first mooted by @xcite , that the oscillation period depends solely on the mass - accretion rate onto the white dwarf .
@xcite observed `` frequency doubling '' of the euv oscillation of ss cyg during the rising branch of its 1996 october outburst , demonstrated that the optical and euv oscillation periods are equal , and that their relative phase delay is consistent with zero . in this communication , we present an analysis of observations of ss cyg in outburst obtained in 1994 june / july with the _ euve _ deep survey ( ds ) photometer and in 2001 january with the _ chandra x - ray observatory _ low energy transmission grating spectrograph ( letgs ) . in
2 we present the observations and data analysis , finding that the euv flux was oscillating at frequencies @xmath0 hz and @xmath16 hz , while the soft x - ray flux was oscillating at frequencies @xmath2 hz , @xmath3 hz , and possibly @xmath17 hz . in 3 we discuss these results in the context of similar results from optical observations of the dwarf nova vw hyi in outburst and x - ray observations of neutron star and black hole low - mass x - ray binaries ( lmxbs ) .
as first pointed out by @xcite , white dwarf , neutron star , and black hole binaries share a common correlation in the frequencies of their qpos , extending over nearly five orders of magnitude in frequency .
the implications of this result for qpo models are discussed , we discuss how additional observations of cvs can provide unique and quantitative tests of qpo models , and close in 4 with a summary .
our pre - approved target - of - opportunity _ chandra _ low energy transmission grating ( letg ) and high resolution camera ( hrc ) observation of ss cyg was performed between 2001 january 16 @xmath18 and january 17 @xmath19 ut during the plateau phase of a wide normal ( asymmetric ) dwarf nova outburst that began on january 12 .
when the observation began , ss cyg had been at maximum optical light ( @xmath20 ) for approximately 3 days ; long enough , given the delay of 1.5 days between the rise of the optical and euv light curves of ss cyg @xcite , for the euv / soft x - ray flux to have reached maximum and for the system to have reached a quasi - steady - state .
indeed , the zero- and @xmath21 first - order letg / hrc count rates were observed to be roughly constant throughout the 47.3 ks observation at 1.5 and 4.9 @xmath22 , respectively .
the files used for this analysis were created on 2002 january 7 by the pipeline data reduction software using ciao 2.0 with asdcs version 6.5.1 .
events were extracted from the level 2 evt file , and the mean spectrum and source and background region masks were extracted from the level 2 pha file .
the mean letg spectrum contains two distinct components : ( 1 ) a hard component shortward of 40 consisting of a bremsstrahlung continuum and emission lines of h- and he - like c , n , o , ne , mg , and si and l - shell fe ( most prominently ) and ( 2 ) a soft component extending from 42 to 120 that appears to consist of a continuum significantly modified by a forest of emission and absorption features ( possibly even p cygni profiles ) . in the 72120 wave band , the letg spectrum is essentially identical to that measured by the _ euve
_ short wavelength ( sw ) spectrometer during the 1993 august @xcite and 1996 october @xcite outbursts of ss cyg .
these hard and soft components of the x - ray spectrum of ss cyg are understood to be due to the optically thin and optically thick portions of the boundary layer between the disk and the surface of the white dwarf . to investigate the temporal properties of the hard and soft x - ray flux of ss cyg , we processed the @xmath23 first - order letg events as follows .
first , we created a list of source and background events by filtering the events with the source and background region masks .
second , we applied a wavelength filter that excluded events near the hrc chip gaps : for the positive order , events with wavelengths @xmath2467.5 were excluded , while for the negative order , events with wavelengths @xmath2557.2 were excluded . without this filter , a portion of the soft component of the spectrum moves on and off the hrc chips , and so produces signals in the power spectra at the spacecraft dither periods of 707 s and 1000 s. third , various wavelength cuts were applied to isolate the hard ( @xmath2642 ) component of the spectrum , the soft ( @xmath27120 ) component of the spectrum , and various subsets thereof ( see below ) .
events passing these filters were used to create background - subtracted light curves with 1 s time resolution , and power spectra of these light curves were examined for evidence of periodic flux modulations .
consistent with more sensitive searches by _ heao-1 _
@xcite , _ asca _
@xcite , _ ginga _
@xcite , and the _ rossi x - ray timing explorer _ ( _ rxte _ ) @xcite , no periodic signal was detected in the hard x - ray light curves .
in contrast , strong periodic signals were detected in the soft x - ray light curves : as shown in the upper panel of figure 1 , the power spectrum of the soft x - ray light curve shows excess power at frequencies @xmath28 hz and @xmath3 hz , indicating the presence of oscillations at periods @xmath29 s and @xmath30 s. to investigate these oscillations more closely , we divided the soft x - ray light curve into 47 consecutive 1 ks intervals and calculated their power spectra .
although the individual power spectra are rather noisy , the @xmath31 hz oscillation typically appears as a single @xmath32 hz peak with a frequency in the range @xmath330.112 hz .
the mean of these 47 consecutive 1 ks power spectra is shown in the middle panel of figure 1 .
in addition to the peaks at @xmath2 hz and @xmath3 hz , there appears to be a shoulder on the higher - frequency peak extending to @xmath34 hz . to account for the observed frequency evolution , we scaled the frequency vectors of the 47 power spectra by the instantaneous frequency of the dominant @xmath35 oscillation .
the mean of these scaled power spectra is shown in the lower panel of figure 1 .
note that most of the power of the @xmath35 oscillation now lies in a single frequency bin , there is no excess power in the first harmonic @xmath36 or subharmonic @xmath37 ( i.e. , the dominant @xmath35 oscillation is sinusoidal to high degree ) , and the lower - frequency peak lies at @xmath38 .
this is not the first time that multiple periodicities have been seen in the short wavelength flux of ss cyg . in the power spectrum of the _ euve _ ds count rate light curve of the 1994 june / july outburst of ss cyg , @xcite noted the presence of oscillations at @xmath39 hz , @xmath1 hz , and its first harmonic @xmath36 .
furthermore , he noted that the harmonic peak was present in data from only the first part of the observation , and suggested that the low - frequency peak is due to the spin of the white dwarf . to revisit these issues
, we reexamined the ds data from the second part of the _ euve _ observation ( between 1994 june 29 @xmath40 and july 3 @xmath41 ut ) .
background - subtracted light curves were derived from 37 15 ks orbits , 37 power spectra were calculated , and the mean power spectrum derived . as shown in the upper panel of figure 2 ,
peaks in this power spectrum occur at frequencies @xmath0 hz and @xmath42 hz , indicating the presence of oscillations in the euv flux of ss cyg at periods @xmath43 s and @xmath44 s. as before , we accounted for the observed frequency evolution from @xmath45 hz to 0.132 hz @xcite by scaling the frequency vectors of the 37 power spectra by the instantaneous frequency of the dominant @xmath35 oscillation .
the mean of these scaled power spectra is shown in the lower panel of figure 2 .
note that most of the power of the @xmath35 oscillation again lies in a single frequency bin , there is no excess power in the first harmonic @xmath46 or subharmonic @xmath37 ( i.e. , the dominant @xmath35 oscillation is sinusoidal to high degree ) , and the lower - frequency peak lies at @xmath47 . before leaving this section
, we note that it is possible to use the _ chandra _ letg data to investigate the energy dependence of the euv / soft x - ray oscillations of ss cyg . while an - by - investigation of the spectrum of the oscillations is beyond the scope of this paper , it is a simple matter to divide the letg soft x - ray bandpass in two .
a division at @xmath48 would halve the counts from the soft component of the x - ray spectrum of ss cyg , but instead we divided the counts at @xmath49 to approximate the bandpass of the _ euve _ ds .
as before , we constructed background - subtracted light curves with 1 s time resolution , divided the light curves into 47 consecutive 1 ks intervals , calculated power spectra , and derived mean power spectra .
the resulting mean power spectra for the @xmath5070 and @xmath49120 bandpasses are shown respectively in the upper and lower panels of figure 3 .
this figure shows that in the softer @xmath49120 bandpass the @xmath51 hz and @xmath3 hz oscillations are slightly stronger , while in the harder @xmath2770 bandpass the shoulder on the @xmath3 hz oscillation appears to resolve into a distinct peak at @xmath52 hz .
summarizing the results from the previous section , we have found that during the 1994 june / july _ euve _ observation of ss cyg in outburst the euv flux was oscillating at frequencies @xmath0 hz and @xmath1 hz , while during the 2001 january _ chandra _ letg observation the soft x - ray flux was oscillating at frequencies @xmath2 hz , @xmath3 hz , and possibly @xmath53 hz .
we note that because the frequency of the @xmath54 oscillation is not constant from 1994 to 2001 , it can not be due to the spin of the white dwarf , as suggested by @xcite .
instead , the near - constancy of the ratio @xmath55 suggests that the @xmath54 oscillation is related to the dominant @xmath35 oscillation . to place these results in a broader context
, we note that other high-@xmath8 cvs have been observed to display multiple periodicities in their power spectra .
first , @xcite list a number of instances in the literature when pairs of dnos have been detected in the optical flux of nova - like variables and dwarf novae in outburst .
second , @xcite detected a pair of dnos at @xmath56 hz and @xmath57 hz in the optical continuum and balmer emission line flux of v2051 oph during its 1998 july outburst .
although it was not possible to detect the difference frequency @xmath58 hz directly in the power spectrum ( d. steeghs 2002 , personal communication ) , the amplitude of the oscillation varied considerably on a timescale equal to the inverse of the difference frequency ( 8 min ) . third , @xcite discuss a number of instances when multiple periodicities have been detected in the optical flux of vw hyi in outburst . during the decline of the 2000 february outburst , dnos with periods @xmath5937 s and qpos with periods @xmath60580 s were detected simultaneously . the ratio @xmath610.071 , which is similar to the ratio observed in the euv and soft x - ray oscillations of ss cyg .
furthermore , in data from the 1972 november outburst of vw hyi , three oscillations were detected simultaneously : a pair of dnos with periods @xmath62 s and @xmath63 s , and a qpo with a period @xmath64 s. the ratio @xmath65 and the difference frequency @xmath66 hz , which is close , but not equal , to the qpo frequency @xmath67 hz .
having established these properties of the qpos of cvs , it is interesting to investigate their possible connection to the qpos of lmxbs .
@xcite provides a comprehensive review of lmxb qpos , so it is sufficient to note here that among neutron star binaries , the luminous z sources have pairs of 2001200 hz `` kilohertz qpos , ''
1560 hz `` horizontal branch oscillations '' ( hbos ) , and 520 hz `` normal branch oscillations , '' while the less luminous atoll sources have 5001250 hz kilohertz qpos , as well as 2060 hz qpos and broad noise components with properties that are similar to hbos .
* hereafter pbk99 ) showed that in five z sources a tight correlation exists between the hbo frequency @xmath68 and the frequency @xmath69 of the lower - frequency member of the pair of khz qpos ( the `` lower khz qpo '' ) .
specifically , when @xmath70 hz , @xmath71 furthermore , by identifying with @xmath68 and @xmath72 the frequencies of various types of peaked noise components in atoll sources , other neutron star binaries , and black hole binaries , pbk99 and subsequently ( * ? ? ?
* hereafter bpk02 ) extended this correlation over nearly three orders of magnitude in frequency .
as noted by @xcite , the optical data for vw hyi in outburst lie on an extrapolation of this correlation , extending it an additional two orders of magnitude in frequency .
figure 4 shows that our euv and soft x - ray data for ss cyg in outburst also lie on an extrapolation of this correlation , further establishing the connection between the cv and lmxb qpos .
this connection identifies the dnos of cvs with the khz qpos of lmxbs , and the qpos of cvs with the hbos ( or the broad noise component identified as such ) of lmxbs .
we note that the frequencies of the dnos of cvs and the khz qpos of neutron star binaries are similar in that they are comparable to the keplerian frequency @xmath73 at the inner edge of the accretion disk of , respectively , a white dwarf and neutron star : @xmath74 hz for a @xmath75 white dwarf with @xmath76 cm , while @xmath77 hz for a @xmath78 neutron star with @xmath79 km , as required by general relativity .
in addition to their frequencies , the dnos of cvs and the khz qpos of neutron star binaries are similar in that they have relatively high coherence and high amplitudes , their frequency scales with the inferred mass - accretion rate , and they sometimes occur in pairs . given the apparent connection between the qpos of white dwarf , neutron star , and black hole binaries , it is appropriate that we investigate the implications for the theories of qpo formation .
in the beat - frequency models , qpos occur at the keplerian frequency @xmath80 at a special radius in the accretion disk , and the beat of this frequency with the stellar rotation frequency @xmath81 . in the magnetospheric beat - frequency model @xcite , the hbo frequency
is identified with the beat frequency between the keplerian frequency at the magnetospheric radius @xmath82 and the stellar rotation frequency : @xmath83 . in the sonic - point beat - frequency model @xcite
, some of the disk plasma makes its way past the magnetospheric radius to the `` sonic point '' radius @xmath84 where the disk is effectively terminated because of either radiation drag or general relativistic corrections to newtonian gravity . in this model , the upper khz qpo frequency is identified with the keplerian frequency at the sonic point radius : @xmath85 , and the lower khz qpo frequency is identified with ( one or two times ) the beat frequency between the keplerian frequency at the sonic point radius and the stellar rotation frequency : @xmath86 $ ] , where @xmath87 or
2 . note that in this model , the stellar rotation frequency @xmath88 . in the relativistic precession model @xcite , the qpo signals are due to the fundamental frequencies of disk plasma orbiting a rapidly rotating compact star in slightly eccentric and tilted orbits . in this model , the upper khz qpo frequency is identified with the keplerian frequency : @xmath89 , the lower khz qpo frequency is identified with the periastron precession frequency : @xmath90 , and the hbo frequency is identified with the nodal precession frequency : @xmath91 . how does the existence of qpos in white dwarf , neutron star , and black hole binaries constrain the theories of qpo formation ?
first , as has been pointed out by other authors , the existence of qpos in both neutron star and black hole binaries excludes both flavors of the beat - frequency models ( as well as _
any _ model requiring the presence or absence of a stellar surface or magnetic field ) .
second , the dnos of cvs exclude the sonic - point beat - frequency model .
first , there is no reason to expect a sonic point in the inner disk of a cv : @xmath92 and radiation drag is unimportant because the luminosity is a small fraction of the eddington rate ( @xmath93 ) and because the flow is only mildly relativistic ( @xmath94 ^{1/2}{{\lower0.75ex\hbox { $ < $ } \atop\raise0.5ex\hbox { $ \sim$ } } } 0.02\ , c$ ] ) .
second , in vw hyi at least , the dno frequency separation is not equal to one or two times the white dwarf spin frequency . using the _ hubble space telescope _ ( _ hst _ ) goddard high resolution spectrograph , @xcite measured the projected rotation velocity of the white dwarf in vw hyi in quiescence to be @xmath95 . with a binary inclination @xmath96 and a white dwarf mass @xmath97 ( hence @xmath98 ) , @xmath99 hz , whereas the dno separation frequency @xmath100 hz .
third , the qpos and dnos of cvs exclude the relativistic precession model .
first consider the nodal precession frequency , which is the sum of the relativistic ( lense - thirring ) precession frequency @xmath101 due to frame dragging around a rapidly rotating compact star and the classical precession frequency @xmath102 due to the quadrupole term in the gravitational potential of an oblate star . because @xmath102 is negative for prograde orbits , @xmath103 , where @xmath104 is the moment of inertia of the star . for a @xmath105 white dwarf , @xmath74 hz and @xmath106 , and since stability requires @xmath107 @xmath108 hz , whereas the observed cv qpo frequencies @xmath109 hz . next consider the periastron precession frequency , which is the difference between the keplerian and epicyclic frequencies .
the latter term is @xmath110 to sufficient accuracy for a white dwarf accretor , so @xmath111{{\lower0.75ex\hbox { $ < $ } \atop\raise0.5ex\hbox { $ \sim$ } } } 10^{-3}$ ] hz , whereas the observed cv dno frequencies @xmath112 hz . in summary , assuming that the same mechanisms produce the qpos in white dwarf , neutron star , and black hole binaries , we find that the data exclude the relativistic precession model and the magnetospheric and sonic - point beat - frequency models ( as well as _
any _ model requiring the presence or absence of a stellar surface or magnetic field ) .
more promising are models , such as the transition layer model of titarchuk and collaborators @xcite , that interpret qpos as manifestations of disk accretion onto any low - magnetic field compact object .
the results from the previous sections suggest that there is a close relationship between the qpos of cvs and lmxbs , with the dnos of cvs being the analogues of the khz qpos of lmxbs and the qpos of cvs being the analogues of the hbos ( or the broad noise component identified as such ) of lmxbs .
the proposed equivalence of cv dnos and lmxb khz qpos strengthens the commonly - held belief that the frequency of the khz qpos is equal to the keplerian frequency at the inner edge of the accretion disk because ( 1 ) the frequency ratio @xmath113 and ( 2 ) in every case the dno frequencies are consistent with the requirement that @xmath114 @xcite .
the proposed equivalence of cv dnos and lmxb khz qpos on one hand and cv qpos and lmxb hbos on the other hand can be strengthened by finding more cvs with pairs of dnos and more cvs in which dnos and qpos are observed simultaneously .
conversely , systems known to show only qpos could be searched more carefully for dnos .
a interesting candidate is the dwarf nova u gem in outburst , whose @xmath115 hz qpo @xcite implies the existence of @xmath116 hz ( 2 s ) dnos ( an oscillation period shorter than the integration times typically employed to search for optical oscillations ) . for the keplerian frequency to be this high
, the mass of the white dwarf in u gem must be quite high : @xmath117 , assuming the @xcite white dwarf mass - radius relation .
this requirement is consistent with the white dwarf mass derived by @xcite ( @xmath118 ) , but is inconsistent with the values derived by @xcite ( @xmath119 ) and @xcite ( @xmath120 ) . aside from filling in the lower - left corner of figure 4 ,
additional observations of cvs are warranted because they provide unique and quantitative tests of qpo models .
data can be obtained from the ground in the optical and from space in the ultraviolet , euv , and soft x - rays ; the system parameters ( binary inclination , white dwarf mass , radius , and rotation velocity ) can be measured ; eclipse mapping in edge - on systems allows the sites of flux modulations to be located and dissected ; dwarf nova outbursts provide a dramatic and systematic variation in the mass - accretion rate ; diagnostic emission lines are available from the optical through soft x - rays , and general relativistic affects are minimal .
furthermore , technological improvements now allow _ spectroscopic _ observations of the flux oscillations of cvs .
@xcite used the keck low resolution imaging spectrograph ( @xmath1219200 ) in continuous readout mode to study the dnos in the optical continuum and balmer line flux of the eclipsing dwarf nova v2051 oph in outburst .
@xcite used the _ hst _ faint object spectrograph with the g160l ( @xmath1222510 ) grating to study the wavelength dependence of the ultraviolet dnos of the eclipsing dwarf nova oy car in outburst . @xcite used the _ hst _ space telescope imaging spectrograph in time - tag mode with the e140h ( @xmath1231690 ) echelle grating to study the ultraviolet dnos of ss cyg in outburst , while @xcite used the _ euve _ sw spectrometer to study the wavelength dependence of the euv dnos of ss cyg in outburst .
such observations shed new light on the qpos of cvs and lmxbs .
we have shown that during the 1994 june / july _ euve _ observation of ss cyg in outburst the euv flux was oscillating at frequencies @xmath124 hz and @xmath1 hz , while during the 2001 january _ chandra _ letg observation the soft x - ray flux was oscillating at frequencies @xmath2 hz , @xmath3 hz , and possibly @xmath125 hz .
these data , combined with the optical data of @xcite for vw hyi , extend the pbk99 and bpk02 @xmath5@xmath126 correlation for neutron star and black hole lmxbs nearly two orders of magnitude in frequency , with @xmath127 .
this correlation identifies the dnos of cvs with the khz qpos of lmxbs , and the qpos of cvs with the hbos ( or the broad noise component identified as such ) of lmxbs . assuming that the same mechanisms produce the qpos in white dwarf , neutron star , and black hole binaries
, we find that the data exclude the relativistic precession model and the magnetospheric and sonic - point beat - frequency models ( as well as _
any _ model requiring the presence or absence of a stellar surface or magnetic field ) .
additional observations of cvs can help establish the proposed equivalence of cv and lmxb qpos , and will provide unique and quantitative tests of qpo models .
our satellite observations of ss cyg were made possible by the optical monitoring and alerts provided by the members , staff ( particularly e.waagen ) , and director , j. mattei , of the american association of variable star observers .
the _ euve _
observation was made possible by the efforts of _ euve _ project scientist r. malina , science planner m. eckert , the staff of the _ euve _ science operations center at the center for euv astrophysics , and the flight operations team at goddard space flight center .
observation was made possible by the efforts of _ chandra _ x - ray observatory center director h. tananbaum , mission planner k. delain , and the _ chandra _ flight operations team at mit .
we acknowledge t. belloni for kindly supplying us with the neutron star and black hole binary data shown in figure 4 ; l. titarchuk , k. wood , and d. steeghs for helpful discussions ; and the anonymous referee for comments which improved the clarity of the manuscript . support for this work was provided by nasa through _ chandra _ award number go1 - 2023a issued by the _
chandra _ x - ray observatory center , which is operated by the smithsonian astrophysical observatory for and on behalf of nasa under contract nas8 - 39073 .
this work was performed under the auspices of the u.s.department of energy by university of california lawrence livermore national laboratory under contract no . w-7405-eng-48 .
alpar , m. a. , & shaham , j. 1985 , nature , 316 , 239 belloni , t. , psaltis , d. , & van der klis , m. 2002 , , 572 , 392 ( bpk02 ) crdova , f. a. , chester , t. j. , mason , k. o. , kahn , s. m. , & garmire , g. p. 1984
, , 278 , 739 crdova , f. a. , chester , t. j. , tuohy , i. r. , & garmire , g. p. 1980
, , 235 , 163 friend , m. t. , martin , j. s. , smith , r. c. , & jones , d. h. p. 1990
, , 246 , 637 hildebrand , r. h. , spillar , e. j. , & stiening , r. f. 1981 , , 243 , 223 horne , k. , & gomer , r. 1980 , , 237 , 845 jones , m. h. , & watson , m. g. 1992 , , 257 , 633 knigge , c. , drake , n. , long , k. s. , wade , r. a. , horne , k. , & baptista , r. 1998 , , 499 , 429 lamb , f. k. , shibazaki , n. , alpar , m. a. , & shaham , j. 1985 , nature , 317 , 681 long , k. s. , & gilliland , r. l. 1999 , , 511 , 916 long , k. s. , mauche , c. w. , raymond , j. c. , szkody , p. , & mattei , j. a. 1996 , , 469 , 841 marsh , t. r. , & horne , k. 1998 , , 299 , 921 mason , k. o. , crdova , f. a. , watson , m g. , & king , a. r. 1988 , , 232 , 779 mauche , c. w. 1996 , , 463 , l87 mauche , c. w. 1997a , , 476 , l85 mauche , c. w. 1997b , in accretion phenomena and related outflows , ed .
d. t. wickramasinghe , l. ferrario , & g. v. bicknell ( san francisco : asp ) , 251 mauche , c. w. 2002a , in continuing the challenge of euv astronomy : current analysis and prospects for the future , ed .
s. howell , j. dupuis , d. golombek , & j. cullison ( san francisco : asp ) , 75 mauche , c. w. 2002b , baps , 47 , # 2 , 151 mauche , c. w. , mattei , j. a. , & bateson , f. m. 2001 , in evolution of binary and multiple stars , ed .
podsiadlowski , et al .
( san francisco : asp ) , 367 mauche , c. w. , raymond , j. c. , & mattei , j. a. 1995 , , 446 , 842 mauche , c. w. , & robinson , e. l. 2001 , , 562 , 508 miller , m. c. , lamb , f. k. , & psaltis , d. 1998 , , 508 , 791 nauenberg , m. 1972 , , 175 , 417 nousek , j. a. , et al .
1994 , , 436 , l19 patterson , j. 1981 , , 45 , 517 patterson , j. , robinson , e. l. , & kiplinger , a. l. 1978 , , 226 , l137 ponman , t. j. , et al .
1995 , , 276 , 495 psaltis , d. , belloni , t. , & van der klis , m. 1999 , , 520 , 262 ( pbk99 ) sion , e. m. , huang , m. , szkody , p. , & cheng , f .- h . 1995 , , 445 , l31 smak , j. i. 2001 , acta astro . , 51 , 279 steeghs , d. , obrien , k. , horne , k. , gomer , r. , & oke , j. b. 2001 , , 323 , 484 stella , l. , & vietri , m. 1998 , , 492 , l59 stella , l. , & vietri , m. 1999 , , 82 , 17 swank , j. h. 1979 , in white dwarfs and variable degenerate stars , ed .
h. m. van horn & v. weidemann ( rochester : univ . of rochester ) , 135 titarchuk , l. , lapidus , i. , & muslimov , a. 1998 , , 499 , 315 titarchuk , l. , osherovich , v. , & kuznetsov , s. 1999 , , 525 , l129 titarchuk , l. , & wood , k. 2002 , , submitted van der klis , m. 2000 , , 38 , 717 van teeseling , a. 1997 , , 324 , l73 warner , b. 1995a , cataclysmic variable stars ( cambridge : cup ) warner , b. 1995b , in cape workshop on magnetic cataclysmic variables , ed .
d. a. h. buckley & b. warner ( san francisco : asp ) , 343 warner , b. , & woudt , p. a. 2002a , in the physics of cataclysmic variables and related object , ed .
b. t. gnsicke , k. beuermann , & k. reinsch ( san francisco : asp ) , 406 warner , b. , & woudt , p. a. 2002b , , in press ( astro - ph/0204484 ) wheatley , p. j. , mauche , c. w. , & mattei , j. a. 2002 , , submitted woudt , p. a. , & warner , b. 2002 , , in press ( astro - ph/0202441 ) | using data obtained in 1994 june / july with the _ extreme ultraviolet explorer _ deep survey photometer and in 2001 january with the _ chandra x - ray observatory _ low energy transmission grating spectrograph , we investigate the extreme - ultraviolet ( euv ) and soft x - ray oscillations of the dwarf nova ss cyg in outburst .
we find quasi - periodic oscillations ( qpos ) at @xmath0 hz and @xmath1 hz in the euv flux and at @xmath2 hz , @xmath3 hz , and possibly @xmath4 hz in the soft x - ray flux .
these data , combined with the optical data of woudt & warner for vw hyi , extend the psaltis , belloni , & van der klis @xmath5@xmath6 correlation for neutron star and black hole low - mass x - ray binaries ( lmxbs ) nearly two orders of magnitude in frequency , with @xmath7 .
this correlation identifies the high - frequency quasi - coherent oscillations ( so - called `` dwarf nova oscillations '' ) of cataclysmic variables ( cvs ) with the kilohertz qpos of lmxbs , and the low - frequency qpos of cvs with the horizontal branch oscillations ( or the broad noise component identified as such ) of lmxbs . assuming that the same mechanisms produce the qpos in white dwarf , neutron star , and black hole binaries
, we find that the data exclude the relativistic precession model and the magnetospheric and sonic - point beat - frequency models ( as well as _
any _ model requiring the presence or absence of a stellar surface or magnetic field ) ; more promising are models that interpret qpos as manifestations of disk accretion onto any low - magnetic field compact object . |
it is known that third generation squark loops can introduce sizeable cp violation in the higgs potential of the minimal supersymmetric standard model ( mssm ) , if the soft - supersymmetry - breaking mass parameters of the third generation are complex ; see , for example , @xcite . as a result
, the neutral higgs bosons will mix to produce three physical mass eigenstates with mixed cp parity , which we denote @xmath12 and @xmath5 in order of increasing mass .
a benchmark scenario of maximal cp violation , called cpx , was introduced in ref .
@xcite . in this scenario |a_t|=|a_b|=2
m_susy,||=4 m_susy , m__3,_3,_3=m_susy,|m_3|=1 tev , [ eq : jan5a ] where @xmath13 are are the soft - supersymmetry - breaking trilinear parameters of the third generation squarks and @xmath14 is the supersymmetric higgsino mass parameter .
the phenomenological consequences of this model may be quite spectacular .
in particular , the @xmath15 coupling of the lightest higgs boson can be significantly suppressed ; see , for example , @xcite and references therein . in this case
, it was shown that the lep2 data do not exclude the existence of a light higgs boson with mass @xmath16 gev ( 40 gev ) in the minimal susy model with @xmath174 ( 23 ) and cp - violating phase _
cpx ( a_t ) = arg(a_b ) = arg(a_)=arg ( m_g ) = 90^ ( 60^ ) .
[ eq : a1 ] since the @xmath2 couplings to the @xmath18 and @xmath19 gauge bosons become rather small , it would be hard to detect the light higgs via the processes @xmath20 or @xmath21 .
it is therefore interesting to consider the possibility of observing a light higgs boson at the lhc or tevatron collider .
however , in general , it will be hard to observe a light higgs at hadron colliders via the @xmath22 decay mode because , in particular , the transverse momenta of the outgoing @xmath3 and @xmath23 jets are not large . as a consequence
the signal is swamped by the qcd @xmath22 background .
therefore it was proposed @xcite to search for a cp - violating light higgs boson in the _ exclusive _ process @xmath24 at hadron colliders , where the @xmath25 signs denote the presence of large rapidity gaps . over the past few years
such exclusive diffractive processes have been considered as a promising way to search for manifestations of new physics in high energy proton - proton collisions ; see , for instance , @xcite .
these processes have both unique experimental and theoretical advantages in hunting for higgs bosons as compared to the traditional non - diffractive approaches . in particular , in the exclusive diffractive reactions the @xmath22 background is suppressed @xcite , and it may be feasible to isolate the signal . in the present paper we discuss the central _ exclusive _ diffractive production ( cedp ) in more detail .
we compare the signal and the background for observing a light neutral higgs boson via @xmath26 and @xmath27 decay modes
. then we evaluate the asymmetry arising from the interference of the p - even and p - odd production amplitudes .
note that this asymmetry is the most direct manifestation of cp - violation in the higgs sector .
finally we consider the exclusive diffractive production of the heavier neutral higgs bosons , @xmath4 and @xmath5 , followed by the decays @xmath4 or @xmath28-jets . + for numerical estimates , we use the formalism to describe central production in diffractive exclusive processes of @xcite , and the parameters ( that is the masses , width and couplings of the higgs bosons ) given by the code `` cpsuperh '' @xcite , where we choose @xmath29 , @xmath30 , @xmath31 tev , ( that is @xmath32 tev , @xmath33 tev , @xmath34 tev ) and the charged higgs boson mass @xmath35 gev so that the mass of the lightest higgs boson , @xmath2 , is @xmath36 gev .
the exclusive process is shown schematically in fig .
[ fig:1 ] .
the cross section may be written@xcite as the product of the effective gluon
gluon luminosity @xmath37 , and the square of the matrix element of the subprocess @xmath38 .
note that the hard subprocess is mediated by the quark / squark triangles . for a cp - violating higgs , there are two different vertices of the higgs
quark interaction : the scalar yukawa vertex and the vertex containing the @xmath39 dirac matrix .
therefore the @xmath40 matrix element contains two terms : and @xmath41 in the mssm with cp - violation see , for example , @xcite . ] = g_s(e_1^e_2^ ) - g_p ^ e_1e_2p_1p_2/(p_1p_2 ) [ eq:1 ] where @xmath42 are the gluon polarisation vectors and @xmath43 is the antisymmetric tensor . in ( [ eq:1 ] )
we have used a simplified form of the matrix element which already accounts for gauge invariance , assuming that the gluon virtualities are small in comparison with the higgs mass . in forward exclusive central production ,
the incoming gluon polarisations are correlated , in such a way that the effective luminosity satisfies the p - even , @xmath44 selection rule @xcite .
therefore only the first term contributes to the strictly forward cross section .
however , at non - zero transverse momenta of the recoil protons , @xmath45 , there is an admixture of the p - odd @xmath44 amplitude of order @xmath46 , on account of the @xmath41 term becoming active .
thus we consider non - zero recoil proton transverse momenta , and demonstrate that the interference between the cp - even ( @xmath47 ) and cp - odd ( @xmath41 ) terms leads to left - right asymmetry in the azimuthal distribution of the outgoing protons .
first , we consider the background .
unfortunately , even in the exclusive process , we show below that the qcd @xmath22 background is too large .
however , we shall see that it may be possible to observe such a cp - violating light higgs boson in the @xmath48 decay mode , where the qed background can be suppressed by selecting events with relatively large outgoing proton transverse momenta , say , @xmath49 mev .
first , we consider the exclusive double - diffractive process ppp+(h)+p [ eq : a2 ] the signal - to - background ratio is given by the ratio of the cross sections for the hard subprocesses , since the effective gluon
gluon luminosity @xmath37 cancels out . the cross section for the @xmath38 subprocess to indicate that each of the incoming gluons belongs to colour - singlet pomeron exchange . here
this notation is assumed to be implicit . ]
@xcite ( ggh ) = ( 1- ) ~(1- ) , [ eq:2]as the width , @xmath50 , behaves as @xmath51 , where @xmath52 is the fermi constant . on the other hand , at leading order , the qcd background is given by the @xmath53 subprocess ~ , [ eq:3 ] where @xmath54 is the transverse energy of the @xmath3 and @xmath23 jets . at leading order ( lo )
, the cross section is suppressed by the @xmath44 selection rule ( which gives rise to the @xmath55 factor ) in comparison with the inclusive process .
the extra factor was crucial to suppress the background .
it was shown in @xcite that it is possible to achieve a signal - to - background ratio of about 3 for the detection of a standard model higgs with mass @xmath56 gev , by selecting @xmath22 exclusive events where the polar angle @xmath57 between the outgoing jets lies in the interval @xmath58 if the missing mass resolution @xmath59 gev .
the situation is much worse for a light higgs , since the signal - to - background ratio behaves as ~ m_^5 [ eq:4 ] where we have used @xmath60 .
the @xmath61 behaviour comes just from dimensional counting . as the experimental resolution
@xmath62 is larger than the width of the higgs , @xmath63 , the higgs cross section ( in the numerator ) is driven by @xmath64 , while the qcd background is proportional to @xmath65 and the size of the @xmath62 interval . to restore the dimensions we have to divide @xmath66 by @xmath67 .
thus , in going from @xmath56 gev to @xmath68 gev , the expected leading - order qcd @xmath22 background increases by a factor of 240 in comparison with that for @xmath69 gev . strictly speaking
, there are other sources of background @xcite .
there is the possibility of the gluon jet being misidentified as either a @xmath3 or a @xmath23 jet , or a contribution from the nlo @xmath70 subprocess , where the extra gluon is not separated from either a @xmath3 or a @xmath23 jet .
these contributions have no @xmath71 suppression , and hence increase only as @xmath72 , and not as @xmath73 , with decreasing @xmath74 .
for @xmath56 gev , the lo @xmath22 qcd production was only about 30% of the total background .
however , for @xmath75 gev , the lo @xmath22 contribution dominates . finally , with the cuts of ref .
@xcite , we predict that the cross section of the @xmath2 signal is of the recoil protons in the gluon loop of fig.1 . for smaller boson masses , @xmath76 gev ,
this leads to a steeper @xmath77 dependence of the amplitude , which emphasizes larger values of the impact parameter , @xmath78 , where the absorptive effects are weaker .
therefore we obtain a larger soft survival factor , @xmath79 , at the lhc energy . however , recall that a factor of 2 difference is within the accuracy of the approach@xcite . ] @xmath80 as compared to the qcd background cross section , with the same cuts - tagging efficiencies and the missing mass resolution in the case of a light higgs boson are the same as for the case of @xmath81 gev @xcite .
likely , this assumption is not well justified . in particular , the missing mass resolution and proton tagging efficiency may worsen at lower masses .
] , of @xmath82 that is the signal - to - background ratio is only @xmath83 , and so even for an integrated luminosity @xmath84 for @xmath85 gev the significance of the signal is only @xmath86 .
here we have taken a @xmath87 factor of @xmath88 for the qcd @xmath22 background , and again used the cuts and efficiencies quoted in ref.@xcite . therefore , to identify a light higgs ,
it is desirable to study a decay mode other than @xmath26 .
the next largest mode is @xmath27 , with a branching fraction of about 0.07 .
the dependence of the results on the mass of the @xmath2 higgs boson is illustrated in table 1 .
clearly the cross section decreases with increasing mass . on the other hand the signal - to - background ratio increases .
therefore for the case @xmath89 gev we see a slightly improved statistical significance of @xmath90 for the @xmath22 decay mode .
.the cross sections ( in fb ) of the central _ exclusive _ diffractive production of @xmath91 neutral higgs bosons , together with those of the qcd(@xmath22 ) and qed(@xmath92 ) backgrounds .
the acceptance cuts applied are ( a ) the polar angle cut @xmath93 in the higgs rest frame , ( b ) @xmath94 mev for the forward outgoing protons and ( c ) the polar angle cut @xmath95 .
the azimuthal asymmetries @xmath96 are defined in eq.(12 ) . [ cols="<,^,^,^,^",options="header " , ]
we have evaluated the cross sections , and the corresponding backgrounds , for the central double - diffractive production of the ( three neutral ) cp - violating higgs bosons at the lhc .
this scenario is of interest since even a very light boson of mass about 30 gev is not experimentally ruled out for some range of the mssm parameters .
we have studied the production of the three states , @xmath97 , both with exclusive kinematics , @xmath24 which we denoted cedp , and in double - diffractive reactions where both the incoming protons may be destroyed , @xmath98 which we denoted cdd . recall that a + sign denotes the presence of a large rapidity gap .
proton taggers are required in the former processes , but not in the latter .
typical results are summarised in tables 1 and 2 , respectively .
the cross sections are not large , but should be accessible at the lhc .
the uncertainties in the calculation of the exclusive cross sections were discussed in refs.@xcite . for the light @xmath2 boson , where the contribution from the low @xmath99 region is more important
, the uncertainty is much larger . recall that for the semi - inclusive cdd processes the effective gluon - gluon ( @xmath100 ) luminosity
is calculated using the lo formula .
thus we can not exclude rather large nlo corrections . on the other hand , for cdd
, the values of the cross sections are practically insensitive to the contributions from the infrared domain . moreover , with the _ skewed _ cdd kinematics , the nlo bfkl corrections are expected to be much smaller than in the forward ( cedp ) case .
so we may expect an uncertainty of the predictions to be about a factor of 3 to 4 , or even better .
it would be very informative to measure the azimuthal angular dependence of the outgoing proton systems , for both the cedp and cdd processes .
such measurements would reveal explicitly any cp - violating effect , via the interference of the scalar and pseudoscalar @xmath38 vertices .
finally , we recall the advantages of diffractive , as compared to the non - diffractive , production of higgs bosons : \i ) a much better mass higgs resolution is obtained by the missing mass method for exclusive events , \ii ) a clean environment , which may be important to identify four @xmath3-jets with transverse momenta @xmath101 gev ( for the non - diffractive process , at the lhc energy , the qcd backgroud may be too large ) , \iii ) a possibility to measure cp - property of the higgs boson and to detect cp - violation ( note that the asymmetries @xmath102 and @xmath103 are explicit manifestations of cp violation at the quark level ) , next , assuming that p and c parities are conserved , \iv ) the existence of the p - even , @xmath44 selection rule for lo central exclusive diffractive production , which means that we observe an object of natural parity ( most probably @xmath104 ) ; the analysis of the azimuthal angular distribution of the outgoing protons may give additional information about the spin of the centrally produced object @xcite , \v ) in addition we know that an object produced by the diffractive process ( that is by pomeron - pomeron fusion ) has positive c - parity , is an isoscalar and a colour singlet resonance@xcite .
a knowledge of its c - parity is important to understand its nature .
if it is a c = + 1 state with spin - parity @xmath104 or @xmath105 then it may be even seen in cdd production with a large rapidity gap on either side of its j/@xmath106 decay .
forward proton tagging would , of course , allow a better spin - parity analysis . ] . the properties listed above should help to distinguish the @xmath4 and @xmath5 four - jet decay channels from the production of a susy particle , followed by a ` cascade'-like decay .
we thank jeff forshaw , risto orava , albert de roeck , sasha nikitenko , apostolos pilaftsis and , especially , brian cox and jae sik lee for useful discussions . adm
thanks the leverhulme trust for an emeritus fellowship and mgr thanks the ippp at the university of durham for hospitality .
this work was supported by the uk particle physics and astronomy research council , by grant rfbr 04 - 02 - 16073 and by the federal program of the russian ministry of industry , science and technology ss-1124.2003.2 .
khoze , a.d .
martin and m.g .
ryskin , hep - ph/0006005 , in _ proc . of 8th int .
workshop on deep inelastic scattering and qcd ( dis2000 ) _ , liverpool , eds .
j. gracey and t. greenshaw ( world scientific , 2001 ) , p.592 . | we study the central diffractive production of the ( three neutral ) higgs bosons , with a rapidity gap on either side , in an mssm scenario with cp - violation .
we consider the @xmath0 and @xmath1 decay for the light @xmath2 boson and the four @xmath3-jet final state for the heavy @xmath4 and @xmath5 bosons , and discuss the corresponding backgrounds .
a direct indication of the existence of cp - violation can come from the observation of either an azimuthal asymmetry in the angular distribution of the tagged forward protons ( for the exclusive @xmath6 process ) or of a sin@xmath7 contribution in the azimuthal correlation between the transverse energy flows in the proton fragmentation regions for the process with the diffractive dissociation of both incoming protons ( @xmath8 ) .
we emphasise the advantage of reactions with the rapidity gaps ( that is production by the pomeron - pomeron fusion ) to probe cp parity and to determine the quantum numbers of the produced central object .
ippp/03/84 + dcpt/03/168 + 12 january 2004 + * hunting a light cp - violating higgs via diffraction at the lhc * v.a .
khoze@xmath9 , a.d . martin@xmath10 and m.g . ryskin@xmath9 + @xmath10 department of physics and institute for particle physics phenomenology , + university of durham , dh1 3le , uk + @xmath11 petersburg nuclear physics institute , gatchina , st .
petersburg , 188300 , russia + |
low - dimensional quantum magnets show very rich and fascinating physical properties @xcite . as a starting point
, one may consider isolated spin-1/2 dimers with an antiferromagnetic coupling @xmath9 causing a singlet ground state and an excited triplet state , which are separated by an energy gap @xmath10 .
if such dimers are magnetically coupled to each other , a multitude of different theoretical models can be constructed , depending on the strength and the geometric arrangement of the interdimer coupling(s ) @xmath11 . as a consequence of one ( or more ) non - zero @xmath11 , finite dispersion(s ) of the triplet state evolve along the respective direction(s ) in reciprocal space , i.e.
, the excited triplets may hop along different directions . of particular interest
are one - dimensional ( 1d ) chains with alternating couplings @xmath9 and @xmath11 between neighboring spins , because of qualitatively different excitation spectra of the alternating ( @xmath12 ) and the uniform ( @xmath13 ) chain @xcite .
another example of 1d coupled spin dimers is represented by so - called spin - ladders with the couplings @xmath14 and @xmath15 along the rungs and legs of the ladder , respectively .
the excitation gap of two - leg ladders is finite , while it vanishes for a three - leg spin - ladder @xcite .
this difference is easily understood in the limit @xmath16 , where the two - leg ladder can be viewed as weakly coupled dimers and the three - leg ladder as an effective @xmath17 chain with uniform chain coupling @xmath15 . with increasing number of legs ,
the spin ladders approach the 2d antiferromagnetic square lattice .
another well - studied system of 2d - coupled spin dimers is the 2d shastry - sutherland model @xcite , which can be generated from the 2d square lattice by introducing one additional diagonal coupling @xmath18 on every second square .
the triangular arrangement of one @xmath18 and two @xmath9 causes a strong frustration and for @xmath19 the product state of singlets on every diagonal is the exact ground state .
the above - mentioned models have been studied very intensively by theoretical as well by experimental physicists during the last decades , since a large number of materials became available which rather well represent various types of these models @xcite .
some cuprate examples are : the 2d square lattice realized by the parent compound la@xmath20cuo@xmath21 of the high - t@xmath22 s , the spin - peierls system cugeo@xmath2 @xcite , spin-1/2 chain and ladder compounds , such as sr@xmath20cuo@xmath2 and srcu@xmath20o@xmath2 , respectively , as well as sr@xmath23cu@xmath24o@xmath25 containing both , spin chains and spin ladders @xcite , or the 2d shastry - sutherland model which is realized in srcu@xmath20(bo@xmath2)@xmath20 @xcite .
tlcucl@xmath2 is another quantum spin system which has been intensively studied in recent years @xcite . from the structural point of view
this compound contains a ladder - like arrangement of cu@xmath26 ions @xcite .
the main magnetic coupling @xmath27 mev is present along the rungs , but there are various additional , rather large couplings @xmath11 present along different other lattice directions @xcite .
thus , the magnetic system of tlcucl@xmath2 should be viewed as a set of three - dimensionally coupled spin dimers .
the inter - dimer couplings @xmath11 cause a strong dispersion of the triplet excitations , and as a consequence the minimum singlet - triplet @xmath28 mev is much smaller than @xmath9 .
a moderate field of about 5 t is already sufficient to close @xmath29 and induces a nel order with staggered magnetization perpendicular to the applied field .
if there is no magnetic anisotropy in the plane perpendicular to the applied field , this transition is in the same universality class as the bose - einstein condensation ( bec ) and it is often termed a bec of magnons . in the zero - temperature limit , it represents an example of a quantum phase transition @xcite , whose control parameter is the magnetic field strength ( see fig . [ phadi ] ) . in the vicinity of a quantum critical point ( qcp )
anomalous temperature dependencies are expected for various physical properties , as e.g. specific heat @xmath1 , susceptibility @xmath30 , thermal expansion @xmath31 , ( and resistivity @xmath32 for metals ) @xcite .
in particular , a divergence of the so - called grneisen parameter @xmath33 is expected , when a pressure - dependent qcp is approached @xcite .
experimentally , a diverging @xmath34 has indeed been observed for different heavy - fermion compounds @xcite .
since the phase transition of tlcucl@xmath2 is extremely sensitive to pressure @xcite and the control parameter may be easily tuned by a variation of the field , this compound is ideally suited to study such generic properties of a qcp
. obtained from magnetostriction ( @xmath35 ) and thermal expansion ( @xmath36 ) .
the dashed line is a fit of the form @xmath37 for @xmath38 k , which yields @xmath39 t and @xmath40 , while the solid line is a fit up to 8 k yielding @xmath41 .
the @xmath42 dependence of @xmath43 is shown in the inset by the plot of @xmath44 versus @xmath42 on double - logarithmic scales , and is also obtained by quantum monte carlo calculations @xcite .
right : schematic view of the phase diagram around the quantum phase transition from a spin - gap to a nel - ordered ground state at @xmath45 .
the dotted line shows the spin - gap closing due to the zeeman splitting , and the dashed lines indicate the region of enhanced quantum fluctuations at @xmath46 .
the thick solid line is the phase boundary @xmath47 and the thin solid lines indicate the region of enhanced thermal fluctuations . ]
we present high - resolution measurements of the uniaxial thermal expansion @xmath48 and the magnetostriction @xmath49/l_i(0)$ ] along different lattice directions @xmath50 ( @xmath51 is the respective sample length along @xmath50 ) as well as specific heat @xmath1 and magnetization @xmath52 data .
the length changes have been measured down to 250 mk by a home - built capacitance dilatometer and @xmath1 by a home - built calorimeter for @xmath53 mk , while the magnetization has been studied by a commercial vibrating sample magnetometer ( quantum design ) for @xmath54 k. all properties have been studied in magnetic fields up to 14 t. since tlcucl@xmath2 easily cleaves along the ( 010 ) and ( 10@xmath55 ) planes of the monoclinic structure , we measured @xmath56 perpendicular to these planes on a single crystal of dimensions @xmath57
mm@xmath58 perpendicular to ( 010 ) and ( 10@xmath55 ) , respectively .
in addition , the [ 201 ] direction , which is perpendicular to both other directions , was measured on a second crystal of length @xmath59}=4.4 $ ] mm .
for all three measurement directions @xmath50 the magnetic field was applied along the same direction , namely perpendicular to the ( 10@xmath55 ) plane .
fig . [ alpha ] shows @xmath60 measured along all three directions for different magnetic fields . in zero field
, @xmath60 has no visible anomalies , but is strongly anisotropic . for @xmath61 t ,
pronounced anomalies develop and shift systematically towards higher @xmath42 with increasing @xmath62 .
the @xmath60 curves for @xmath63 and @xmath64 agree well with our previous results measured on a different crystal for @xmath65 k @xcite .
the anomalies of @xmath60 signal large uniaxial pressure dependencies of @xmath66 , which are related to @xmath31 and @xmath1 via ehrenfest s relations @xmath67 here , @xmath68 is the molar volume and @xmath69 and @xmath70 denote the respective mean - field jumps at @xmath66 .
the second expression of eq . ( [ ehr ] ) relates the pressure dependencies of the transition field to the jumps of @xmath71 and of the differential magnetic susceptibility @xmath72 . for @xmath63 and @xmath64 ,
the @xmath73 largely cancel each other under hydrostatic pressure , since the anomalies are of comparable magnitudes but of opposite signs .
thus , the hydrostatic pressure dependence @xmath74 is essentially determined by the sign and size of the anomaly of @xmath75 . obviously , the anomalies of @xmath75 are the largest ones and their positive signs mean that @xmath66 drastically increases for uniaxial pressure along @xmath76 $ ] as well as for hydrostatic pressure .
[ ms ] shows the field derivatives of the magnetostriction .
again , we find the characteristic anisotropy that the anomalies of @xmath77 and @xmath78 are of similar sizes but opposite signs , while significantly larger anomalies are present for @xmath79 .
thus , the hydrostatic pressure dependence of @xmath80 is again essentially identical to that for uniaxial pressure along @xmath76 $ ] .
cleavage plane of tlcucl@xmath2 .
note the different scale for the @xmath76 $ ] direction . ] in ref .
@xcite we have shown that the pressure - dependencies of @xmath66 , @xmath80 , and the magnetic susceptibility in the paramagnetic phase can be traced back to the pressure - dependent changes of a single parameter , which in the case of tlcucl@xmath2 is the intra - dimer coupling @xmath9 .
this conclusion was based on the measurements perpendicular to the @xmath81 and @xmath64 planes , and is fully confirmed by the additional new data along the @xmath76 $ ] direction . from the zero - temperature extrapolations of the uniaxial pressure dependencies of @xmath80 for @xmath63 and @xmath64 we estimated @xmath82 % /gpa and @xmath83 % /gpa .
since the anomalies for the @xmath76 $ ] direction are about twice as large , we obtain @xmath84 % /gpa for pressure along @xmath76 $ ] and @xmath85 % /gpa for hydrostatic pressure .
the latter value is in reasonable agreement with direct measurements under hydrostatic pressure , which yield @xmath86 % /gpa for the initial slope at ambient pressure @xcite .
the shape of the @xmath60 anomalies is typical for a second - order phase transition with a pronounced mean - field contribution causing a jump @xmath87 at @xmath66 , superimposed by fluctuations causing a divergence @xmath88 with the reduced temperature @xmath89 and the critical exponents @xmath90 depending on the universality class of the phase transition . on approaching @xmath91 t
, the @xmath60 anomalies broaden to some extent ( see below ) .
the @xmath92 anomalies also show a pronounced fluctuation contribution for @xmath93 k , but become more jump - like for lower @xmath42 .
the changing shapes of both , the @xmath92 and the @xmath60 anomalies can be intuitively understood from fig .
[ phadi ] , because ( i ) the absolute temperature region around the phase boundary where fluctuations become important decreases with decreasing @xmath66 , and ( ii ) close to @xmath45 the phase boundary is crossed with a very small slope as a function of @xmath42 , so that the fact that @xmath80 is not infinitely sharp becomes more and more important . as a criterion for @xmath80 , we used the maximum of the second derivatives @xmath94 , whose full widths at half maximum amount to @xmath95 t ( see inset of fig . [ ms ] ) .
we suspect that this width mainly arises from internal stresses , which broaden the transition due to the strong ( uniaxial ) pressure dependencies of @xmath80 .
we have also investigated , whether there is a finite hysteresis of @xmath80 by measuring @xmath96 with increasing and decreasing @xmath62 . for a drift rate of @xmath97
t / min we obtain a difference @xmath98 t , which does hardly change with temperature . at @xmath99
k we also measured with @xmath100 t / min and found @xmath101 t , i.e. , the observed hysteresis partly arises from the finite field drift .
thus , we regard the phase transition as a second - order one with a weak first - order contribution .
the first - order contribution most probably arises from the large spin - lattice coupling , which may drive a second - order into a first - order transition @xcite .
it is also possible that a transition transforms from second to first order , when @xmath102 is suppressed towards 0 k by an external parameter .
however , the weak temperature - dependence of the hysteresis observed in tlcucl@xmath2 does not give any evidence for such a scenario , and we suspect that the transition of tlcucl@xmath2 remains ( almost ) continuous down to lowest @xmath42 . measured along different directions @xmath50 at various temperatures .
in all cases the magnetic field has been applied perpendicular to the @xmath103 plane ( see text ) . ] as mentioned above , a diverging grneisen parameter @xmath104 has been predicted at @xmath105 @xcite . before comparing our experimental data to this prediction , we will discuss from a phenomenological point of view what may be expected for @xmath106 when the qcp is approached along different routes in the phase diagram .
using maxwell s relations one finds that @xmath107 , while @xmath108 .
if a thermodynamic system can be described by a single energy scale @xmath109 , its entropy @xmath110 only depends on the ratio @xmath111 , i.e. @xmath112 with a model - dependent function @xmath113 .
comparing the @xmath42- and @xmath114-derivatives of @xmath115 yields the grneisen scaling @xmath116 which is temperature independent .
prominent examples of such systems are the debye model with @xmath117 , the almost free electron gas with @xmath118 , or magnetically ordered states with exchange coupling @xmath119 ( for @xmath120 ) . in real systems , one usually observes a weakly @xmath42-dependent @xmath34 , which is due to the fact that the above - mentioned single - parameter models only consider the leading energy scale and neglect others .
if several energy scales @xmath121 are equally important , the individual @xmath121 usually have different pressure dependencies , and therefore the @xmath114- and @xmath42-derivatives are , in general , not ( almost ) proportional to each other . however , it is possible that for different temperature regions different @xmath121 s are dominant and in the respective regions @xmath122 holds .
an example is a coupled spin - dimer system as it is realized in tlcucl@xmath2 . for high temperatures ( @xmath123 ) ,
the behavior is determined by the average spin gap @xmath124 , while the minimum gap @xmath125 becomes dominant at @xmath126 .
thus , @xmath34 varies from @xmath127 to @xmath128 with increasing @xmath42 . in tlcucl@xmath2 ,
the zero - field gap @xmath129 k linearly decreases with @xmath62 @xcite , i.e. @xmath130 where @xmath131 , @xmath132 is the @xmath132-factor and @xmath133 boltzmann s constant .
thus , at a given field @xmath134 the temperature , below which @xmath106 is expected to approach a constant , decreases with @xmath62 . for @xmath135 , however , @xmath136 diverges , because @xmath137 . in finite and zero magnetic field .
the inset shows the anomaly heights as a function of field .
right : expanded view of the low - temperature @xmath138 for different fields below and above the critical field @xmath3 t. the arrow indicates increasing magnetic - field strength .
( see text ) ] for @xmath139 , i.e. in the ordered phase , the characteristic low - temperature energy scale is given by the spin wave velocity @xmath140 .
to our knowledge , the exact dependence of @xmath140 on @xmath141 is not known , but it is quite natural that @xmath140 disappears when @xmath45 is approached . for simplicity , we assume @xmath142 with @xmath143 and derive @xmath144 . thus , the grneisen parameter is again expected to approach a constant , which diverges for @xmath135 , but the sign of the divergence for @xmath139 is opposite to that for @xmath134 . in a next step
we will approach the qcp along the phase boundary @xmath47 for @xmath139 .
for clarity , we use the approximation @xmath145 with @xmath146 for @xmath147 ( see fig . [ phadi ] and refs .
@xcite ) , and calculate @xmath148 .
obviously , @xmath149 diverges for @xmath147 for @xmath150 , and the sign of this divergence is opposite to the sign of @xmath151 .
we emphasize that this result follows from the infinite slope of the phase boundary for @xmath152 , and does not depend on the particular choice of @xmath47 . since the pressure dependence of @xmath66 is given by ehrenfest s relation ( [ ehr ] )
, the ratio @xmath153 of the thermal - expansion and specific - heat anomalies at @xmath66 is expected to diverge for @xmath135 . from eq .
( [ ehr ] ) , it is also clear that the vanishing @xmath154 would cause a divergence of @xmath153 even if the slope of the phase boundary was finite and @xmath155 would thus not diverge for @xmath135 .
let us summarize the above considerations .
on approaching @xmath45 we expect ( i ) for @xmath134 that @xmath156 , which diverges for @xmath135 , ( ii ) for @xmath139 a similar divergence of @xmath157 , but of the opposite sign , and ( iii ) a divergence of the ratio @xmath158 , which has the same ( opposite ) sign as the divergence of @xmath106 above ( below ) @xmath45 .
since the above considerations are rather general , one may expect a divergence of @xmath34 close to many kinds of transitions , whose @xmath102 is suppressed to 0 k. to obtain more information about a quantum phase transition , one has to consider the actual temperature dependencies @xmath7 , @xmath8 , and/or @xmath34 , as it has been done e.g. by the authors of refs.@xcite . at different temperatures .
the finite magnetization in the gap region can be well reproduced by brillouin functions ( lines ) assuming magnetic impurities with spin @xmath159 .
( see text ) ] in fig .
[ cp ] we show the specific heat anomalies for different magnetic fields . in agreement with ref .
@xcite we find rather small anomalies even for the largest field , and their magnitude rapidly decreases when @xmath45 is approached ( see inset of fig .
since the magnitude of the respective @xmath60 anomalies changes much less with field ( see fig.[alpha ] ) , the expected divergence of @xmath73 for @xmath135 is obviously confirmed by the experimental data , since the denominator in eq .
( [ ehr ] ) vanishes .
this is the case for all three directions of uniaxial as well as for hydrostatic pressure . in the right panel of fig .
[ cp ] we show an expanded view of the low - temperature behavior of some @xmath160 curves . for zero field , the onset of an anomaly can be clearly seen @xcite , and this anomaly is suppressed above about 3 t. we suspect that this anomaly arises from an ordering of magnetic impurities .
the presence of such impurities is also evident from the finite magnetization in the gap region @xmath161 at low @xmath42 , which can be well reproduced by brillouin functions ( see fig . [ mag ] ) .
the corresponding fits of the data at @xmath162 k and 2.5 k yield 0.4% of magnetic impurities with spin @xmath159 .
the brillouin function calculated for the same parameters and @xmath163 k is somewhat smaller than the experimental data for @xmath164 t. this is expected , because at this higher temperature a sizeable magnetization from excited triplets is already present . probably , the @xmath159 impurities are mostly ferromagnetically aligned spin dimers , because the intra - dimer coupling between the spin-1/2 cu@xmath26 ions arises from a @xmath165 superexchange via the @xmath114 orbitals of the cl@xmath166 ions , which is very sensitive to changes in the bond angle .
the impurities strongly influence the low - temperature behavior of @xmath8 for low fields , but become much less influential at higher fields because the moments are almost completely saturated for @xmath167 t and @xmath168 k. it is , however , unclear to what extent the impurities may change the ( critical ) behavior close to the phase transition . plane for different magnetic fields below ( left ) and above ( right ) the critical field .
the arrows indicate increasing magnetic - field strength . the bottom panels show some @xmath169 curves for the other directions . ]
[ alpt ] shows an expanded view of the low-@xmath42 behavior of @xmath169 . in zero field ,
@xmath169 continuously approaches zero for @xmath4 , while for larger @xmath62 it shows a pronounced shoulder , which systematically increases and reaches a maximum slightly below @xmath45 . for larger fields clear anomalies with a sign change of @xmath169 occur , and these anomalies systematically sharpen with further increasing field .
the behavior is essentially the same for all three directions , only the magnitudes and signs are different @xcite .
as already mentioned , we attribute the broadening of the anomalies when @xmath45 is approached from larger fields to the finite width of the phase transition .
this also explains that the @xmath170 curves show anomalies already for @xmath171 t , i.e. below @xmath3 t determined by the magnetostriction measurements . in fig .
[ ga ] we present @xmath172 for different magnetic fields .
for all three directions @xmath172 shows the tendency to diverge with decreasing @xmath42 for @xmath173 t. for lower @xmath62 , the @xmath174 follow the same curve at higher @xmath42 , but seem to approach finite values for @xmath4 .
the magnitudes of these limiting values increases with increasing field and the temperature , below which the deviation sets in , decreases .
the @xmath174 for @xmath139 also follow the general curve at higher @xmath42 , until a large anomaly signals the crossing of the phase boundary .
for all three directions , the magnitudes of these anomalies drastically increase with decreasing @xmath62 and the signs are opposite to the respective signs of the diverging @xmath174 for @xmath175 . on this qualitative level , our experimental data of @xmath174
very well confirm the behavior , which one can expect from the above considerations of the gap closing for @xmath134 and the shape of the phase boundary for @xmath139 .
for a deeper understanding , one has to compare the experimental data quantitatively to theoretical predictions . according to ref .
@xcite the following temperature dependencies are expected @xmath105 : @xmath176 obtained from the thermal expansion along different measurement directions @xmath50 in magnetic fields below ( symbols , increasing field strength is indicated by the arrows ) and above ( lines ) the qcp at @xmath3 t. the upper right panel shows @xmath174 for @xmath177 on double - logarithmic scales . here
, the negative @xmath174 for @xmath178 and @xmath76 $ ] have been divided by a factor of @xmath179 and @xmath180 , respectively .
the solid line is a power - law fit of @xmath181 yielding @xmath182 and , for comparison , the expected @xmath183 behavior is shown by the dashed line ( see text ) . ] in the upper right panel of fig .
[ ga ] we show the diverging @xmath174 for @xmath177 on double - logarithmic scales .
because of the negative signs , @xmath184 and @xmath185 have been divided by the factors @xmath179 and @xmath180 , respectively . apparently , the slope is the same within experimental accuracy .
the solid line is a power - law fit of @xmath181 which yields @xmath186 and describes the experimental data reasonably well for about one decade . for comparison ,
the predicted @xmath183 behavior is also shown ( dashed line ) .
in view of the fact that the theoretical prediction only considers the irregular contributions of @xmath1 and @xmath60 , while the experimental data also contain the phononic contributions of @xmath1 and @xmath60 , one may tend to the conclusion that our data nicely confirm the theoretical prediction for @xmath34 .
however , the agreement between theory and experiment becomes much worse when the individual temperature dependencies of @xmath187 and @xmath169 are considered . neither the data of fig .
[ cp ] nor those of fig .
[ alpt ] give any indication to follow the predicted temperature dependencies of eq .
( [ temps ] ) . concerning the specific heat data , one might argue that the predicted @xmath188 behavior is difficult to identify because of the phononic contribution and the influence of the magnetic impurities .
this argument is less convincing for @xmath189 , since ( i ) the predicted divergence should be seen despite a ( regular ) phononic contribution and ( ii ) the ordering of the magnetic impurities does not cause a sizeable anomaly in the zero - field data @xcite .
thus , we conclude that our present experimental data of tlcucl@xmath2 do not confirm the theoretical predictions @xcite .
however , our data do not disprove the theoretical predictions either .
experimentally , one can suspect that in order to observe the predicted temperature dependencies it would be necessary to study ( i ) samples with significantly reduced transition widths and less magnetic impurities , and ( ii ) it might be necessary to extend the measurements to lower temperatures .
moreover , the theoretical predictions have been calculated for clean and isotropic systems .
it is not clear to what extent the temperature dependencies of @xmath31 and @xmath1 are influenced by disorder or a finite magnetic anisotropy .
the latter is reflected in the @xmath190% anisotropy of the @xmath132 factors of tlcucl@xmath2 for different magnetic field orientations @xcite .
in summary , we have presented high - resolution measurements of thermal expansion and magnetostriction along different lattice directions of tlcucl@xmath2 .
both quantities show very pronounced and strongly anisotropic anomalies at the phase boundary of the field - induced nel order @xmath47 for @xmath191 , and signal very large and strongly anisotropic uniaxial pressure dependencies of the transition temperatures and fields .
the hydrostatic pressure dependence of the spin gap @xmath29 for @xmath152 obtained from our data using ehrenfest s relations is in reasonable agreement with the value observed by direct measurements under hydrostatic pressure .
in addition , our data confirm the diverging pressure dependencies @xmath192 for @xmath135 , which are expected from the infinite slope @xmath193 of the phase boundary for @xmath135 , i.e. when the quantum critical point is approached . for @xmath175 ,
the grneisen parameters @xmath194 are expected to approach constant values for @xmath195 , which diverge for @xmath135 , and @xmath196 has been predicted at @xmath197 .
in fact , the experimental @xmath174 for all three measurement directions are in qualitative agreement with these expectations . however , the temperature dependencies predicted for the individual quantities @xmath170 and @xmath8 are not observed experimentally .
for @xmath5 the low-@xmath42 behavior of both @xmath60 and @xmath1 is influenced by the finite transition width and for lower fields at least @xmath8 is also affected by the presence of magnetic impurities .
thus , future measurements on samples of improved quality as well as calculations considering the influence of disorder and weak magnetic anisotropy may clarify the reasons for the puzzling temperature dependencies of @xmath170 , @xmath8 , and @xmath174 .
such an anomaly is not seen in ref .
@xcite , where @xmath8 versus @xmath42 is plotted for @xmath198 k. plotting in the same representation and temperature range , this anomaly is not visible in our data , too . | we present thermal expansion @xmath0 , magnetostriction and specific heat @xmath1 measurements of tlcucl@xmath2 , which shows a quantum phase transition from a spin - gap phase to a nel - ordered ground state as a function of magnetic field around @xmath3 t. using ehrenfest s relation , we find huge pressure dependencies of the spin gap for uniaxial as well as for hydrostatic pressure . for @xmath4 and @xmath5
we observe a diverging grneisen parameter @xmath6 , in qualitative agreement with theoretical predictions .
however , the predicted individual temperature dependencies @xmath7 and @xmath8 are not reproduced by our experimental data . 75.30.kz , 75.80.+q , 65.40.de low - dimensional magnets , bose - einstein condensation , magnetoelastic coupling , quantum phase transition |
various proposals of diverse characters have been suggested by the physicists during the past decades for going beyond , or modifying , einstein general relativity and often for few viable reasons . the most important motivation coming from high - energy physics for adding higher order invariants to the gravitational action , as well as a general motivation coming from cosmology and astrophysics @xcite for seeking generalizations of einstein gravity .
the so - called modified gravities constructed by adding correction terms to the usual einstein - hilbert action ( e.g : @xcite ) , is only one endeavor among others to go beyond einstein general relativity and have opened a new window to study the origin of the current accelerated expansion of the universe @xcite .
for example in lovelock gravity , there have been some attempts for understanding the role of the higher curvature terms from various points of view @xcite . as an another example , a special case of considering the effect of higher curvature corrections
, we will deal with in this paper the so - called @xmath0 gravity @xcite ( and for a review , see , @xcite ) , whose action is an arbitrary function of curvature scalar @xmath2 . when @xmath3 , the einstein s general relativity is recovered .
some of the main reasons to consider @xmath0 gravity are as follows : first of all , there is simplicity : @xmath0 actions are sufficiently general to encapsulate some of the basic characteristics of higher - order gravity , but at the same time they are simple enough to be easy to handle .
second , there are serious reasons to believe that @xmath0 theories are unique among higher - order gravity theories , in the sense that they seem to be the only ones which can avoid the long known and fatal ostrogradski instability @xcite .
third , @xmath0 theories have no ghosts @xcite , and the stability condition @xmath4 of @xcite essentially amounts to guarantee that the scalaron is not a ghost . in @xmath0 gravity , einstein equations
posses extra terms induced from geometry which , when moved to the right hand side , may be interpreted as a matter source represented by the energy - momentum tensor @xmath5 , see equation ( [ fe],[tcurv ] ) . in a similar fashion ,
the space - time - matter ( @xmath6 ) theory , discussed below , results in einstein equations in @xmath7dimension with some extra matter terms showed by the energy - momentum tensor @xmath8 .
it therefore seems plausible to make a correspondence between the matter terms in @xmath6 theory and geometrical terms in @xmath5 resulting in @xmath0 gravity . from the other point of view
, straightforward generalization of the electromagnetic field to higher dimensions one essential property of it is lost , namely , conformal invariance . in ref .
@xcite , it has been shown that the conformal excitations of the extra - dimensional space components have the form of massive scalar fields living in the external ( our ) spacetime .
maxwell theory can be studied in a gauge which is invariant under conformal rescalings of the metric , and firstly , has been proposed by eastwood and singer @xcite .
also , there exists a conformally invariant extension of the maxwell action in higher dimensions ( generalized maxwell field , gmf ) , if one uses the lagrangian of the @xmath9 gauge field in the form @xcite @xmath10 where @xmath11 is the maxwell tensor and @xmath12 is an arbitrary constant .
it is straightforward to show that for @xmath13 , the action ( [ igmf ] ) is invariant under conformal transformation ( @xmath14 and @xmath15 ) and for @xmath16 , the action ( [ igmf ] ) reduces to the maxwell action as it should be .
the idea is to take advantage of the conformal symmetry to construct the analogues of the four - dimensional reissner - nordstrm black hole solutions in higher dimensions .
the main scope of this work is to present the correspondence between @xmath0 gravity and einstein - conformally invariant maxwell theory .
as we show later , these solutions have some interesting properties , specially in the electromagnetic fields , which do not occur in einstein gravity in the present of ordinary maxwell field .
the outline of our paper is as follows . in section [ fieldf(r ) ] we present a short review of field equations of @xmath17-dimensional @xmath0 gravity . in section [ solf(r ) ] the field equations of @xmath1 gravity are solved in the absence of matter field and the resulting solutions are interpreted as black hole .
the solutions of einstein - conformally invariant maxwell source are considered in section [ enonmax ] .
conclusions are drawn in the last section .
the action of @xmath0 gravity , in the presence of matter field has the form of @xmath18 , \label{action}\ ] ] where @xmath2 is the scalar curvature and @xmath0 is an arbitrary function of @xmath2 , and @xmath19 is the lagrangian of matter fields .
variation with respect to metric @xmath20 , leads to the field equations @xmath21 where @xmath22 is the einstein tensor and the gravitational stress - energy tensor is @xmath23 with @xmath24 and @xmath25 the standard matter stress - energy tensor derived from the matter lagrangian @xmath19 in the action ( [ action ] ) .
one can consider geometrical terms in the left hand side of the field equation , and therefore eq .
( [ fe ] ) reduces to @xmath26 the trace of eq .
( [ fieldeq1 ] ) reduces to @xmath27 here , we consider the special case of @xmath0 gravity , namely , @xmath28 gravity and therefore the field equation ( [ fieldeq1 ] ) reduces to @xmath29 r^{n-1}+n\left [ g_{\mu \nu } \box -\nabla _ { \mu } \nabla _ { \nu } \right ] r^{n-1}=\mathrm{t}_{\mu \nu } ^{matt } \label{fieldeq2}\ ] ] it is easy to show that for @xmath30 , eq . ( [ fieldeq2 ] ) reduces to familiar einstein gravity .
here we want to obtain the @xmath17-dimensional static solutions of eqs .
( [ fieldeq2 ] ) without any matter field ( @xmath31 , and then @xmath32 ) with @xmath33 .
we assume that the metric has the following form @xmath34 where @xmath35 which represents the line element of an @xmath36-dimensional hypersurface with constant curvature @xmath37 and volume @xmath38 . to find the function @xmath39
, one may use any components of eq .
( [ fieldeq2 ] ) .
the solution which satisfies all the components of the gravitational field equations ( [ fieldeq2 ] ) , can be written as @xmath40 where @xmath41 and @xmath42 are integration constants which proportional to the mass and charge parameter respectively . in order to study the general structure of this solution
, we first look for the curvature singularities .
it is easy to show that the kretschmann scalar @xmath43 diverges at @xmath44 , it is finite for @xmath45 and goes to zero as @xmath46 .
thus , there is an essential singularity located at @xmath44 .
the event horizon(s ) , if there exists any , is ( are ) located at the root(s ) of @xmath47 @xmath48 the temperature may be obtained through the use of the definition of surface gravity .
one obtains @xmath49 using the fact that the temperature of the extreme black hole is zero , it is easy to show that the condition for having an extreme black hole is that the mass parameter is equal to @xmath50 , where @xmath50 is given as @xmath51 the metric of eqs .
( [ metric ] ) , ( [ domega ] ) and ( [ g(r ) ] ) presents a black hole solution with inner and outer horizons , provided the mass parameter @xmath41 is greater than @xmath50 , an extreme black hole for @xmath52 , and a naked singularity otherwise ( see fig([fig1 ] ) for more details ) .
this behavior is the same as reissner - nordstrm black holes , but , the metric function @xmath39 , presented here , differ from the standard higher - dimensional reissner - nordstrm solutions since the electric charge term in the metric function is proportional to @xmath53 and in the standard higher dimensional charged black hole solutions is proportional to @xmath54 .
it is notable that for @xmath30 ( @xmath3 : einstein gravity ) , the third term in eq .
( [ g(r ) ] ) vanishes and the solutions reduce to schwarzschild like solutions .
in this section , we consider the @xmath17-dimensional action in which gravity is coupled to nonlinear electrodynamics field with an action @xmath55 , \label{ig}\ ] ] where @xmath56 is a coupling constant and the exponent @xmath57 represented the nonlinear power of the electromagnetic field .
varying the action ( [ ig ] ) with respect to the metric tensor @xmath20 and the electromagnetic field @xmath58 , the equations of gravitational and electromagnetic fields may be obtained as @xmath59 @xmath60 = 0 , \label{maxeq}\ ] ] where @xmath61 \label{tnonmax}\ ] ] it is easy to show that when @xmath57 goes to @xmath62 , the eqs .
( [ ig])([maxeq ] ) , reduce to the in einstein - standard maxwell gravity in higher dimensions . the maxwell equation ( [ maxeq ] ) with metric ( [ metric ] )
can be integrated immediately to give @xmath63 where @xmath64 , an integration constant where the electric charge of the spacetime is related to this constant . inserting the maxwell fields ( [ ftr ] ) and the metric ( [ metric ] ) in the field equation ( [ graveq ] )
, one can show that these equations have the following solutions @xmath65 ^{s } , & \text{otherwise } \end{array } \right .
, \label{g(r)2}\ ] ] where @xmath41 is the integration constant which is related to mass parameter . in the linear case ( @xmath66 )
, the solutions reduce to the higher dimensional reissner - nordstrm solutions with linear maxwell source as they should be .
straightforward calculation of kretschmann scalar shows that there is an curvature singularity located at @xmath44 . before studying in details the spacetime ,
we first specify the sign of the coupling constant @xmath56 in term of the exponent @xmath57 in order to ensure a physical interpretation of our future solutions . in fact
, the sign of the coupling constant in the action ( [ ig ] ) can be chosen such that the energy density , i.e. the @xmath67 component of the energy - momentum tensor in the ortonormal frame , is positive @xmath68 this condition selects two branches depending on the range of the nonlinear parameter @xmath57 , @xmath69 while the cases @xmath70 is excluded because in these cases @xmath71 ( and charge term in ( [ g(r)2 ] ) ) vanishes .
now , we want to investigate the special case , such that the electromagnetic field equation be invariant under conformal transformation ( @xmath72 and @xmath73 ) . the idea is to take advantage of the conformal symmetry to construct the analogues of the four dimensional reissner - nordstrm solutions in higher dimensions .
it is easy to show that for lagrangian in the form @xmath74 in @xmath17-dimensions , @xmath75 $ ] ; so @xmath76 implies @xmath77 .
indeed , in our case ( @xmath78 ) , for @xmath13 , the maxwell action ( [ graveq ] ) enjoys the conformal symmetry in arbitrary dimensions . in this case
the metric function @xmath79 and electromagnetic field @xmath71 reduce to @xcite @xmath80 @xmath81 here , we calculate the electric charge of the einstein - conformally invariant maxwell solutions . to determine the electric field , we should consider the projections of the electromagnetic field tensor on special hypersurfaces
. the electric charge per unit volume of the black hole can be found by calculating the flux of the electromagnetic field at infinity , obtaining @xmath82 which confirm that @xmath64 is related to the electrical charge of the spacetime .
comparing eq .
( [ g(r ) ] ) with eq . ( [ gconformal ] ) ( and let @xmath83/(n-1)$ ] ) ,
show that there is a correspondence between the einstein - conformally invariant maxwell solutions and the solutions of @xmath0 gravity without matter field .
one can think about the same effects of @xmath84 in @xmath0 gravity ( eq .
( [ fe ] ) without @xmath85 ) and @xmath25 in einstein - conformally invariant maxwell gravity ( eq.([graveq ] ) ) and therefore @xmath42 is related to the charge parameter @xmath64 .
in the presented paper , we considered the special case of @xmath0 gravity , so - called @xmath28 gravity and obtained its topological black hole solutions in higher dimensions .
we foud that , such as charged black hole solutions , these solutions may be interpreted as black hole solutions with two event horizons , extreme black hole and naked singularity provide that the mass parameter @xmath41 is greater than an extremal value @xmath50 , @xmath86 and @xmath87 respectively .
but the presented solutions differ from the standard higher - dimensional reissner - nordstrm solutions since the electric charge term in obtained metric function is proportional to @xmath53 and in the standard higher dimensional charged black hole solutions is proportional to @xmath54 .
next , we presented topological black hole solutions in einstein gravity with nonlinear electromagnetic field .
then , we restricted ourself to the special case , namely , the conformally invariant maxwell field coupled to einstein gravity and discuss about the correspondence between the einstein - conformally invariant maxwell solutions and the solutions of @xmath0 gravity without matter field in arbitrary dimensions .
it is easy to show that presented solutions reduce to reissner - nordstrm black hole in four dimension .
it is a pleasure to acknowledge discussions with , and/or comments by , r. g. cai , m. h. dehghani and s. otarod .
this work has been supported financially by research institute for astronomy and astrophysics of maragha .
99 s. capozziello , int . j. mod
d 11 ( 2002 ) 483 ; s. capozziello , v. f. cardone , s. carloni and a. troisi , int .
d 12 ( 2003 ) 1969 ; s. capozziello , s. nojiri , s.d .
odintsov and a. troisi , phys . lett .
b 639 ( 2006 ) 135 ; s. nojiri and s. d. odintsov , j. phys .
66 ( 2007 ) 012005 ; s. nojiri and s. d. odintsov , j. phys . a 40 ( 2007 ) 6725 ; f. briscese , e. elizalde , s. nojiri and s. d. odintsov , phys .
b 646 ( 2007 ) 105 ; s. nojiri , s.d . odintsov and h. stefancic ,
d 74 , ( 2006 ) 086009 ; m. e. soussa and r. p. woodard , gen .
rel . grav .
36 ( 2004 ) 855 ; r. dick , gen . rel .
36 ( 2004 ) 217 ; a. e. dominguez and d. e. barraco , phys . rev .
d 70 , ( 2004 ) 043505 ; v. faraoni , phys . rev .
d 75 ( 2007 ) 067302 ; d. a. easson , int . j. mod .
phys . a 19 ( 2004 ) 5343 ; g. j. olmo , phys . rev .
95 ( 2005 ) 261102 ; g. allemandi , m. francaviglia , m. l. ruggiero and a. tartaglia , gen .
rel . grav .
37 ( 2005 ) 1891 ; s. capozziello and a. troisi , phys .
d 72 ( 2005 ) 044022 ; t. clifton and j.d .
barrow , phys .
d 72 ( 2005 ) 103005 ; t. p. sotiriou , gen .
38 ( 2006 ) 1407 ; s. capozziello , a. stabile and a. troisi , mod .
lett . a 21 ( 2006 ) 2291 ; a. dolgov and d. n. pelliccia , nucl .
b 734 ( 2006 ) 208 ; s. nojiri and s. d. odintsov , int . j. geom .
phys . 4 ( 2007 ) 115 .
m. akbar and rong - gen cai , phys .
b635 ( 2006 ) 7 ; m. akbar and rong - gen cai , phys .
b648 ( 2007 ) 243 ; j. c. c. de souza and v. faraoni , class .
( 2007 ) 3637 ; k. atazadeh and h.r .
sepangi , int . j. mod
d 16 ( 2007 ) 687 ; k. atazadeh , m. farhoudi and h. r. sepangi , phys .
b 660 ( 2008 ) 275 ; s. nojiri and s.d .
odintsov , phys .
d 78 ( 2008 ) 046006 ; g. cognola , e. elizalde , s. nojiri , s.d .
odintsov , l. sebastiani and s. zerbini , phys .
d77 ( 2008 ) 046009 ; k. bamba and s.d .
odintsov , jcap 0804 ( 2008 ) 024 .
d. kastor and r. b. mann , jhep 0604 ( 2006 ) 048;m .
h. dehghani and s. h. hendi , phys .
d 73 ( 2006 ) 084021 ; r. g. cai and n. ohta , phys .
d74 ( 2006 ) 064001 ; m. h. dehghani and s. h. hendi , int
d 16 ( 2007 ) 1829 ; m. h. dehghani , n. alinejadi and s. h. hendi , phys .
d 77 ( 2008 ) 104025 ; s. h. hendi and m. h. dehghani , phys . lett .
b 666 ( 2008 ) 116 ; m. h. dehghani , n. bostani and s. h. hendi , phys . rev .
d 78 ( 2008 ) 064031 ; r. g. cai , l. m. cao , y. p. hu and s. p. kim , phys . rev . d78 ( 2008 ) 124012 ; s. h. mazharimousavi and m. halilsoy , phys .
lett . b665 ( 2008 ) 125 ; q. exirifard and m. m. sheikh - jabbari , phys .
b661 ( 2008 ) 158 ; g. a. s. dias , s. gao and j. p. s. lemos , phys . rev .
d75 ( 2007 ) 024030 ; m. farhoudi , gen .
grav.41 ( 2009 ) 117 .
i. zlatev , l. wang and p. j. steinhardt , phys .
82 ( 1999 ) 896 ; p. j. steinhardt , l. wang , i. zlatev , phys .
d 59 ( 1999 ) 123504 ; m. s. turner , int .
a 17s1 ( 2002 ) 180 ; v. sahni , class .
19 ( 2002 ) 3435 .
r. r. caldwell , m. kamionkowski , n. n. weinberg , phys .
91 ( 2003 ) 071301 ; r. r. caldwell , phys .
b 545 ( 2002 ) 23 ; p. singh , m. sami , n. dadhich , phys .
d 68 ( 2003 ) 023522 ; j.g .
hao , x.z .
li , phys .
d 67 ( 2003 ) 107303 . c. armendariz - picon , t. damour and v. mukhanov , phy . lett .
b 458 ( 1999 ) 209 ; m. malquarti , e. j. copeland , a. r. liddle and m. trodden , phys . rev .
d 67 ( 2003 ) 123503 ; t. chiba , phys . rev .
d 66 ( 2002 ) 063514 .
i. l. buchbinder , s. d. odintsov and i. l. shapiro , effective actions in quantum gravity ( iop publishing , bristol ( 1992 ) ) ; m. ferraris , m. francaviglia and g. magnano , class .
5 ( 1988 ) l95 ; k. s. stelle , gen .
( 1978 ) 353 ; k. s. stelle , phys . rev .
d16 ( 1977 ) 953 ; a. strominger , phys . rev .
d30 ( 1984 ) 2257 ; r. utiyama and b. s. dewitt , j. math . phys .
3 1962 608 ; g. a. vilkovisky , class .
quant . grav . 9
( 1992 ) 895 .
a. d. dolgov and m. kawasaki , phys .
lett . b573 ( 2003 ) 1 ; v. faraoni , phys .
d74 ( 2006 ) 104017 ; s. nojiri and s.d .
odintsov , phys .
b652 ( 2007 ) 343 ; s. nojiri and s.d .
odintsov , phys .
b657 ( 2007 ) 238 . | in this paper , we consider the special case of @xmath0 gravity , in which @xmath1 and obtain its topological black hole solutions in higher dimensions . we show that , the same as higher dimensional charged black hole , these solutions may be interpreted as black hole solutions with two event horizons , extreme black hole and naked singularity provided the parameters of the solutions are chosen suitably .
but , the presented black hole is different from the standard higher - dimensional reissner - nordstrm solutions .
next , we present the conformally invariant maxwell field coupled to einstein gravity and discuss about its black hole solutions . comparing these two class of solutions , shows that there is a correspondence between the einstein - conformally invariant maxwell solutions and the solutions of @xmath0 gravity without matter field in arbitrary dimensions . |
the last decade has seen the emergence of highly popular online social networks like myspace , orkut and facebook .
myspace was founded in 2003 and it gained its peak popularity in 2008 @xcite .
however , most of the users have started abandoning it since 2008 @xcite .
cannarella and spechler @xcite have modified the sir model of disease spread @xcite to explain this phenomenon . the disease spread model has also been used to study the arrival and departure dynamics of the users in social networks @xcite . in this paper , we give an alternative explanation of the rise and fall of myspace . to begin with
, we note that people join social networks due to the presence of their friends in these networks and leave them due to the inactivity of these friends @xcite .
it is also known that a large number of the social network friendships are not active or strong friendships between the users @xcite , where a _ strong friendship _ between a pair of social network friends is indicated by regular communication between them .
in fact , the well - known dunbar s number @xcite says that an individual can comfortably maintain a stable relationship with around 150 other people only .
it means that the strong friendships are limited in a social network , even though there is no bound on the number of friends one can have in most social networks . in order to identify the strong friendships between people in a social network , onella et .
al @xcite did an empirical study to establish that greater neighbourhood overlap between a pair of friends corresponds to stronger friendships between them .
the _ neighbourhood overlap _ between a pair of users @xmath0 and @xmath1 in a network is defined as the number of nodes who are neighbours of both @xmath0 and @xmath1 divided by the number of nodes who are neighbours of at least one of them . since
strong or active friendships are important for a user to remain in a social network @xcite , we study the change in the number of users who have at least a certain minimum number of strong friendships in an evolving social network created by the barabasi - albert random graph model @xcite , where we define a pair of users to be strong friends if their neighbourhood overlap is larger than a certain constant .
we also study the change in the size of the largest connected component in the strong friendship subgraph of the evolving barabasi - albert random graph .
this is motivated by the observation that the nodes in the core of a social network are more likely to survive than the nodes at the periphery @xcite .
we consider the largest connected component in the strong friendship subgraph of a social network to be its _ core _ that is important to retain most of its users .
our model is based on the well - known barabasi - albert model @xcite of social networks .
the barabasi - albert model is an algorithm to generate random scale - free networks based on preferential attachment and growth .
the preferential attachment is the property that a node with a higher degree is more likely to get connected to new nodes as the network grows . starting with an initial collection of @xmath2 connected nodes
, the barabasi - albert algorithm adds one node at a time to the network .
each new node is connected to @xmath3 existing nodes as follows .
the probability that the new node is connected to node @xmath4 is @xmath5 , where @xmath6 is the degree of node @xmath4 and @xmath7 is the sum of the degrees of all the nodes in the current network .
the degree distribution resulting out of barabasi - albert model follows power law .
let @xmath8 be the initial graph with @xmath2 vertices , and @xmath9 be the random graph after @xmath10 nodes have been added by the barabasi - albert algorithm . for a constant @xmath11 , we define two users in @xmath9 to be strong friends if their neighbourhood overlap is more than @xmath12 . for a given @xmath12 , the strong friendship subgraph @xmath13 of @xmath9 is defined as follows .
the graph @xmath13 contains the same set of nodes as @xmath9 , and a pair of nodes @xmath0 and @xmath1 are connected by an edge in @xmath13 if and only if they are connected by an edge in @xmath9 and their neighbourhood overlap w.r.t .
@xmath9 is more than @xmath12 .
we run our experiments with three different values of @xmath12 , the threshold for the strong friendship between a pair of users . in each of the cases , we start with a complete graph with @xmath14 vertices , where @xmath15 is the the number of nodes every new node is connected to . in figures
[ fig1 ] , [ fig2 ] and [ fig3 ] , we plot the number of the users having at least @xmath16 ( @xmath16 and @xmath17 , respectively ) strong friends in @xmath13 against the size of the graph @xmath13 for a given @xmath15 and @xmath12 . in each of these plots
, we observe that this number increases till a point before starting to decrease . in figures
[ fig4 ] , [ fig5 ] and [ fig6 ] , we plot the size of the largest connected component ( lcc ) in @xmath13 against the size of the graph @xmath13 for a given @xmath15 and @xmath12 . in each of these plots , we observe that the size of the largest connected component increases till a point , before it reaches a peak and starts decreasing . also , the decline is sharper for higher values of @xmath12 .
our observations indicate that one possible explanation for the fall of myspace is that many of the users started to abandon it after they had a few strong friendships left in the network .
moreover , the remaining users found it difficult to survive as the core of the strong friendship subgraph started reducing in size .
since this might be the effect of preferential attachment where a popular user is likely to befriend a large number of other users , it would be interesting to see whether the same observation does nt hold for random evolving networks that have a restriction on the number of friends each person can have .
it is also an interesting open problem to estimate the size of the network where the plots turn downward , as a function of @xmath15 and @xmath12 , for either of the properties mentioned above .
j. p. onnela , j. saramaki , j. hyvonen , g. szabo , d. lazer , k. kaski , j. kertesz , and a. l. barabasi .
structure and tie strengths in mobile communication networks .
usa , 104 : 7332 - 7336 , 2007 .
s. wu , a. das sarma , a. fabrikant , s. lattanzi and a. tomkins .
arrival and departure dynamics in social networks .
proceedings of the sixth acm international conference on web search and data mining ( wsdm ) , 233 - 242 , 2013 . | the rise and fall of online social networks has recently generated an enormous amount of interest among people , both inside and outside of academia .
gillette [ businessweek magazine , 2011 ] did a detailed analysis of myspace , which started losing its popularity since 2008 .
recently , cannarella and spechler [ arxiv , 2014 ] used a model of disease spread to explain this rise and fall of myspace . in this paper
, we give an alternative explanation for the same .
our explanation is based on the well - known barabasi - albert model of generating random scale - free networks using preferential attachment mechanism . |
in this section we prove eqs .
( 2 ) and ( 4 ) from the paper .
we have : @xmath130 where @xmath131 and we use the normalisation of the state @xmath132 . on the other hand , for a family of pure states @xmath133 the symmetric logarithmic derivative is @xmath134 . therefore @xmath135 in order to prove eq .
( 4 ) , let us consider the discretisation of the master dynamics described below eq . ( 3 ) of the paper .
we have : @xmath136 where @xmath45 , @xmath137 , @xmath138 , and @xmath48 is the initial state of the system . in the limit @xmath139 ,
analogously as tracing out the output in @xmath140 gives the state of the system @xmath141 : @xmath142 the fidelity becomes @xmath143 , where @xmath144 is a modified master operator defined in eq .
( 5 ) in the paper .
the same result can also be derived by using the continuous mps state which describes the state of the system and the output in continuous time : @xmath145 where @xmath146 is the effective hamiltonian .
in this section we first discuss the general dependence of the qfi of the mps state @xmath50 on time @xmath12 .
this will enable us to prove the asymptotic linear behaviour of the qfi in the case of dynamics with a unique stationary state , see eq .
( 6 ) in the paper . using the general time dependence of the qfi
, we then prove the existence of a quadratic scaling regime of the qfi ( cf .
( 7 ) in the paper ) for dynamics near a dpt .
finally , for a system displaying a first - order dpt in photon emissions , we argue how the quadratic scaling of the qfi for phase estimation with emitted photons , can be related to difference in photon emission rates between dynamical phases , cf .
( 10 ) in the paper . in order to express the qfi of the mps state @xmath50 , we use eqs .
( 2 ) and ( 4 ) in the paper , and obtain : @xmath147 where @xmath148 , @xmath149 is the master operator , see eq .
( 3 ) in the paper , and @xmath144 is the modified master operator , see eq .
( 5 ) in the paper . for clarity of further presentation
we assume that @xmath150 can be diagonalised and has one stationary state @xmath151 , i.e @xmath152 , where @xmath153 is the dimension of the system hilbert space @xmath154 and @xmath155 , @xmath156 stand for @xmath157-th right and left eigenvectors of @xmath149 , ordered so that @xmath158 and normalised as @xmath159 , @xmath160 .
the stationary state @xmath151 and other eigenvectors @xmath161 with corresponding eigenvalues @xmath162 depend on @xmath2 .
in general one should consider a jordan decomposition of @xmath53 , but the following discussion would be similar for that case . due to the fact that eq .
involves integrals of @xmath163 , we need to consider the @xmath164-eigenspace of @xmath165 , i.e. the stationary state @xmath151 , separately from all the rest of eigenvectors , whose eigenvalues are different from 0 .
we introduce the projection on the complement of the stationary state @xmath166 and denote the restriction of an operator @xmath167 to the complement of @xmath151 by @xmath168_{\mathcal{p}_1}:=\mathcal{p}_1 \mathcal{x } \mathcal{p}_1 $ ] .
we now express the finite time behaviour of qfi using derivatives of the modified master operator and the diagonal decomposition of the original master operator @xmath53 . from eq .
it follows : @xmath169_{\mathcal{p}_1}|\chi\rangle\langle\chi|\right\}\right|_{g_1=g}^2 \nonumber\\ & & + 4\ , \left(\,t\,\mathrm{tr}\,\left\{\partial^2_{g_1g_2 } \mathcal{l } _ { g_1,g_2 } \,\rho_{ss}\right\}+\mathrm{tr}\,\left\{\partial^2_{g_1g_2 } \mathcal{l } _ { g_1,g_2}\left[\frac{e^{t\mathcal{l}_g}-\mathcal{i}}{\mathcal{l}_g } \right]_{\mathcal{p}_1}|\chi\rangle\langle\chi|\right\}\right)_{g_1=g_2=g}\nonumber\\ & & + 4\ , t^2 \,\left|\mathrm{tr}\,\left\{\partial_{g_1 } \mathcal{l } _ { g_1,g } \,\rho_{ss}\right\}\right|^2 + 8 \,\mathrm{re}\ , \mathrm{tr}\,\left\{\partial_{g_1 } \mathcal{l } _ { g_1,g } \,\rho_{ss}\right\}\mathrm{tr}\,\left\{\partial_{g_2 } \mathcal{l } _ { g , g_2}\left[\frac{e^{t\mathcal{l}_g}-\mathcal{i}-t\mathcal{l}_g}{\mathcal{l}_g^2 } \right]_{\mathcal{p}_1}|\chi\rangle\langle\chi|\right\}_{g_1=g_2=g } \nonumber\\ & & + 8 \,\mathrm{re } \,\mathrm{tr}\,\left\{\partial_{g_1 } \mathcal{l } _ { g_1,g}\left [ \frac{e^{t\mathcal{l}_g}-\mathcal{i}-t\mathcal{l}_g}{\mathcal{l}_g^2 } \ , \right]_{\mathcal{p}_1}\partial_{g_2 } \mathcal{l } _ { g , g_2}\ , \rho_{ss } \right\}_{g_1=g_2=g } \nonumber\\ & & -8 \,\mathrm{re } \,\mathrm{tr}\,\left\ { \partial_{g_1 } \mathcal{l } _ { g_1,g } \left[\mathcal{l}_g^{-1 } \right]_{\mathcal{p}_1}\partial_{g_2 } \mathcal{l } _ { g , g_2}\left [ \frac{e^{t\mathcal{l}_g}-\mathcal{i}}{\mathcal{l}_g } \right]_{\mathcal{p}_1}|\chi\rangle\langle\chi|\right\}_{g_1=g_2=g}\nonumber\\ & & + 8 \,\mathrm{re } \,\mathrm{tr}\,\left\{\partial_{g_1 } \mathcal{l } _ { g_1,g}\left [ \frac{e^{t\mathcal{l}_g}}{\mathcal{l}_g } \ , \left(\int_{0}^{t } \mathrm{d}t'\ , e^{-t ' \mathcal{l}_{g}}\ , \partial_{g_2 } \mathcal{l}_{g , g_2 } \ , e^{t ' \mathcal{l } _ { g}}\right ) \right]_{\mathcal{p}_1}|\chi\rangle\langle\chi|\right\}_{g_1=g_2=g } , \label{sm : qfitime2}\end{aligned}\ ] ] and one can show that @xmath170_{\mathcal{p}_1}&= & \ , t\,\sum_{k=2}^{d^2}\frac{e^{t\lambda_k}}{\lambda_k}\langle l_k | \partial_{g_2 } \mathcal{l } _ { g , g_2 } |\rho_k\rangle \ , |\rho_k\rangle \langle l_k |\\ & & + \
\sum_{j\neq k , j , k>1}^{d^2 } \frac{e^{t \lambda_j } -e^{t\lambda_k}}{\lambda_k(\lambda_j-\lambda_k ) } \langle l_k | \partial_{g_2 } \mathcal{l } _ { g , g_2 } |\rho_j\rangle \ , |\rho_k\rangle \langle l_j |.\end{aligned}\ ] ] the first line and the second line in eq . correspond to the first and the second line in eq . , respectively .
all other terms in eq .
correspond to the third line in eq . .
we see that the quadratic contribution @xmath171 cancels out and for one stationary state there is no explicit quadratic behaviour .
eq . will be used for investigating the asymptotic and the quadratic time regime of qfi .
+ we note that as an alternative route , one can use the eigendecomposition of the modified master operator @xmath144 defined in eq .
( 5 ) in the paper : @xmath172 where @xmath173 and in the case of a single stationary state we have @xmath174 , @xmath175 , @xmath176 , which follows from the normalisation of the eigenbasis of @xmath149 to @xmath177 , i.e. @xmath178 for @xmath179 . from eqs . ( 2 ) and ( 4 ) in the paper , for a single stationary state ,
we obtain : @xmath180 where @xmath181 and @xmath182 , @xmath183 .
the first line correponds to the first line of eq . and
the second to the rest of terms in eq . .
we see again that quadratic terms @xmath184 cancel out and there is no explicit quadratic behaviour .
+ here we assume that the dynamics has a unique stationary state , i.e. the second eigenvalue of the master operator @xmath149 is different from 0 , @xmath185 . in order to find the asymptotic behaviour of the qfi of the state @xmath121
, we consider the limit @xmath186 when we have @xmath187_{\mathcal{p}_1}=0 $ ] and from eq .
we obtain : @xmath188_{\mathcal{p}_1}\partial_{g_2 } \mathcal{l } _ { g , g_2}\ , \,\rho_{ss}\right\}_{g_1=g_2=g } , \label{sm : qfitimealinear}\ ] ] since the limit is finite , this shows that the qfi has an asymptotic linear behaviour in the case of a single stationary state .
this result was also obtained using different methods in @xcite .
we see that eq . can diverge at a first - order dpt when @xmath189 for @xmath190 , as @xmath191 $ ] has then a diverging eigenvalue @xmath192 .
the asymptotic linear behaviour of the qfi can be also obtained from eq .
: @xmath193 which corresponds to eq .
( 6 ) in the paper . by comparing it to eq .
, we see that closing of the gap @xmath189 when approching @xmath190 , can cause non - analyticity of the eigenvalues and eigenvectors of @xmath144 at @xmath194 . in this subsection
we describe the quadratic regime in the qfi scaling with time , which can be present for systems at and near a first - order dpt , and is given by eq .
( 7 ) in the paper .
+ _ quadratic behaviour near a dpt_. we consider a system near a dpt when the gap is much smaller than the gap associated with the rest of the spectrum , i.e. @xmath195 .
this allows to introduce a time regime @xmath196 . for simplicity
let us assume @xmath197 , which gives @xmath198 .
we introduce the projection @xmath199 on the subspace spanned by the stationary state @xmath151 and the second eigenvector @xmath200 .
we also introduce the projection on their complement @xmath201 and denote by @xmath202_{\mathcal{p}_2}=(\mathcal{i - p})x ( \mathcal{i - p})$ ] the restriction of an operator @xmath203 to this complement .
we would like to simplify eq . in the regime @xmath204 .
in this regime we expect the second eigenvector @xmath200 of @xmath149 to be almost stationary and determine , both with the stationary state @xmath151 , dominant terms in the behaviour of the qfi .
we assume that other eigenvectors do not play significant role , by which we understand that @xmath205_{\mathcal{p}_2}\right\rvert_1\approx0 $ ] , where @xmath206 is the trace - norm of operators on @xmath154 , which for density matrices is always @xmath207 .
the general behaviour of the qfi given by eq . simplifies to : @xmath208_{\mathcal{p}_2}|\chi\rangle\langle\chi|\right\}\right|_{g_1=g}^2 \nonumber\\ & & + 4\ , \left(\,t\,\mathrm{tr}\,\left\{\partial^2_{g_1g_2 } \mathcal{l } _ { g_1,g_2 } \,\mathcal{p } \,|\chi\rangle\langle\chi|\right\}-\mathrm{tr}\,\left\{\partial^2_{g_1g_2 } \mathcal{l } _ { g_1,g_2}\left[\mathcal{l}_g^{-1}\right]_{\mathcal{p}_2}|\chi\rangle\langle\chi|\right\}\right)_{g_1=g_2=g}\nonumber\\ & & + 4\ , t^2 \,\mathrm{re}\,\mathrm{tr}\,\left\{\partial_{g_1 } \mathcal{l } _ { g_1,g } \,\mathcal{p } \,\partial_{g_2 } \mathcal{l } _ { g , g_2}\,\mathcal{p } \,|\chi\rangle\langle\chi| \right\}_{g_1=g_2=g } \nonumber\\&&-8 \,\mathrm{re}\ , \mathrm{tr}\,\left\{\partial_{g_1 } \mathcal{l } _ { g_1,g } \,\mathcal{p}\,\partial_{g_2 } \mathcal{l } _ { g , g_2}\left[\frac{\mathcal{i}+t\mathcal{l}_g}{\mathcal{l}_g^2 } \right]_{\mathcal{p}_2}|\chi\rangle\langle\chi|\right\}_{g_1=g_2=g } \nonumber\\ & & - 8 \,\mathrm{re } \,\mathrm{tr}\,\left\{\partial_{g_1 } \mathcal{l } _ { g_1,g}\left[\frac{\mathcal{i}+t\mathcal{l}_g}{\mathcal{l}_g^2 } \right]_{\mathcal{p}_2}\partial_{g_2 } \mathcal{l } _ { g , g_2}\ , \mathcal{p}\ , |\chi\rangle\langle\chi|\right\}_{g_1=g_2=g } \nonumber\\ & & + 8 \,\mathrm{re } \,\mathrm{tr}\,\left\{\partial_{g_1 } \mathcal{l } _ { g_1,g } \left[\mathcal{l}_g^{-1 } \right]_{\mathcal{p}_2}\partial_{g_2 } \mathcal{l } _ { g , g_2}\left[\mathcal{l}_g^{-1 } \right]_{\mathcal{p}_2}|\chi\rangle\langle\chi|\right\}_{g_1=g_2=g}\nonumber\\ & & \quad \quad \,+\ , t^2 \ , \mathcal{o } ( \lambda_2 t)\,\mathcal{o}\left(c_2 ( c_2 + 1)\ , c_1 ^ 2\right ) \nonumber\\ & & \quad\quad \,+\ , t\ , \left\{\mathcal{o } ( \lambda_2 t)\ , \left[\mathcal{o}\left(c_2 \,c_1 ^ 2\,c_2\right)\,+\,\mathcal{o}\left(c_2 \,c_3\right ) \right ] \,+\,\mathcal{o}\left((1+c_2 ) \,c_1 ^ 2\,c_2\left\lvert \left[e^{t \mathcal{l}_g}\right]_{\mathcal{p}_2}\right\rvert_1\right)\,+\,\mathcal{o}\left(c_2c_1 ^ 2 c_2 \right)\mathcal{o}\left(\frac{\lambda_2}{\lambda_3}\right)\right\}\nonumber\\ \nonumber\\ & & \quad \quad\,+\,\mathcal{o}(\lambda_2 t)\,\mathcal{o}(c_2\,c_1 ^ 2\,c_2 ^ 2)+\mathcal{o}\left((1+c_2)\,c_1 ^ 2\,c_2 ^ 2\left\lvert \left[e^{t \mathcal{l}_g}\right]_{\mathcal{p}_2}\right\rvert_1\right)\,+\,\mathcal{o}\left(c_2\,c_3\left\lvert \left[e^{t \mathcal{l}_g}\right]_{\mathcal{p}_2}\right\rvert_1\right)\nonumber\\ & & \quad\quad\,+ \mathcal{o}(c_2\,c_1 ^
2\,c_2 ^ 2)\,\mathcal{o}\left(\frac{\lambda_2}{\lambda_3}\right)\,+\,\mathcal{o}\left(c_1 \left\lvert \sum_{j\neq k , j , k>2}^{d^2 } \frac{e^{t \lambda_j } -e^{t \lambda_k}}{\lambda_k(\lambda_j-\lambda_k ) } |\rho_k\rangle\langle l_k |\,\partial_{g_2 } \mathcal{l } _ { g , g_2}\ , |\rho_j\rangle \langle l_j| \right\rvert_1 \right)_{g_2=g}\label{sm : qfiquadratic},\end{aligned}\ ] ] where corrections in the approximation are given by @xmath209 , @xmath210 , @xmath211_{\mathcal{p}_2}\right\rvert_1 $ ] and @xmath212 .
we note that the estimate of the approximation error in eq .
is very rough and implies strong conditions on the master dynamics to be near a dpt , i.e. the corrections to be negligible . for a particular model one should check the approximation by comparing to the exact results given by eq . .
the quadratic terms in eq . correspond to eq .
( 7 ) in the paper .
let us note that using eq
. does not provide clear results in the quadratic regime . from comparing eq . to eq .
, we see that when @xmath195 , many terms in eq . can diverge . in order to obtain the way they diverge one needs to go back to the operators @xmath213 and @xmath214 and therefore to eqs . and .
+ _ quadratic behaviour at a first - order dpt_. at a first - order dpt we have @xmath215 and the considered time regime is infinitely long due to @xmath216 . since we have @xmath217_{\mathcal{p}_2}\right\rvert_1=0 $ ] , in the limit of long time @xmath12 eq .
becomes exact , as all the corrections disappear .
therefore , eq . gives asymptotic quadratic behaviour of the qfi : @xmath218 see also eq .
( 7 ) in the paper .
+ here we consider a system at a first - order dpt with respect to photon counting statistics .
we show how the quadratic behaviour of the qfi emerges from the bimodality of the total photon number which which the generator of the phase transformation .
this relation is given by eq .
( 10 ) in the paper .
a similar behaviour holds in the general case of arbitrary parameter defence , but this will be discussed in later work @xcite .
we consider the estimated parameter to be the phase @xmath71 encoded on photons emitted by a system , which can be formalised by defining @xmath219 , where @xmath220 is a jump operator of the master operator @xmath221 , see the paper . in that case
the master operator does not depend on @xmath74 , @xmath222 .
we consider the system to be at a first - order dpt , and for simplicity we restrict to the case where the zero eigenvalue of the master operator @xmath221 has degeneracy two .
this means that there exist two stationary states @xmath223 and @xmath224 which are supported on orthogonal subspaces @xmath225 , @xmath226 , so that @xmath227
. moreover , the jump and hamiltonian operators have a block diagonal form in this decomposition @xmath228 , @xmath229 and @xmath230 , @xmath231 , acting on @xmath227 .
let @xmath232 , @xmath233 denote orthogonal projections on @xmath225 , @xmath226 , respectively .
we note that the block - diagonal structure is preserved for @xmath234 , see eq .
( 8) in the paper .
thus , @xmath235 , which leads to the following simple form of the asymptotic qfi : @xmath236 where @xmath237 and @xmath238 . we will now show how the stationary states @xmath223 , @xmath224 correspond to the bimodal distribution of the generator , which is the total photon count @xmath80 , when measured on @xmath50 . for @xmath239
, we have @xmath240 .
we can then express the photon emission average as follows : @xmath241 we represent the initial state of the system as @xmath242 , where @xmath243 and @xmath244 and @xmath245 . by setting @xmath246 or @xmath247
we arrive at @xmath248 and @xmath249 , where @xmath99 , @xmath100 are the asymptotic rates of @xmath80 when the system is initially in the state @xmath243 , @xmath250 , respectively .
. becomes : @xmath251 which corresponds to eq .
( 10 ) in the paper .
let us assume @xmath252 .
we see that in the asymptotic limit @xmath253 we can define two _ dynamical phases _ corresponding to active ( a ) and inactive ( i ) mode in total photon count @xmath80 distribution , to be any mps states @xmath254 , @xmath255 , which after tracing out the output are supported only on @xmath225 , @xmath226 , respectively . in order to ensure quadratic scaling of the qfi @xmath256 , the initial state @xmath105 of the system needs to be a superposition of states from @xmath225 and @xmath226 ,
so that both @xmath257 .
when the asymptotic emission rates are equal @xmath258 , there is no quadratic scaling of the qfi , as the asymptotic distribution of photon counts @xmath80 is unimodal with variance scaling linearly with time @xmath12 .
+ _ quadratic behaviour and approximate bimodality for an intermittent system_. near a dpt in photon emissions , the system dynamics is intermittent and switches between long time intervals with different emission rates .
the typical length of those intervals is given by the correlation time @xmath259 .
therefore , we expect to be able to construct approximate stationary states for the quadratic qfi regime @xmath260
. let us note that the master operator @xmath221 has only one stationary state @xmath151 of @xmath221 and its second eigenvector @xmath200 fulfills @xmath261 due to normalisation of eigenvectors .
nevertheless , @xmath221 is degenerate up to order @xmath262 .
below we briefly present a construction of two approximately stationary states with different emission rates as linear combinations of @xmath151 and @xmath200 .
the construction closely follows the theory of classical non - equilibrium first - order phase transitions @xcite .
we leave rigorous proofs and discussion of eq . in that case for later work @xcite .
consider first the case when a first - order dpt is approached by changing parameters in the master operator @xmath221 , which will then guide the more general considerations near a dpt .
when approaching the dpt , the first two eigenvectors converge to @xmath263 and @xmath200 , such that @xmath264 , @xmath265 and @xmath266 .
one can show that in that case @xmath267 , @xmath268 and @xmath269 , where @xmath270 , @xmath223 , @xmath224 are the stationary states supported on orthogonal subspaces @xmath225 , @xmath226 , respectively , and @xmath232 , @xmath233 are the orthogonal projections on these subspaces . in the case of the system near a dpt ,
the construction of approximate stationary states is as follows .
the master operator @xmath221 acts almost block - diagonally , i.e. @xmath40 , @xmath41 , @xmath271 are approximately block - diagonal with respect to an orthogonal splitting into subspaces @xmath272 and @xmath273 . for pure initial states @xmath274 ,
@xmath275 supported in these subspaces , the corresponding evolved states @xmath276 , @xmath277 will still be supported in the blocks for time @xmath278 in quadratic regime of qfi , @xmath279 , and will be well approximated by linear combinations of @xmath151 and @xmath200 ( due to @xmath280 ) .
moreover , @xmath276 , @xmath277 are almost stationary w.r.t @xmath281 , where @xmath282 . in order to define the approximate blocks , @xmath272 , @xmath273 , and the initial states @xmath274 , @xmath275 using the master operator @xmath221
, we assume @xmath283 for simplicity . in this case both @xmath200 and @xmath284 are hermitian matrices on @xmath154 , i.e. they diagonalise and their spectra are real .
first , inspired by the form of the second eigenvector at a first - order dpt , @xmath269 , we define the subspaces @xmath272 , @xmath273 in the following way .
@xmath272 is spanned by the eigenvectors of @xmath284 which correspond to positive eigenvalues close to the maximal eigenvalue of @xmath284 , while @xmath273 is spanned by eigenvectors of @xmath284 corresponding to negative eigenvalues close to the minimal eigenvalue of @xmath284 .
next , the initial states @xmath274 and @xmath275 are chosen to be the eigenvectors corresponding to maximal and minimal eigenvalue of @xmath284 , respectively . finally , the two approximate dynamical phases in photon emissions are defined as any mps states which after tracing out the output are supported mostly on @xmath272 , @xmath273 , respectively . using the above decomposition
, it can be shown that the quadratic behaviour of the qfi is again related to the two approximate modes in the counting distribution given by the approximate dynamical phases , see eq .
( 10 ) in the paper . | we consider the general problem of estimating an unknown control parameter of an open quantum system .
we establish a direct relation between the evolution of both system and environment and the precision with which the parameter can be estimated .
we show that when the open quantum system undergoes a first - order dynamical phase transition the quantum fisher information ( qfi ) , which gives the upper bound on the achievable precision of any measurement of the system and environment , becomes quadratic in observation time ( cf .
`` heisenberg scaling '' ) .
in fact , the qfi is identical to the variance of the dynamical observable that characterises the phases that coexist at the transition , and enhanced scaling is a consequence of the divergence of the variance of this observable at the transition point .
this identification allows to establish the finite time scaling of the qfi .
near the transition the qfi is quadratic in time for times shorter than the correlation time of the dynamics . in the regime of
enhanced scaling the optimal measurement whose precision is given by the qfi involves measuring both system and output . as a particular realisation of these ideas
, we describe a theoretical scheme for quantum enhanced phase estimation using the photons being emitted from a quantum system near the coexistence of dynamical phases with distinct photon emission rates . _
introduction . _
the estimation of unknown parameters is a crucial task for quantum technology applications such as state tomography @xcite , system identification @xcite , and quantum metrology @xcite .
enhancement in precision can be achieved by using highly correlated / entangled quantum states which encode the unknown parameter , like the greenberger - horne - zeilinger ( ghz ) state @xmath0 constructed out of @xmath1 qubits .
with such a state as a resource an unknown parameter @xmath2 can be encoded as @xmath3 . since the phase effectively encoded in the state
is @xmath4 , the estimation error on @xmath2 scales as @xmath5 ( referred to as heisenberg scaling @xcite ) instead of the standard @xmath6 scaling for a non - correlated state @xmath7 .
the key property that makes correlated states such as @xmath8 useful for enhanced metrology is that they can be thought of as `` bimodal '' , in the sense that the probability of an appropriate observable is peaked in two ( or more ) `` phases '' ( the states @xmath9 and @xmath10 in the case of @xmath8 ) .
this bimodality is reminiscent of what occurs near a ( first - order ) phase transition .
in fact , enhanced parameter estimation can be achieved with pure states at quantum phase transitions @xcite . for large @xmath1 ,
highly correlated _ pure _ states are challenging to prepare in practice @xcite , either as the ground state of a closed many - body system , or as the stationary state of some dissipative dynamics @xcite .
typically , the latter requires careful system engineering , since generic open quantum systems have _ mixed _ rather than pure stationary states . in general
one therefore has to deal with mixed states .
these however have an additional complication since the best possible measurement is difficult to formulate in general , except for particular cases such as thermal states @xcite .
this means that with mixed states it is often difficult to compute the best possible precision of parameter estimation . in this paper
we show theoretically how to exploit the dynamics of open quantum systems ( for example , driven atomic or molecular ensembles emitting photons @xcite , or quantum dots @xcite ) to generate states for quantum enhanced metrology .
our approach connects to recent work on parameter estimation with single stationary states of open quantum systems @xcite .
we overcome the problem of mixed states by considering the combined state of the system and output .
this is a pure quantum state actually a matrix product state ( mps ) @xcite which encodes the state of the system as well as the record of emissions for the whole observation time .
this allows us to find the best estimation precision using the system - output as a resource .
this approach has several advantages .
first , it provides improved precision due to the fact that the effective `` size '' of the system and output is now @xmath11 where @xmath12 is the observation time and @xmath1 is the system size .
the second advantage arises from the fact that open systems can feature _ dynamical _ phase transitions ( dpts ) @xcite which , in contrast to static transitions , are characterised by singular changes in observables on the whole dynamical evolution and not just on the state of the system . we show that at a first - order dpt @xcite the quantum fisher information ( qfi ) of the system - and - output may become quadratic in @xmath12 giving rise to heisenberg scaling .
we also clarify the behaviour away from the transition point . here
heisenberg scaling of the qfi is present for times shorter than the correlation time of the dynamics , while asymptotically linear scaling is recovered .
moreover , due to the pure form of the system - output state we can always ( formally ) construct the optimal measurement .
we discuss our ideas in a specific setting for quantum enhancement in phase estimation using an intermittent system near a dynamical first - order transition as shown in fig .
[ fig : system](a ) .
_ elements of quantum metrology .
_ we first review some essential aspects of quantum parameter estimation .
suppose that we wish to estimate a parameter @xmath2 encoded in a quantum state @xmath13 , by measuring an observable @xmath14 .
the estimation precision is given by the signal to noise ratio @xcite , @xmath15 , where @xmath16 and @xmath17 are the mean and variance of measuring @xmath14 on @xmath13 , respectively .
the observable with the optimal snr is ( up to linear transformations ) given by the so - called symmetric logarithmic derivative , @xmath18 , defined by the relation @xcite @xmath19 except for very particular forms of @xmath20 , the optimal measurement @xmath18 is difficult to engineer .
nevertheless , the snr for this observable is given by the quantum fisher information @xcite , @xmath21 , which bounds the precision of any measurement that can be performed in practice .
this bound is in fact given by the variance of @xmath18 , i.e. @xmath22 . in general , the qfi is hard to compute , but for a pure state , @xmath23 , it can be obtained from the fidelity @xmath24 @xcite according to , @xmath25 a situation which is relevant for what follows is when the parameter is encoded as a phase in a unitary transformation on a pure state , @xmath26 . here
the fidelity @xmath24 is the characteristic function of @xmath27 at @xmath28 , and the qfi is given by its variance , @xmath29 .
note that while the qfi is given by the variance of both @xmath18 and @xmath27 , these two operators play very different roles .
the optimal measurement to recover the parameter @xmath2 is @xmath18 , and its snr is maximal , @xmath30 .
in contrast , @xmath27 encodes @xmath2 in the quantum state , but measuring it provides no information about @xmath2 since @xmath31 .
for example , for the state @xmath32 the generator is @xmath33 and the optimal measurement @xmath34 , where @xmath35 are pauli operators acting on qubit @xmath36 .
the qfi for the ghz state then obeys heisenberg scaling , @xmath37 .
this is related to the fact that the distributions of both @xmath27 and @xmath18 are bimodal .
in contrast , the qfi of the uncorrelated state is standard , @xmath38 , given by the fact that the corresponding distributions are unimodal .
below we show that an analogous change from bimodal to unimodal also accompanies a change in the scaling with time of the qfi when approaching a first - order dpt . _
open dynamics , mps and dpts . _
our goal is to explore open quantum systems as resources for parameter estimation .
we consider systems whose reduced dynamics , after tracing out the environment , is given by a markovian master equation @xcite @xmath39 + \sum_{j=1}^k \left ( l_j \rho l_j^\dagger - \frac{1}{2}\{l_j^\dagger l_j , \rho\ } \right ) \label{eq : master}\end{aligned}\ ] ] where @xmath40 is the system s hamiltonian , and @xmath41 are the jump operators ( @xmath42 ) . in the input - output formalism @xcite , the _ joint system and output state _ is given by a continuous mps ( cmps ) @xcite . for clarity ,
we discretise time by @xmath43 , and the cmps is approximated by a regular mps @xcite , @xmath44 , where @xmath45 , @xmath46 , @xmath47 , and @xmath48 is the initial state of the system .
the output state @xmath49 describes the time record of emissions into the environment , as sketched in fig .
1(a ) .
the state @xmath50 can have a singular change when varying a parameter in .
this could either correspond to a static phase transition in the stationary state of the system , or to a dynamical phase transition in the system and output .
both kinds of transitions are captured by discontinuities in the average , or a higher cumulant , of an observable that acts on the whole of @xmath50 .
_ relation between dpts and qfi .
we now assume that the dynamics depends on the parameter @xmath2 to be estimated , see fig .
1(a ) .
this means that the hamiltonian , @xmath51 and jump operators , @xmath52 , and consequently the master operator , @xmath53 , eq .
( [ eq : master ] ) , may depend on @xmath2 .
it follows then that the mps , @xmath54 , also depends on @xmath2 , and so does the fidelity @xcite , @xmath55 where @xmath56 is a deformation of the master operator @xcite , @xmath57 \nonumber.\end{aligned}\ ] ] thus , in the long time limit the qfi of @xmath54 is related to the largest eigenvalue @xmath58 of @xmath56 , @xmath59 one can already see that something interesting will occur as the system approaches a dpt , so that the gap between the two leading eigenvalues of @xmath53 closes at some @xmath2 .
when the gap is small , for example close to a dpt , there is a time regime where the qfi is quadratic in time , @xmath60 where @xmath61 is a projection onto the first two eigenvectors of @xmath53 corresponding to the two eigenvalues with the largest real part , @xmath62 .
the gap is given by @xmath63 .
this approximation of eq .
is valid for @xmath64 , where @xmath65 is the correlation time given by the gap , @xmath66 , while @xmath67 is the longest timescale associated with the rest of the spectrum , @xmath68 .
the quadratic time dependence of the qfi is a consequence of time - correlations in the system - output mps .
furthermore , at a dtp @xmath69 and the asymptotic scaling of eq .
is no longer valid .
instead the qfi is quadratic in time and this _ heisenberg scaling _ is given exactly by the eq . for all @xmath70 .
_ enhanced phase estimation and intermittency .
_ we now use the ideas above for the case of a system with intermittent dynamics @xcite used as a resource for parameter estimation , see fig .
1(a ) .
the parameter here is a phase @xmath71 encoded in the jump operator @xmath72 , by defining @xmath73 . for concreteness ,
note that the quantum jump associated with @xmath72 is the emission of a photon .
this means that a phase @xmath74 is imprinted on each outgoing photon .
as we now show , if the system displays intermittent photon emission associated to a ( first - order ) dpt in counting statistics @xcite , then it will be an efficient resource for quantum metrology . with the above choice
, the master operator is independent of @xmath74 , @xmath75 . in turn ,
the deformed generator @xmath76 , eq . , from which the qfi is obtained , reads ( @xmath77 ) @xmath78 with these definitions there is a direct connection to a photon counting problem @xcite .
the phase @xmath74 is encoded in a unitary transformation of the mps with generator @xmath79 , where @xmath80 is the operator that counts the number of photons emitted up to time @xmath12 , so that @xmath81 .
the fidelity @xmath82 is the characteristic function of @xmath80 , the logarithm of which encodes all its cumulants .
the cumulants are also encoded in the cumulant generating function ( cgf ) , @xmath83 , where @xmath84 is the probability of observing @xmath85 photons in time @xmath12 .
the cgf can be related to a deformation of the master operator , @xmath86 , where @xmath87 is the same as with @xmath88 .
the long time limit of the cgf , @xmath89 , plays the role of a dynamical free - energy for the ensemble of trajectories of photon emissions @xcite .
a singularity of @xmath90 at some @xmath91 is an indication of a phase transition in the ensemble of quantum jump trajectories , and when @xmath92 we have what we term a dpt , i.e. , a singular change in the actual dynamics of the open system associated with a vanishing of the spectral gap @xmath93 @xcite . the asymptotic qfi becomes , @xmath94 when the function @xmath90 has a first - order singularity at some @xmath95 , i.e. we are near a dpt , eq
. will be large at @xmath96 . in this a case
the system will display an intermittent dynamics that switches between long periods with very distinct emission characteristics .
such a situation can be understood in terms of the coexistence of dynamical phases with significantly different photon count rates @xcite , see fig .
1(a ) .
the qfi of @xmath97 is proportional to the variance of the photon counting generator @xmath79 .
for times shorter than the correlation time @xmath65 the system is mostly in one of the two phases , the distribution of the photon count is approximately bimodal , and the dynamics displays large fluctuations in the total photon emission , see fig .
1(b ) .
this implies a quadratic increase of the qfi with time , with eq .
reducing to , @xmath98 here @xmath99 and @xmath100 are the average counting rates , @xmath101 , in the two phases ( which we term `` active '' and `` inactive '' as we assume @xmath102 ) , while @xmath103 and @xmath104 are the probability of the initial state @xmath105 being in either phase .
the above approximation holds for @xmath106 , and becomes valid for all times at a dpt . for times longer than @xmath65 , dynamics switches between the two phases , giving rise to intermittent behaviour , and eventual normal ( unimodal ) distribution of the photon count around the overall average
; see fig .
1(b ) and derivations in @xcite .
the above shows that an intermittent system near a dpt can be used as a photon source for quantum enhanced phase estimation .
the situation is then similar to that of ghz states : the total photon count distribution is bimodal for times up to the correlation time @xmath65 and imprints an effective macroscopic phase difference of @xmath107 between the active and inactive dynamical phases ; see discussion after eq . .
_ enhanced metrology and dpt in general .
_ we now extend the above discussion to the case where the dynamics has an arbitrary dependence on the parameter @xmath2 to be estimated . in this case , @xmath2 is encoded in the action of a generator " @xmath108 , @xmath109 where @xmath108 is the time - integral of a local - in - time observable , just like @xmath80 in the photon counting case . in terms of @xmath108
the fidelity reads , @xmath110 , where @xmath111 is the @xmath2-ordering ( cf .
time - ordering ) operator , see also @xcite .
the qfi is then the variance of @xmath108 .
it follows that if we have a system which displays a first - order dpt where the dynamical phases are characterised by @xmath108 , then , in the @xmath64 time regime , the qfi follows eq . , where @xmath112 are the averages of @xmath108 per unit time in the two coexisting dynamical phases @xcite .
again this emphasises the connection between dynamical bimodality and enhanced quantum sensitivity .
the @xmath113 behaviour of the qfi is an intrinsically quantum feature .
this behaviour can not occur in systems for which the associated mps is real and therefore can not accumulate any quantum phase .
note that this includes all classical systems .
in such a case the average of @xmath114 is zero , cf .
. , and only terms linear in @xmath12 will survive in eq . .
_ measurement schemes .
_ we have shown that near a dpt the system - output state can have a large qfi . but to exploit this , and achieve quantum enhanced sensitivity , it is necessary to measure an appropriately chosen observable .
the optimal observable is known to be the symmetric logarithmic derivative @xmath18 defined by ( [ eq : log ] ) , which for pure states can be written explicitly as @xmath115 .
however , the measurement of @xmath18 will be difficult to engineer in most practical situations .
one needs therefore to find an alternative which is both practical and whose snr is as close as possible to the qfi . despite the fact that the intricacy of the optimal measurement makes it impractical
, we can still formulate general characteristics for a measurement that achieves enhanced precision .
the first consideration is whether the measurement should be on the system or output , or both .
in fact , in the regime of enhanced scaling the optimal measurement whose precision is given by the qfi involves measuring both system and output .
the reason is that the precision achievable by measuring only the output is bounded by @xmath116 , which scales linearly in time . here
@xmath117 are the mps states associated to the individual active / inactive stationary states , and @xmath118 are their probabilities , see eq . .
this last result is the precision of an idealised protocol given by a first measurement of the system to project onto one of the subspaces associated with the competing stationary states , followed by an optimal measurement of the conditioned system - output state @xmath117 .
the second consideration is what should be the time extension @xmath12 of a single measurement run .
here we imagine that the total time available to the experiment is @xmath119 and one performs @xmath120 independent repetitions of an efficient system - output measurement of the state @xmath121 .
this corresponds to a measurement of the joint state @xmath122 , and the optimal time @xmath12 is that which maximises the qfi of the joint state , @xmath123 .
equation tells us that this optimal time is of the order of the correlation time , @xmath124 . for the case of phase estimation at a dpt
, the bimodality of the system - output state means that it is essentially of the form of a `` schrdinger cat '' state .
assuming for simplicity that the competing stationary states are pure , it reads , @xmath125 where @xmath126 are coherent states with amplitudes @xmath127 , where @xmath128 are the photon emission rates of the phases , see eq . and fig .
1(c ) .
in fact , as shown in ref .
@xcite , the state is approximately a ghz state with relative phase @xmath129 .
note that for neither counting nor homodyne measurements achieve heisenberg scaling , which highlights the general challenge of identifying optimal measurements .
however , one might think of instead employing interferometric protocols , related to the ones put forward in refs .
@xcite for superpositions of coherent states , in order to exploit the enhanced precision scaling .
_ conclusions .
_ we have shown that , close to a dynamical phase transition , the output of an open quantum system can be seen as a resource for quantum metrology applications . for times of the order of the correlation time
, the system - output qfi scales quadratically with time , while in the long time limit the qfi scales linearly in time with rate which diverges when the spectral gap closes , as in a dtp .
it remains an open issue how to exploit in a general and systematic way the large qfi of the system - output close to a dpt .
this work was supported by the leverhulme trust ( grant no .
f/00114/bg ) , epsrc ( grant no . ep / j009776/1 ) and the european research council under the european union s seventh framework programme ( fp/2007 - 2013 ) through erc grant agreement no . 335266 ( escquma ) and the eu - fet grant no .
512862 ( hairs ) .
i.l . acknowledges discussions with k. mlmer and e.m .
kessler .
10 h. hffner , w. hnsel , c. f. roos , j. benhelm , d. chek - al - kar , m. chwalla , t. krber , u. d. rapol , m. riebe , p. o. schmidt , c. becher , o. ghne , w. dr and r. blatt , nature * 438 * , 643 - 646 ( 2005 ) .
d. burgarth and k. yuasa , phys .
rev .
lett . * 108 * , 080502 ( 2012 ) . for a review
see , v. giovannetti , s. lloyd and l. maccone , nature photon . * 5 * , 222 ( 2011 ) .
ligo scientific collaboration , nature phys .
* 7 * , 962 ( 2011 ) .
d.j .
wineland , j.j .
bollinger , w.m .
itano , f.l .
moore and d.j .
heinzen , phys .
rev . a , * 46 * , r6797 ( 1992 ) ; d. leibfried , m.d .
barrett , t. schaetz , j. britton , j. chiaverini , w.m .
itano , j.d .
jost , c. langer and d.j .
wineland , science * 304 * , 1476 ( 2004 ) ; c.f .
roos , m. chwalla , k. kim , m. riebe and r. blatt , nature * 443 * , 316 ( 2006 ) .
c.m .
caves , phys .
rev .
d * 23 * , 1693 ( 1981 ) .
p. zanardi , p. giorda , m. cozzini , phys .
rev .
lett . * 99 * , 100603 ( 2007 ) ; p. zanardi , m.g.a .
paris , l. campos venuti , phys .
rev .
a * 78 * , 042105 ( 2008 ) .
r. blatt , g.j .
milburn and a. lvovksy , j. phys .
b * 46 * 100201 ( 2013 ) . s. diehl , a. micheli , a. kantian , b. kraus , h.p .
bchler and p. zoller , nature physics , * 4 * , 878 ( 2008 ) ; b. kraus , h.p .
bchler , s. diehl , a. kantian , a. micheli , and p. zoller , phys .
rev .
a 78 , 042307 ( 2008 ) .
z. jiang , phys .
rev , a * 89 * , 032128 ( 2014 ) .
j.c .
bergquist , r.g .
hulet , w.m .
itano and d.j .
wineland , phys .
rev .
lett . * 57 * , 1699 ( 1986 ) .
e. barkai , y.j .
jung , and r. silbey , annu .
rev .
phys . chem . *
55 * , 457 ( 2004 ) .
m.m . wolf , g. ortiz , f. verstraete , j.i .
cirac , phys .
rev .
lett . * 97 * , 110403 ( 2006 ) ; f. verstraete and j. i. cirac , phys .
rev .
lett . * 104 * , 190405 , ( 2010 ) .
i. lesanovsky , m. van horssen , m. guta , and j.p .
garrahan , phys .
rev .
lett . * 110 * , 150401 ( 2013 ) .
j.p .
garrahan , i. lesanovsky , phys .
rev .
lett . * 104 * , 160601 ( 2010 ) . c. ates , b. olmos , j.p .
garrahan , and i. lesanovsky , phys .
rev .
a * 85 * , 043620 ( 2012 ) .
m. guta , phys , rev .
a , * 83 * , 062324 ( 2011 ) . s. gammelmark , k. mlmer , phys .
rev .
lett .
* 112 * , 170401 ( 2014 ) . c. catana , l. bouten , m. guta , arxiv:1407.5131 .
b. m. escher , arxiv:1212.2533 . c. w. helstrom , phys
. lett . *
25a * , 101 ( 1967 ) ; ieee trans .
inf .
theory * 14 * , 234 ( 1968 ) ; s. l. braunstein and c.m .
caves , phys .
rev .
lett . * 72 * , 3439 ( 1994 ) . v. gorini , a. kossakowski , and e.c.g .
sudarshan , j. math . phys . * 17 * , 821 ( 1976 ) ; g. lindblad , commun .
math .
phys . * 48 * , 119 ( 1976 ) .
c.w .
gardiner and p. zoller , _ quantum noise _ ( springer - verlag , berlin , 2004 ) ( 3rd edition ) .
see supplemental material .
m.b .
plenio and p. l. knight , rev .
mod .
phys . * 70 * , 101 ( 1998 ) .
a. de pasquale , d. rossini , p. facchi , v. giovannetti , phys .
rev .
a * 88 * , 052117 ( 2013 ) .
m. guta , j. kiukas , arxiv:1402.3535 .
k. macieszczak , m. guta , i. lesanovsky , j.p .
garrahan , [ in preparation ] .
t.c .
ralph , phys .
rev .
a * 65 * , 042313 ( 2002 ) .
j. joo , w.j .
munro and t.p .
spiller , phys .
rev .
lett . * 107 * , 083601 ( 2011 ) .
c.c .
gerry , j. mimih , phys .
rev .
a , * 82 * , 013831 ( 2010 ) . b. gaveau , l. s. schulman , j. math . phys . * 39 * , 3 ( 1998 ) . |
the energy and the momentum spectrum of a hadron @xmath16 produced in the annihilation process @xmath17 is described by a fragmentation function @xmath18 . here
@xmath19 is either the fractional momentum , @xmath20 , or the fractional energy , @xmath21 , carried by a hadron @xmath16 , and @xmath22 is the centre - of - mass energy of the annihilation process with total hadronic cross - section @xmath23 . in the case of unpolarized e@xmath24 beams , and
averaging over the polarization of the hadron @xmath16 , the fragmentation function receives contributions from the transverse ( @xmath25 ) and longitudinal ( @xmath26 ) polarization states of the intermediate electroweak vector bosons , @xmath27 and , and from their interference yielding an asymmetric contribution ( @xmath28 ) .
the most general form of the differential cross - section for the inclusive single - particle production in annihilation is @xcite @xmath29 where @xmath30 is the polar angle between the direction of the incoming and the outgoing hadron @xmath16 . at centre - of - mass energies much larger than the mass of the produced quark @xmath31 , the longitudinal contribution is negligible @xcite due to the helicity structure at the quarks production vertex .
a sizeable contribution to the longitudinal fragmentation function , however , comes from gluon radiation from the @xmath32 system in the final state @xcite .
the asymmetric contribution is largest at energies below and above but very small at the z peak . even though it is @xmath33-@xmath34% of the transverse cross - section @xcite in the energy range of @xmath0 through @xmath1 gev considered for this analysis , the required experimental distinction of quark and antiquark renders a measurement of the asymmetric contribution virtually impossible .
it , therefore , was not considered for this analysis .
the fragmentation functions are related to the perturbatively calculable ratios of the longitudinal , @xmath2 , and transverse , @xmath3 , cross - sections to the total cross - section .
integrating eq .
( [ eqn - costheta - dependence ] ) over @xmath35 and respecting energy conservation for the integral over @xmath19 yields @xcite @xmath36 where @xmath37 the contribution of gluon radiation to @xmath38 has been calculated in second order of @xmath7 @xcite . at first order
the qcd correction in the total cross - section contributes only to the longitudinal part @xcite : @xmath39 where @xmath40 is the born level cross - section .
this allows tests of qcd and determinations of @xmath7 from measurements of @xmath41 and @xmath42 .
the longitudinal and transverse fragmentation functions were already investigated by the slac / lbl magnetic detector collaboration at the spear collider at @xmath43 gev @xcite , by the tasso collaboration at the petra collider at @xmath44 , @xmath45 , and @xmath46 gev @xcite , and by the opal , aleph , and delphi collaborations at the lep collider at @xmath47 @xcite .
ratios @xmath41 to derive @xmath48 were only determined by the opal and delphi collaborations @xcite . due to the sum rule eq .
( [ eqn - sumrule ] ) all details of the unknown fragmentation functions disappear from the longitudinal and transverse cross - sections up to corrections suppressed by some power of @xmath49 .
a measurement of these cross - sections at centre - of - mass energies different from @xmath50 is , therefore , indispensable to experimentally investigate the question whether power corrections to eq .
( [ eqn - sigma_tot ] ) are required and whether these are of the form @xmath49 or @xmath51 @xcite . the analysis presented in the
following determines @xmath41 , @xmath42 , and @xmath7 at an average energy of @xmath52 gev .
it uses data measured with the jade detector @xcite at the petra collider to be introduced in section [ sec - detector ] .
the measurement of the @xmath35 distribution and the investigation of the experimental systematics are detailed in section [ sec - costheta ] .
section [ sec - observables ] presents the results from the fits to the measured @xmath35 distribution .
power corrections to eq .
( [ eqn - sigma_tot ] ) are considered in section [ sec - powercorrection ] .
our final results for @xmath41 , @xmath42 , and @xmath7 are summarized in section [ sec - summary ] .
.[tab - selection - cuts ] the main cuts are listed for the selection of multihadronic events which were varied to assess systematic uncertainties ( see text ) . @xmath53 and @xmath54 are energy and 3-momentum of tracks and clusters . [ cols="<,<",options="header " , ] in this table , the contributions of the statistical uncertainties due to data and monte carlo simulation are given separately to illustrate the possible gain from a larger sample of simulated events on the total errors .
[ fig - sigmal - vs - q ] shows our result and those obtained at the z peak @xcite for the ratio of the longitudinal and total cross - section versus the centre - of - mass energy .
neither the measurement of the slac / lbl magnetic detector @xcite nor that of the tasso collaboration @xcite were considered for the study of power corrections . the results published by the slac / lbl collaboration , from which @xmath55(stat . ) at @xmath56 gev can be derived , are solely quoted with statistical uncertainties .
the tasso collaboration used for their investigations a limited the @xmath19 range of @xmath57-@xmath58 only .
longitudinal cross - section relative to the total cross - section is presented versus centre - of - mass energy .
the results obtained at the z peak are also shown .
the dotted and solid lines show , respectively , the jetset @xcite expectation for partons and stable particles .
the dashed line is the second order qcd prediction , eq .
( [ eqn - qcd - prediction ] ) , using @xmath59 @xcite .
, scaledwidth=100.0% ] even though the application of the sum rule , eq .
( [ eqn - sumrule ] ) , should absorb all the details about the fragmentation functions a large difference exists between the jetset expectations at parton and at hadron level .
this indicates a substantial hadronization correction which is expected to behave as a leading order @xmath49 power correction @xcite . for the study of the power correction to the longitudinal cross - section we used the parametrization of @xcite where the power correction is given by @xmath60 here
@xmath61 is the milan factor , @xmath62 , usually chosen to be @xmath63 gev , is the infrared matching scale of the non - perturbative term with the perturbative terms .
the non - calculable parameter @xmath64 is to be determined from a fit to the data .
the coefficient of the power correction for @xmath2 given in @xcite is @xmath65 when being adapted for the parametrization chosen in eq .
( [ eqn - powcor - formula ] ) . since the available @xmath41 data are not sufficiently precise for a detailed test of the power corrections , we solely quote for illustrative purposes the result @xmath66 and @xmath67 from fitting the second order plus power correction prediction .
this yielded : @xmath68 where the errors propagated from the total uncertainties of the @xmath41 data . with the large uncertainties and without further data no definite conclusion about the power correction for the longitudinal cross - section is possible .
the size of the power correction expected at the z peak was estimated to be @xmath69 @xcite using the measured value of @xmath41 at @xmath70 quoted in @xcite .
this paper presents the first measurement of the longitudinal and transverse cross - sections at petra energies of 35 through 44 gev .
values of @xmath71 were obtained for the longitudinal and transverse cross - sections relative to the total hadronic cross - section , where the errors are statistical and systematics .
the two results are exactly anti - correlated since the relative transverse cross - section was obtained using the relation ( [ eqn - sigma_tot ] ) .
the statistical errors are due to the data and the limited monte carlo simulation statistics which are about equal in size . using the second order qcd prediction for the relative longitudinal cross - section ,
a value of @xmath72 was determined for the strong coupling constant at the luminosity weighted average centre - of - mass energy of @xmath4 gev . evolved to the z peak using the 3-loop formula from @xcite
this result corresponds to @xmath73 with total errors , which is in agreement with the current average @xmath74 @xcite .
power corrections to the longitudinal cross - section were considered .
the available measurements at @xmath4 gev and @xmath75 gev are not sufficient yet for a definite conclusion .
m.b . thanks the members of the iii .
physikalisches institut of the rwth aachen for their generous hospitality and many helpful discussions .
t. sjstrand : comput .
* 39 * ( 1986 ) 347 , + t. sjstrand and m. bengtsson : comput .
* 43 * ( 1987 ) 367 .
+ we used version 6.3 of jetset as used by the jade collaboration with parameters from @xcite to generate the events . | an investigation of the polar angle distribution of charged hadrons is presented using data taken by the jade experiment at the petra collider at centre - of - mass energies of @xmath0 and @xmath1 gev . from fits to the polar angle distribution the longitudinal , @xmath2 , and transverse , @xmath3 , cross - section relative to the total hadronic are determined at an average energy scale of @xmath4 gev .
the results are @xmath5 where total errors are given and the results are exactly anti - correlated . using the next - to - leading order qcd prediction for the longitudinal cross - section , the value @xmath6 of the strong coupling constant is obtained in agreement with the world average value of @xmath7 evolved to an energy scale of @xmath4 gev . to * measurement of the + longitudinal and transverse cross - section + in e@xmath8e@xmath9 annihilation at @xmath10-@xmath11 gev
* m. blumenstengel@xmath12 , o. biebel@xmath12 , p.a .
movilla fernndez@xmath12 , p. pfeifenschneider@xmath13 , s. bethke@xmath12 , s. kluth@xmath12 and the jade collaboration@xmath14 submitted to phys .
lett .
b @xmath12 max - planck - institut fr physik , d-80805 mnchen , germany + contact e - mail : [email protected] + @xmath14 for a full list of members of the jade collaboration see reference @xcite + @xmath15 now at sap deutschland ag & co. kg , neurottstrae 15 , 69190 walldorf , germany |
blazars are the most luminous persistent objects in the sky ( up to @xmath3 erg s@xmath4 ; e.g. , @xcite ) .
they exhibit variability across the electromagnetic spectrum on timescales ranging from minutes to years ( e.g. , @xcite and references therein ) .
blazars constitute a sub - class of active galactic nuclei ( agns ) whose relativistic plasma jets are thought to be closely aligned to our line of sight @xcite .
the emission from blazars is predominantly non - thermal , including radio synchrotron radiation and x - ray and @xmath0-ray inverse - compton scattering .
the central engines of blazars , which are believed to be the ultimate source of this variability , are unresolved and opaque to radio waves .
hence , they are not imaged directly , even with the @xmath5 mas resolution obtained with the very long baseline array ( vlba ) at 43 ghz . despite this limitation , by monitoring and subsequently modeling flares in the high - energy emission from these agn , we can potentially gain insight into the sub - parsec - scale physics of the jets close to the central engines .
many theoretical models have been developed to explain both the spectral energy distributions ( seds ) and high - energy variability of blazars .
the seds of most blazars take the form of two humps - one in the radio / uv portion of the spectrum ( synchrotron ) and the other at higher energies in the x - ray/@xmath0-ray portion of the spectrum ( interpreted here as inverse - compton emission ) .
@xcite , @xcite , and @xcite proposed that shock acceleration within the jet could result in the observed variations in radio emission .
@xcite presented a leptonic model in which the high - energy hump of the sed is produced by comptonization of accretion disk photons by electrons contained within a plasmoid ( `` blob '' ) moving relativistically along the jet .
alternatively , @xcite formulated a hadronic model in which the high - energy hump is produced through @xmath6 pion production from relativistic protons within the jet encountering both internal and external photon fields .
@xcite also explored the effects of comptonization of external radiation fields from the accretion disk and the surrounding broad emission - line region ( blr ) by electrons within the jet .
@xcite presented an internal shock model for blazar variability in which shells of plasma within the jet collide , producing forward and reverse shocks that result in particle acceleration within the jet . while these models can explain many aspects of the radiative mechanisms potentially at work in blazars , there remain aspects of high - energy variability that are not accounted for by these theories .
recent observations obtained with the _ fermi _ large area telescope ( lat ) and the vlba have shown that a large number of @xmath0-ray flares in blazars are strongly correlated with the passage of superluminal knots through the radio cores of these objects @xcite .
the majority of these @xmath0-ray flares tend to be associated with flares detected at optical / uv wavelengths .
the high cadence of observations provided by the _ fermi _ spacecraft , however , has also allowed the identification of a small sub - set of @xmath0-flares that seem to occur in relative isolation , with little or no variability detected in the other bands .
these isolated events are termed `` orphan '' flares . in one case , @xcite present observations of an orphan @xmath0-ray flare that occurred during the passage of a superluminal knot through the inner jet of the blazar pks 1510@xmath1089 ( see figure [ fig1 ] ) . based on the orphan @xmath0-ray flare highlighted in figure [ fig1 ] , here
we develop a model of blazar variability to account for this type of behavior .
in our model , a blob consisting of a power - law distribution of electrons propagates relativistically along the jet axis of the blazar and passes through a synchrotron - emitting ring of electrons , which represents a shocked segment of a more slowly moving jet sheath ( see figure [ fig2 ] and the sketch presented in figure [ fig9 ] below ) .
the ring creates a very localized source of seed photons that are inverse - compton scattered by the electrons in the moving blob , creating an orphan @xmath0-ray flare as the blob passes through the ring .
we point out that our model is distinct from compton mirror models ( see e.g. , @xcite , @xcite , and @xcite ) that require a reflecting cloud close to the jet axis to produce @xmath0-ray flares .
the presence of a non- to mildly - relativistic sheath of plasma enshrouding the relativistic spine of a blazar jet has been discussed by several authors . from a theoretical standpoint
, @xcite investigated the radiative interplay between these two regions within the jet via the inverse - compton mechanism .
more recently , the seds of several blazars have been fit with models that utilize the presence of a jet sheath ( see , e.g. , @xcite and @xcite ) .
in addition , @xcite discuss the potential of sheaths of plasma within blazars to produce the high - energy neutrinos being detected by icecube .
several authors have also discussed what the observational signature of a sheath of plasma would be within a blazar jet ( see * ? ? ?
velocity shear between the jet and the ambient medium into which the jet propagates is expected to align the magnetic field on the outer edges of the jet with the jet axis so that the sheath polarization should be high ( tens of percent ) relative to the spine .
the electric vector position angles ( evpas ) associated with this polarization should be perpendicular to the jet axis .
this polarimetric signature has indeed been detected in a number of blazars ( see * ? ? ?
we present polarimetric observations of pks 1510@xmath1089 that point to the existence of a jet sheath within this blazar , lending support to our model of orphan @xmath0-ray flares .
this paper is organized as follows : in 2 we outline the theory relating to the radiative transfer modeled in our code . in
3 we describe the numerical simulation performed and list all of our model parameters . in
4 we compare synthetic light curves produced by the model to the observed optical and @xmath0-ray light curves of the orphan flare of pks 1510@xmath1089 . in 5 we present vlba polarimetric observations that point to the existence of a jet sheath within pks 1510@xmath1089 . in 6 we present our conclusions and summary .
we adopt the following cosmological parameters : @xmath7 , @xmath8 , and @xmath9 .
models of blazar emission , whether leptonic or hadronic , rely on the presence of an emission region ( a `` blob '' ) moving relativistically along the jet .
for the purposes of this work we adopt the leptonic scenario and treat the blob ( and ring ) as consisting of electrons with a power - law energy distribution , @xmath10 , over the range @xmath11 to @xmath12 , where the energy of a given electron is @xmath13 .
there are three main components of the non - thermal emission produced by the passage of a blob of plasma down the spine of a jet in our model .
the first is synchrotron radiation ( syn ) produced by electrons gyrating around magnetic field lines within the blob .
these electrons also inverse - compton scatter the synchrotron photons they produce , a process known as synchrotron self - compton ( ssc ) emission .
the third component is inverse - compton scattering of the external ( to the blob ) ring synchrotron photons by the blob s electrons , a process known as external compton scattering ( ec ) .
as the blob approaches the ring , the photon density in the co - moving frame of the plasma increases , resulting in an ec driven @xmath0-ray flare that peaks and decays as the blob passes through and then moves away from the ring .
the photon production rates @xmath14 for these three physical processes , @xmath15 , @xmath16 , and @xmath17 , are given below ( @xmath18 is defined as a dimensionless photon energy ) . in our radiative transfer code , we evaluate these photon production rates in the co - moving frame of the blob , after which a series of lorentz transformations is applied to obtain the resultant flux in the observer s frame . in all of the following equations ,
the dimensionless electron velocity , @xmath19 , has been set to unity .
the synchrotron photon production rate per unit volume for electrons within the blob or the ring is given by : @xmath20 ( e.g. , @xcite ) , where @xmath21 is the ratio of the observed frequency to the critical frequency ( @xmath22 ) , which is given by @xmath23 ; and @xmath24 is an approximation to a bessel function ( see * ? ? ?
the ssc photon production rate per unit volume within the blob is given by : @xmath25 ( e.g. , @xcite ) , where @xmath26 is the synchrotron photon density produced by the electrons in the co - moving frame of the blob
. the isotropic scattering cross - section , @xmath27 , is given by : @xmath28 ~ & \textrm{for $ \epsilon \leq \epsilon_{\rm s } \leq
\frac{4 \epsilon \gamma^{2}}{1 + 4 \epsilon \gamma}$ } , \end{array } \right.\end{aligned}\ ] ] where @xmath29 is the thomson cross - section , @xmath30 is the incoming photon energy , @xmath31 is the scattered photon energy , and @xmath32 , @xmath33 , and @xmath34 are defined as follows : @xmath35 where @xmath36 @xcite .
the general ec photon production rate per unit volume within the blob for the scenario presented in figure [ fig2 ] is given by : @xmath37 @xcite , where @xmath38 is the external ring photon density per unit volume within the co - moving frame of the blob , @xmath39 is the angle between the scattering electron in the blob and the seed photon from the ring in the co - moving frame of the blob , @xmath40 is the differential scattering cross - section , and @xmath41 and @xmath42 are the differential solid angles in the photon and electron directions , respectively . in order to evaluate @xmath38 in the co - moving frame of the blob
, we follow the method outlined by @xcite , and make use of the following lorentz invariants : @xmath43 ( where @xmath44 denotes the host galaxy / stationary ring frame and unprimed notation refers to the co - moving frame of the blob ) and @xmath45 , where @xmath46 is the bulk lorentz factor of the blob , @xmath47 is the blob s velocity in units of @xmath48 , and @xmath49 is the cosine of the angle that the incoming ring photons make with respect to the blob in the co - moving frame of the blob . the external ring photon density ( in @xmath50 ) evaluated in the host galaxy frame
can be shown to be : @xmath51 @xcite , where @xmath52 and @xmath53 , and where @xmath54 is the distance of the blob from the center of the ring ( see figure [ fig2 ] ) . here , @xmath55 is the photon production rate ( in @xmath56 ) of the ring due to synchrotron radiation .
if one further assumes that the ring is optically thin ; we can write @xmath57 where @xmath58 is the thickness of the ring in the @xmath54 direction , and where @xmath59 and @xmath60 denote the inner and outer radii of the ring , respectively ( see figure [ fig2 ] ) . here
, @xmath61 is the synchrotron photon production rate of the ring ( see equation ( [ eq1 ] ) ) evaluated in the host galaxy frame .
it then follows that @xmath38 can be evaluated as follows : @xmath62 lc global parameter & value + @xmath63 & @xmath64 + @xmath65 & @xmath66 + @xmath67 & 0.361 + @xmath68 & @xmath69 + @xmath70 & 1.5 ( mjy ) + @xmath71 & 6.5 ( @xmath72 ) + blob parameter & value + @xmath73 & 0.01 ( pc ) + @xmath74 & 4 + @xmath75 & 19 + @xmath76 & -0.1 ( pc ) + @xmath77 & 0.1 ( pc ) + @xmath78 & 4.0 + @xmath79 & 0.03 ( g ) + @xmath80 & @xmath81 ( @xmath82 ) + ring parameter & value + @xmath83 & 0.0 ( pc ) + @xmath84 & 0.09 ( pc ) + @xmath85 & 0.18 ( pc ) + @xmath86 & 0.018 ( pc ) + @xmath87 & 4.0 + @xmath88 & 0.12 ( g ) + a head - on approximation is adopted , in which the scattered photon direction is set equal to the electron direction in the blob frame , which then results in the scattering angle : @xmath89 @xcite , where @xmath90 , and @xmath91 , where @xmath92 is the angle between the jet axis and our line of sight ( see figure [ fig2 ] ) , and @xmath93 is the azimuthal angle about the jet axis . in order to keep the numerical integration to a minimum
, a delta - function approximation of the differential scattering cross - section in the thomson regime is adopted : @xmath94 @xcite , where @xmath95 refers to the solid angle subtended by the scattered photons , @xmath96 , and @xmath97 is a heaviside function given by : @xmath98 as discussed in @xcite , equation ( [ eq6 ] ) is a valid approximation of the scattering cross - section provided @xmath99 .
the delta - function @xmath100 in equation ( [ eq6 ] ) implies @xmath101 .
the dimensionless scattered photon energy ( @xmath31 ) is defined as @xmath102 .
the orphan flare ( presented in figure [ fig1 ] ) we model with our code occurs at an observational frequency of @xmath103 . due to the relativistic motion of the blob , the scattered photon frequency ( @xmath104 ) in the co - moving frame of the blob is related to the observed frequency by @xmath105 , where @xmath106 is a relativistic doppler boosting factor given by @xmath107 . with these quantities
defined , we compute the limiting values of @xmath108 for the range of scattered photon and electron energies modeled in our code ( see table [ tab1 ] ) , and we find : @xmath109 .
this range of @xmath108 falls well within the thomson regime . in the future
, the code will incorporate the effects of the reduced klein
nishina scattering cross - section at higher energies .
since @xmath110 , equations ( [ eq4])([eq6 ] ) can be combined to yield a simplified version of the external compton photon production rate : @xmath111 where now @xmath112 and @xmath113 . the distribution of electrons in the blob is evolved each time step by solving the fokker planck equation : @xmath114 @xcite , where @xmath115 is the electron radiative cooling rate in @xmath116 , and @xmath117 is the electron injection rate per unit dimensionless energy in @xmath118 .
in general , @xmath119 ; however , here we assume that @xmath120 because ssc is only a minor component in the seds produced by the runs of our code presented here ( see figure [ fig3 ] ) . in the future
we intend to carry out a more sophisticated calculation that includes a proper treatment of the ssc cooling , which , as pointed out by @xcite , can strongly influence the evolution of @xmath121 in time - dependent calculations of the non - thermal emission from some blazars . for simplicity , we neglect aging in the ring electrons . the synchrotron cooling rate for the blob electrons is given by : @xmath122 @xcite .
the external compton cooling rate is given by : @xmath123 @xcite , where the upper limit of the integration ensures that the cooling rate is evaluated within the thomson regime .
the injection rate @xmath117 is computed by defining an injection power : @xmath124 the injection power parameterizes a physical mechanism at work within the blob , for example , turbulence or diffusive shock acceleration , that continuously rejuvenates the aging blob electron distribution .
this injection power is a free parameter within our model and has been set at @xmath125 in the calculations presented here .
equation ( [ eq9 ] ) has the following analytical solution : @xmath126 & \textrm{for $ \xi < 1$}\\ \\
\frac { q \gamma^{-(s+1 ) } } { m_{\rm e } c^{2 } b ( s-1 ) } & \textrm{for $ \xi \ge 1$}~ , \vspace{5 mm } \end{array } \right . \vspace{5mm}\ ] ] where @xmath127 @xcite .
the initial relativistic electron number density , @xmath128 , of the blob is determined by assuming equipartition between the magnetic and electron energy densities within the blob : @xmath129 after the photon production rates have been computed for a range of photon energies at each time step , the flux is computed in the observer s frame by multiplying the sum of the production rates ( @xmath130 ) by the volume of the blob , the energy of the photons , and a @xmath131 conversion factor : @xmath132 where @xmath67 is the redshift , @xmath133 is the luminosity distance , and @xmath134 is a relativistic doppler boosting factor , given by @xmath135 .
the numerical calculations were carried out using the blaze code ( people.bu.edu/nmacdona/research.html ) . a spherical blob of radius @xmath136 accelerates from a bulk lorentz factor of @xmath137 to @xmath138 , after which it continues to propagate at constant speed along the spine of the jet .
the parameters governing the blob acceleration have been chosen to reflect the acceleration derived for the emission knot associated with the orphan @xmath0-ray flare in pks 1510@xmath1089 ( see figure [ fig1 ] ) based on the changing rate of rotation of the observed optical polarization vector ( @xcite , @xcite ) .
the blob begins its acceleration toward the ring ( located at @xmath139 in the simulation ) at @xmath140 and reaches @xmath141 at @xmath142 after passing through the ring .
the blob is tracked out to a distance of @xmath143 , after which the simulation is terminated .
although in actuality the sheath ( and by extension the ring ) will have some relatively low velocity along the jet , for the purposes of the numerical simulation , we treat the ring as stationary .
the dimensions of the ring used in the simulation , namely , its minimum ( @xmath59 ) and maximum ( @xmath60 ) radius and its thickness ( @xmath144 ) ( see figure [ fig2 ] ) , are listed in table [ tab1 ] .
the size of the sheath ( and by extension the ring ) is thought to be a substantial fraction of the cross - sectional width of the jet ( @xcite ; @xcite ) and the values chosen for @xmath59 and @xmath60 reflect the width of the jet in pks 1510@xmath1089 . since we are positing that the ring represents a shocked segment of a continuous jet sheath ( see figure [ fig9 ] below ) , the thickness of the ring ( @xmath144 ) in our model reflects the length scale over which particle acceleration acts on a small section of the sheath , perhaps as it propagated through a standing shock upstream of its current location within the jet .
while this parameter is the least constrained in our model , @xmath144 simply scales the external ring photon density up or down , depending on our choice of its value ( see equation ( [ eq5 ] ) ) .
@xcite make a case against the ability of a sheath of plasma surrounding the jet to provide sufficient numbers of seed photons to produce the orphan flares seen in pks 1510@xmath1089 ( i.e. , figure [ fig1 ] ) without the sheath itself outshining the observed flux .
we find no conflict , however , between our model and observations of this blazar .
as discussed in their paper , the observed luminosity of the sheath can be shown to be ( their equation ( c3 ) ) : @xmath145 where @xmath146 is the distance along the jet , @xmath46 is the blob s bulk lorentz factor , and @xmath147 and @xmath148 are the bulk lorentz factor and opening angle of the sheath , respectively . as @xcite point out , the strong dependence of @xmath149 on @xmath147 means that for even mildly relativistic sheaths the observed luminosity can become quite large .
the ring of sheath plasma in our model , however , is located upstream of the radio core in pks 1510@xmath1089 . in particular , the radio knot correlated with the orphan flare was traveling down the jet at @xmath150 in our frame @xcite . from figure [ fig1 ] , it is apparent that the onset of the orphan flare preceded the large optical flare associated with the knot passing the radio core by about 20 days . assuming that the knot moved at this observed rate along the jet for those 20 days
, that would place our ring @xmath151 upstream of the radio core .
if we further assume that the radio core is at roughly a scale of @xmath152 from the central engine , that would place the ring at roughly a distance of @xmath153 along the jet .
this locates our emission region between the two commonly considered regions of @xmath0-ray production in blazar jets , namely , the near - dissipation region ( @xmath154 ) , where the external photon fields are dominated by emission from a dusty torus and the blr , and the far - dissipation region ( @xmath155 ) near the core seen at 43 ghz . at this distance , utilizing the dimensions of our ring as corresponding to the jet sheath , we compute a sheath opening angle of @xmath156 ( where @xmath157 ) .
here , @xmath158 and @xmath159 refer to the inner and outer angles subtended by the ring relative to the jet apex , which we take to be @xmath160 upstream of the ring
see figure [ fig9]and can be shown to be : @xmath161 and @xmath162 , respectively .
we point out that this value for the sheath opening angle is in rough agreement with the sheath opening angle of @xmath163 used by @xcite to explain variability in the polarization of the blazar oj 287 .
if we further assume that the sheath at this scale is only mildly relativistic ( @xmath164 ) and take our final value for the bulk lorentz factor of the blob ( @xmath165 ) , equation ( [ eq16 ] ) yields @xmath166 , which is two orders of magnitude smaller than the observed @xmath0-ray luminosity for pks 1510@xmath1089 during this flaring epoch ( @xmath167 ) and is roughly a factor of 3.6 smaller than the observed synchrotron luminosity ( @xmath168 ) .
the blob and ring electron populations are both modeled as identical power - law distributions , @xmath169 , with identical power - law indices @xmath170 ( listed in table [ tab1 ] ) , in rough agreement with the observed @xmath0-ray photon index of @xmath171 ( @xmath172 ) reported in @xcite . by setting the power - law index to this value
, we are making the assumption of an unbroken power - law of electron energies in both regions of our model .
the range of electron energies over which we assume the power - law applies ( @xmath11 through @xmath12 ) is listed in table [ tab1 ] .
constant magnetic fields are taken to permeate both the blob ( @xmath79 ) and the ring ( @xmath88 ) with the values used in the simulation listed in table [ tab1 ] .
as discussed in @xcite , the jet sheath is thought to have a higher electron density than the spine .
since we assume equipartition , this implies that @xmath173 .
in addition , a lower magnetic field within the blob is needed to keep the radiative losses experienced by the blob electrons out of the fast cooling regime ( see equation ( [ eq13 ] ) ) ; however , the value used in our simulation ( @xmath174 ) is at odds with the observed value of @xmath175 for pks 1510@xmath1089 reported in @xcite .
this difference in field strength might be reconciled by the fact that , as discussed above , we treat the ring as being located upstream of the radio core in pks 1510@xmath1089 .
it is consistent for the magnetic fields in the jet upstream of the core to be smaller if indeed the core at 43 ghz is associated with a recollimation shock that amplifies the field strength through shock compression ( see figure [ fig9 ] below ) .
we also approximate that the magnetic fields within the blob and the ring do not vary appreciably on the timescale required for the blob to pass through the ring ( @xmath176 in the agn frame ) .
this requires that the opening angle of the jet @xmath177 , which agrees with @xmath178 found by @xcite .
the numerical integrations were performed using linear grids of 15 radial and azimuthal zones and logarithmic grids of 60 intervals in @xmath30 and in @xmath0 . in order to convert the resultant flux into the observer s frame ,
the jet is taken to lie at an inclination angle of @xmath179 @xcite , with a redshift of @xmath180 @xcite , and a luminosity distance of @xmath181 .
these values are based on the observed properties of pks 1510@xmath1089 and yield a projected scale of @xmath182 .
to better fit the observed light curves presented in figure [ fig1 ] , we include a baseline flux in each band , which is added to the variable emission produced by the model .
these baseline fluxes , listed in table [ tab1 ] , reflect the more slowly varying emission from the jet plasma and are approximated to be constant .
finally , the time step is chosen so that 24 samplings of the sed are made by the end of the simulation of the blob passing through the ring .
figure [ fig3 ] shows an sed produced by the model during the peak of the simulated orphan @xmath0-ray flare .
the three non - thermal emission components included in the calculation are differentiated by color .
the model broadly reproduces the observed optical and @xmath0-ray peaks of pks 1510@xmath1089 s sed ( shown in black ) , but does not reproduce the observed optical - ultraviolet excess ( known as the big blue bump ) , nor the observed x - ray flux levels . the former spectral feature is believed to be associated with emission from the accretion disk surrounding the central engine , while the latter may be ssc emission from regions farther down the jet ( as indicated by the weak correlation between the x - ray and @xmath0-ray light curves ; see @xcite ) .
neither of these components are included in the calculation , hence these spectral features are absent from the synthetic seds produced by the model .
the resultant @xmath0-ray and optical light curves produced by the blob s passage through the ring are shown in figure [ fig4 ] ( highlighted in red ) overlaid upon the observed data from @xcite . in particular , the baseline flux level ( as discussed in 3 ) is demarcated by the black dashed line in each plot , whereas the variable emission produced by the model is indicated by the black dotted line .
the superposition of these two lines in each plot creates the synthetic light - curves shown in red .
the synthetic light curves reproduce , in general , the observed orphan @xmath0-ray flare without exceeding the observed optical flux during the flare .
the lack of a strong optical flare demonstrates the model s key ability to produce orphan @xmath0-ray flares .
the peak flux in the simulated @xmath0-ray flare occurs when the blob is at a distance of @xmath183 downstream of the ring .
this is a result of the interplay between the doppler boost gained by the relativistic motion of the blob , the external ring photon density within the co - moving frame of the blob , and the effects of radiative cooling .
it should be pointed out that there is a small optical `` bump '' in our synthetic light curve ( lower panel of figure [ fig4 ] ) .
this gradual increase in emission is a result of the doppler boost gained by the blob s acceleration along the spine of the jet .
the model s ability to produce @xmath0-ray flares with steep rise and decay times ( as illustrated in the upper panel of figure [ fig4 ] ) reflects the anisotropic source of seed photons used in the simulation ( i.e. , the ring ) .
this is in contrast to other structured jet models that invoke isotropic seed photon fields from a continuous jet sheath .
the shape of the orphan @xmath0-ray flare shown in figure [ fig4 ] is very sensitive to the nature of the blob s acceleration toward and through the ring . to further explore the effect the blob s motion along
the jet has on the orphan @xmath0-ray flares produced by this model , the parameters governing the blob s acceleration along the spine of the jet were varied , namely , the blob s initial ( @xmath74 ) and final ( @xmath75 ) bulk lorentz factors . in figure
[ fig5 ] , a series of comparative light curves is shown illustrating the effect that different blob accelerations ( or lack of acceleration ) have on the orphan flares produced by this model .
it is apparent from figure [ fig5 ] that by varying the blob s acceleration through the ring , this model is able to produce orphan flares both with steeper and more gradual onsets and decays .
the differences in these light curves highlight the potential to use the observed shape of orphan @xmath0-ray flares to infer the nature of the acceleration of the emission region through blazar jets upstream of the 43 ghz core .
the presence of a sheath of plasma surrounding the spine of a blazar jet is a central component to our model of orphan @xmath0-ray flares .
several authors have discussed what the observational signature of a jet sheath within a blazar would be ( see @xcite and references therein ) .
they posit that a jet sheath represents a shear layer between the plasma of the jet and the ambient medium into which the jet propagates . in this scenario ,
the initially helical or chaotic magnetic field lines of the outer layers of the jet are `` stretched '' out due to the velocity gradient between the jet and the ambient medium .
this shear , if it exists , would result in a buildup of magnetic field on the outer edges of the jet , with the orientation of that field aligned nearly parallel to the jet boundary @xcite .
the direction of the magnetic field within a blazar is inferred from the orientation of the evpas measured by the vlba , which , barring the effects of opacity and relativistic aberration ( see * ? ? ?
* ) , are assumed to be perpendicular to the projection of the magnetic field onto the plane of the sky .
therefore , the observational signature of a jet sheath would be an increase in the fractional polarization toward the edges of the jet , with the evpa perpendicular to the jet boundary , and therefore roughly perpendicular to the jet axis .
we point out that an alternative interpretation ( as discussed in @xcite ) is that the evpa being perpendicular to the jet boundary is the result of the jet flow carrying a large scale helical magnetic field
. a measure of the faraday rotation gradient , if it exists , ( see * ? ? ?
* ) across the width of the jet of pks 1510@xmath1089 , would potentially be able to distinguish between these two scenarios .
we adopt the prior interpretation for this paper .
a complication in detecting the polarimetric signature of a jet sheath within a blazar is discussed by @xcite . due to the effects of doppler beaming ,
the relativistic spine of the jet ( @xmath184 ) is intrinsically far brighter than the surrounding sheath ( which is usually assumed to be non- to mildly - relativistic : @xmath185 ) . to overcome this vast difference in intrinsic brightness ,
several authors have implemented a method of `` stacking '' radio maps ( see , e.g. , @xcite ; @xcite ) .
this stacking method involves aligning individual radio images of a blazar based on the location of the radio core ( which is assumed to be an essentially stationary feature within the jet at a given frequency ) .
the images are then added together , with the expectation that noise and transient phenomena ( such as superluminal blobs ) will smooth out in the stacked map , and , in contrast , any standing features within the jet ( such as a jet sheath ) will be highlighted as more images are added to the stacked map . with this in mind , we stacked vlba total intensity and polarization images of pks 1510@xmath1089 at 43 ghz in order to see whether the polarimetric signature of a jet sheath is present in this blazar .
in particular , we selected images from twenty epochs of observation from 2008 to 2013 made with vlba data from the vlba - bu - blazar program ( see www.bu.edu/blazars/vlbaproject.html and @xcite for a general description of the observations and data analysis ) .
the chosen epochs correspond to periods of relatively quiescent jet activity .
our stacked map for pks 1510@xmath1089 is shown in figure [ fig6 ] .
we do see an increase in the degree of polarization toward the edges of the jet .
the evpas plotted in figure [ fig6 ] are inclined to the jet spine axis ( highlighted in red ) and are predominantly perpendicular to that axis ( as predicted by the sheath scenario ) in both the southern half and the upper northern half of the jet .
we examine ten transverse slices through the data ( shown in blue ) .
the variations of the stokes parameters along the first slice are shown in figure [ fig7 ] .
we indeed find that the fractional polarization increases toward the edges of the jet , as predicted by the sheath scenario .
these two features , namely , the increase in fractional polarization and the orientation of that polarization perpendicular to the jet axis , are consistent with a sheath in pks 1510@xmath1089 .
the variations of the stokes parameters along the other nine slices are qualitatively similar to the first , with the exception that after the fifth slice we no longer see signs of a sheath component in the southern half of the jet .
the total and polarized intensity profiles along the slices ( highlighted in figure [ fig7 ] ) are used to determine the nominal location of the sheath within the jet , as illustrated in figure [ fig8 ] . in particular
, we align the total intensity and polarized intensity profiles for each slice and use the double - peaked nature of the polarized intensity ( a signature of the jet sheath ) to highlight the location of the sheath within each slice ( the shaded gray regions shown in figure [ fig8 ] ) .
the width of the sheath on each side of the slice is determined , arbitrarily , by the location where the polarized flux falls below 0.85 times the peak value ( the solid black vertical lines in figure [ fig8 ] ) .
this percentage yields a conservative estimate of the extent of the sheath .
the total intensity of the sheath plasma on each side of the jet is approximated by the values of the total intensity profile ( the red circles in figure [ fig8 ] ) that are aligned with the peaks of the polarized intensity profile ( the dashed black vertical lines in figure [ fig8 ] ) .
we then repeat this procedure for each slice throughout our stacked map , thus tracing out the sheath ( shaded gray regions shown in figure [ fig6 ] ) .
an estimate of the bolometric luminosity ( @xmath186 ) of the sheath in the observer s frame is computed by adding all of the flux contained within the gray shaded regions of our stacked map ( figure [ fig6 ] ) .
based on the distance to pks 1510@xmath1089 , we convert this inferred sheath flux at 43 ghz into a spectral luminosity ( @xmath187 ) and assume a power - law of the form @xmath188 . after solving for the constant of proportionality ( @xmath189 )
, we then integrate to obtain : @xmath190 where we have assumed a spectral index of @xmath191 ( in rough agreement with the mean spectral index of a large sample of flat spectrum radio quasar jets found by @xcite ) and limits of integration of @xmath192 and @xmath193 .
it is interesting to note the similarity between this value and the value obtained in 3 for the upstream sheath luminosity using equation ( [ eq16 ] ) .
the above estimate indicates that the sheath within pks 1510@xmath1089 is potentially a very important source of seed photons even at parsec scales along the jet . as a further comparison , we compute the bolometric luminosity ( in the observer s frame ) of the ring of sheath plasma in our model assuming , in contrast to the simulation , that the ring moves along the jet with a bulk lorentz factor of @xmath194 .
we find @xmath195 , roughly an order of magnitude smaller than the above sheath estimate .
finally , a crude estimate of both the inner ( @xmath59 ) and outer ( @xmath60 ) radii of the sheath was made by averaging the values obtained for these parameters relative to the jet spine ( demarcated by the dashed red vertical line in figure [ fig8 ] ) for each slice through the stacked map .
we find that @xmath196 and @xmath197 .
these values are roughly a factor of @xmath198 larger than the inner and outer radii used in our model of the ring ( see table [ tab1 ] ) that we posit to exist upstream of the region highlighted in figure [ fig6 ] .
it is interesting to note that the opening angle of the sheath used in our model ( @xmath199 as discussed in 3 ) predicts that @xmath200 and @xmath201 for the sheath at a distance of @xmath202 from the central engine , which is roughly the scale where the radio core is situated in our stacked map at 43 ghz @xcite .
although the plasma in this sheath needs to move relativistically for beaming to allow us to detect it , given this transverse velocity gradient , we infer the presence of an even slower sheath upstream , where the ring in our model could be located ( as discussed in 3 and illustrated in figure [ fig9 ] ) .
an enhancement , such as a ring of shocked plasma , in this more slowly moving sheath of plasma could potentially supply the seed photons that our model requires to explain the observed orphan flares within pks 1510@xmath1089 .
we have developed a model to explain the origins of orphan @xmath0-ray flares from blazars and have applied this model to a specific orphan flare within the blazar pks 1510@xmath1089 . in this model , a shocked segment of a slow jet sheath provides a localized source of seed photons that are inverse - compton scattered by electrons contained within a blob of plasma moving relativistically along the spine of the jet .
this inverse - compton scattering results in the production of a sharply peaked orphan @xmath0-ray flare .
synthetic light curves produced by this model are able to reproduce the general features of the observed @xmath0-ray light curve during an orphan flare event associated with the ejection of a superluminal knot from the base of the jet of pks 1510@xmath1089 .
the model is also able to reproduce the double - peaked profile seen in the seds of most blazars .
this model can potentially be used to discern the nature of the motion of the blob along the spine of the jet based on the shape of the observed orphan flare .
we have stacked radio images of pks 1510@xmath1089 obtained with the vlba at 43 ghz at multiple epochs .
we find the polarimetric signature of a sheath in our stacked map , thus lending observational support to the plausibility of this model of orphan @xmath0-ray flares within blazars . in the future
, we plan to extend our modeling efforts in an attempt to reproduce orphan @xmath0-ray flares observed in the blazars 3c 273 , 3c 279 , and 4c 71.07 .
we also plan to create stacked maps of the jets of these objects to investigate whether we can detect additional spine - sheath polarimetric signatures .
funding for this research was provided by a canadian nserc pgs d2 doctoral fellowship and by nasa fermi guest investigator grants nnx11aq03 g , nnx12ao79 g , nnx12ao59 g and nnx14aq58 g .
the authors are grateful to karen williamson for supplying the multi - wavelength data used in figure [ fig3 ] and to the anonymous referee for a thorough and helpful review .
the vlba is an instrument of the national radio astronomy observatory .
the national radio astronomy observatory is a facility of the national science foundation operated under cooperative agreement by associated universities , inc .
marscher , a. p. , jorstad , s. g. , agudo , i. , macdonald , n. r. , & scott , t. l. 2012 , in econf c1111101 , fermi & jansky : our evolving understanding of agn , ed .
r. ojha , d. thompson , & c. d. dermer ( arxiv:1204.6707 ) | blazars exhibit flares across the electromagnetic spectrum . many @xmath0-ray flares
are highly correlated with flares detected at optical wavelengths ; however , a small subset appears to occur in isolation , with little or no variability detected at longer wavelengths .
these `` orphan '' @xmath0-ray flares challenge current models of blazar variability , most of which are unable to reproduce this type of behavior .
we present numerical calculations of the time - variable emission of a blazar based on a proposal by @xcite to explain such events . in this model ,
a plasmoid ( `` blob '' ) propagates relativistically along the spine of a blazar jet and passes through a synchrotron - emitting ring of electrons representing a shocked portion of the jet sheath .
this ring supplies a source of seed photons that are inverse - compton scattered by the electrons in the moving blob .
the model includes the effects of radiative cooling , a spatially varying magnetic field , and acceleration of the blob s bulk velocity .
synthetic light curves produced by our model are compared to the observed light curves from an orphan flare that was coincident with the passage of a superluminal knot through the inner jet of the blazar pks 1510@xmath1089 .
in addition , we present very long baseline array polarimetric observations that point to the existence of a jet sheath in pks 1510@xmath1089 , thus providing further observational support for the plausibility of our model .
an estimate of the bolometric luminosity of the sheath within pks 1510@xmath1089 is made , yielding @xmath2 .
this indicates that the sheath within pks 1510@xmath1089 is potentially a very important source of seed photons .
= 1 |
let @xmath4 and @xmath5 be random vectors in @xmath1 .
we ask the following question : if the distributions of @xmath4 and @xmath5 are close in certain sense , can we guarantee that their differential entropies are close as well ?
for example , one can ask whether @xmath6 one motivation comes from multi - user information theory , where frequently one user causes interference to the other and in proving the converse one wants to replace the complicated non - i.i.d .
interference by a simpler i.i.d . approximation . as a concrete example
, we consider the so - called `` missing corner point '' problem in the capacity region of the two - user gaussian interference channels ( gic ) @xcite . perhaps due to the explosion in the number of interfering radio devices , this problem has attracted renewed attention recently @xcite . for further information on capacity region of gic and especially the problem of corner points
, we refer to a comprehensive account just published by igal sason @xcite .
mathematically , the key question for settling `` missing corner point '' is the following : given independent @xmath7-dimensional random vectors @xmath8 with the latter two being gaussian , is it true that @xmath9 to illustrate the nature of the problem , we first note that the answer to is in fact negative as the counterexample of @xmath10 and @xmath11 demonstrates , in which case the divergence is @xmath12 but the differential entropies differ by @xmath13 .
therefore even for very smooth densities the difference in entropies is not controlled by the divergence .
the situation for discrete alphabets is very similar , in the sense that the gap of shannon entropies can not be bounded by divergence in general ( with essentially the same counterexample as that in the continuous case : @xmath4 and @xmath5 being uniform on one and two hamming spheres respectively ) .
the rationale of the above discussion is two - fold : a ) certain regularity conditions of the distributions must be imposed ; b ) distances other than kl divergence might be more suited for bounding the entropy difference .
correspondingly , the main contribution of this paper is the following : under suitable regularity conditions , the difference in entropy ( in both continuous and discrete cases ) can in fact be bounded by the _ wasserstein distance _ , a notion originating from optimal transportation theory which turns out to be the main tool of this paper .
we start with the definition of the wasserstein distance on the euclidean space .
given probability measures @xmath14 on @xmath1 , define their @xmath15-wasserstein distance ( @xmath16 ) as @xmath17)^{1/p } , \label{eq : wp}\ ] ] where @xmath18 denotes the euclidean distance and the infimum is taken over all couplings of @xmath19 and @xmath20 , i.e. , joint distributions @xmath21 whose marginals satisfy @xmath22 and @xmath23 .
the following dual representation of the @xmath24 distance is useful : @xmath25 similar to , it is easy to see that in order to control @xmath26 by means of @xmath27 , one necessarily needs to assume some regularity properties of @xmath28 and @xmath29 ; otherwise , choosing one to be a fine quantization of the other creates infinite gap between differential entropies , while keeping the @xmath30 distance arbitrarily small .
our main result in shows that under moment constraints and certain conditions on the densities ( which are in particular satisfied by convolutions with gaussians ) , various information measures such as differential entropy and mutual information on @xmath1 are in fact @xmath31-lipschitz continuous with respect to the @xmath30-distance .
these results have natural counterparts in the discrete case where the euclidean distance is replaced by hamming distance ( ) .
furthermore , _ transportation - information inequalities _ , such as those due to marton @xcite and talagrand @xcite , allow us to bound the wasserstein distance by the kl divergence ( see , e.g. , @xcite for a review ) .
for example , talagrand s inequality states that if @xmath32 , then @xmath33 where @xmath34 denotes the maximal singular value of @xmath35 . invoking ( [ eq : talagrand ] ) in conjunction with the wasserstein continuity of the differential entropy
, we establish ( [ eq : costa_req ] ) and prove a new outer bound for the capacity region of the two - user gic , finally settling the missing corner point in @xcite . see for details .
one interesting by - product is an estimate that goes in the reverse direction of ( [ eq : talagrand ] ) .
namely , under regularity conditions on @xmath19 and @xmath20 we have , denote @xmath36 if @xmath37 is at most some universal constant . ]
@xmath38 see and in the next section .
we want to emphasize that there are a number of estimates of the form @xmath39 where @xmath40 are independent of a standard gaussian vector @xmath41 , cf .
* chapter 9 , remark 9.4 ) .
the key difference of these estimates from ( [ eq : dgg ] ) is that the @xmath30 distance is measured _ after _ convolving with @xmath42 .
[ [ notations ] ] notations + + + + + + + + + throughout this paper @xmath43 is with respect to an arbitrary base , which also specifies the units of differential entropy @xmath44 , shannon entropy @xmath45 , mutual information @xmath46 and divergence @xmath47 .
the natural logarithm is denoted by @xmath48 .
the norm of @xmath49 is denoted by @xmath50 . for random variables @xmath4 and @xmath51 , let @xmath52 denote their independence .
we say that a probability density function @xmath15 on @xmath53 is @xmath54-regular if @xmath55 and @xmath56 notice that in particular , regular density is never zero and furthermore @xmath57 therefore , if @xmath4 has a regular density and finite second moment then @xmath58 + { c_1\over2 } \ee[\|x\|^2 ] < \infty\,.\ ] ] [ prop : ppr ] let @xmath59 and @xmath60 be random vectors with finite second moments . if @xmath60 has a @xmath54-regular density @xmath61 , then there exists a coupling @xmath62 , such that @xmath63 \le \delta \,,\ ] ] where @xmath64 } + \frac{c_1}{2}\sqrt{\ee[\|v\|^2]}+c_2\big ) w_2(p_{u } , p_{v})\,.\ ] ] consequently , @xmath65 if both @xmath59 and @xmath60 are @xmath54-regular , then @xmath66 first notice : @xmath67 where follows from cauchy - schwartz inequality and the @xmath54-regularity of @xmath61 .
taking expectation of ( [ eq : ppr1a ] ) with respect to @xmath68 distributed according to the optimal @xmath30-coupling of @xmath69 and @xmath70 and then applying cauchy - schwartz and triangle inequality for @xmath71-norm , we obtain . to show ( [ eq : ppr2 ] ) notice that by finiteness of second moment @xmath72 .
if @xmath73 then there is nothing to prove .
so assume otherwise , then in identity @xmath74\ ] ] all terms are finite and hence ( [ eq : ppr2 ] ) follows . clearly , ( [ eq : ppr2 ] ) implies ( [ eq : ppr3 ] ) ( when applied with @xmath59 and @xmath60 interchanged ) .
finally , for ( [ eq : ppr4 ] ) just add the identity ( [ eq : ppr2a ] ) to itself with @xmath59 and @xmath60 interchanged to obtain @xmath75 + \ee\left[\log { p_{u}(u)\over p_{u}(v)}\right]\ ] ] and estimate both terms via ( [ eq : ppr1 ] ) .
the key question now is what densities are regular .
it turns out that convolution with sufficiently smooth density , such as gaussians , produces a regular density .
[ prop : pqr ] let @xmath76 where @xmath77 and @xmath78 < \infty$ ] .
then the density of @xmath60 is @xmath54-regular with @xmath79 and @xmath80 $ ] .
first notice that whenever density @xmath81 of @xmath41 is differentiable and non - vanishing , we have : @xmath82}{p_v(v ) } = \ee[\nabla \log p_{z}(v - b)|v = v]\,,\ ] ] where @xmath83 $ ] is the density of @xmath60 . for @xmath84
, we have @xmath85 so the proof is completed by showing @xmath86 \le 3\|v\| + 4\ee[\|b\|]\,.\ ] ] for this , we mirror the proof in ( * ? ? ? * lemma 4 ) .
indeed , we have @xmath87 & = \ee{\left [ \|b - v\| { p_{z}(b - v)\over p_{v}(v ) } \right ] } \\ & \le 2\ee[\|b - v\|1\{a(b , v)\le 2\ } ] + \ee[\|b - v\|a(b , v ) 1\{a(b , v)>2\}]\ , , \label{eq : pqr1}\end{aligned}\ ] ] where we denoted @xmath88 next , notice that @xmath89 thus since @xmath90 = p_{v}(v)$ ] we have an upper bound for the second term in ( [ eq : pqr1 ] ) as follows @xmath91 \le \sqrt{2}\sigma \sqrt{\ln^+{1\over(2\pi\sigma^2)^{{n\over2 } } 2 p_{v}(v)}}\,,\ ] ] where @xmath92 . from markov inequality
we have @xmath93 \ge 1/2 $ ] and therefore @xmath94)^2\over 2\sigma^2}}\,.\ ] ] using this estimate in ( [ eq : pqr2 ] ) we get @xmath95 \le \|v\| + 2 \ee[\|b\|]\,.\ ] ] upper - bounding the first term in ( [ eq : pqr1 ] ) by @xmath96 + 2\|v\|$ ] we finish the proof of ( [ eq : pqr0 ] ) .
another useful criterion for regularity is the following : if @xmath97 has @xmath54-regular density and @xmath98 satisfies @xmath99 then @xmath100 has @xmath101-regular density . apply ( [ eq : pqrx ] ) and the estimate : @xmath102 \le c_1(\|v\| + \sqrt{np } ) + c_2 . \qedhere\ ] ] as a consequence of regularity , we show that when smoothed by gaussian noise , mutual information , differential entropy and divergence are lipschitz with respect to the @xmath30-distance under average power constraints : [ cor : w2lip ] assume that @xmath103 , with @xmath104 , \ee[\|{{\tilde{x}}}\|^2 ] \leq { np } $ ] and @xmath105 .
then @xmath106 where @xmath107 .
since @xmath108 \leq \sqrt{np}$ ] , by proposition [ prop : pqr ] , the densities of @xmath109 and @xmath110 are both @xmath111-regular .
the desired statement then follows from applying ( [ eq : ppr3])- ( [ eq : ppr4 ] ) to @xmath112 and @xmath113 .
the lipschitz constant @xmath31 is order - optimal as the example of gaussian @xmath4 and @xmath114 with different variances ( one of them could be zero ) demonstrates .
the linear dependence on @xmath30 is also optimal . to see this ,
consider @xmath115 and @xmath116 in one dimension .
then @xmath117 and @xmath118 , as @xmath119 . in fact , to get the best constants for applications to interference channels it is best to forgo the notion of regular density and deal directly with ( [ eq : pqrx ] ) .
indeed , when the inputs has bounded norms , the next result gives a sharpened version of what can be obtained by combining with [ prop : pqr ] .
[ prop : best ] let @xmath120 satisfying and @xmath121 be independent .
let @xmath122 .
then for any @xmath59 , @xmath123 - \ee[\|v\|^2 ] + 2 \sqrt{np } w_1(p_{u } , p_{v})\right)\,.\label{eq : bd_best}\end{aligned}\ ] ] plugging gaussian density @xmath124 into ( [ eq : pqrx ] ) we get @xmath125 where @xmath126 = \frac{{\mathbb{e}}[b p_g(v - b)]}{{\mathbb{e}}[p_g(v - b)]}$ ] satisfies @xmath127 since @xmath128 almost surely .
next we use @xmath129 taking expectation of the last equation under the @xmath24-optimal coupling and in view of , we obtain . to get slightly better constants in one - sided version of ( [ eq : tpr1 ] ) we apply proposition [ prop : best ] : [ cor : best ] let @xmath130 be independent , with @xmath131 , @xmath132 and @xmath120 satisfying ( [ eq : power_as ] ) .
then for every @xmath133 $ ] we have : @xmath134 + 2 { \left \langle { \mathbb{e}}[a ] , { \mathbb{e}}[b ] \right\rangle}- \ee[\|g\|^2]\right ) + { \sqrt{2 n p ( \sigma_g^2 + c^2\sigma_z^2)\log e}\over \sigma_g^2 + \sigma_z^2 } \sqrt { d(p_{a + c z } \| p_{g + c z})}\end{aligned}\ ] ] first , notice that by definition wasserstein distance is non - increasing under convolutions , i.e. , @xmath135 . since @xmath136 and gaussian distribution is stable
, we have @xmath137 which , in turn , can be bounded via talagrand s inequality by @xmath138 from here we apply proposition [ prop : best ] with @xmath139 replaced by @xmath140 ( and @xmath141 by @xmath142 ) .
in this section we assume that all random variables have bounded norms : @xmath143 \simleq n\,.\ ] ] first , an interesting question is whether for @xmath144 we can prove @xmath145 the answer is negative .
a counterexample can be given when @xmath10 and @xmath11 .
then divergence is @xmath146 but the differential entropies differ by @xmath13 .
so even for very smooth densities the difference in entropies is not controlled by the divergence . however , it turns out that @xmath30 distance does work : mutual information , differential entropy and divergence are @xmath31-lipschitz with respect to @xmath30 : @xmath147 notice that @xmath148 \simleq \sqrt{n}\ ] ] and thus by proposition [ prop : pqr ] the density of @xmath109 is @xmath149-regular
. then the statement follows from ( [ eq : ppr3])- ( [ eq : ppr4 ] ) .
the assumption of regularity is not superfluous . indeed , to have inequality of the sort @xmath150 one necessarily needs to assume some regularity properties of @xmath151 and @xmath152 .
otherwise , choosing one to be a fine quantization of another creates infinite gap between differential entropies , while keeping the @xmath30 distance arbitrarily small .
the lipschitz constant @xmath31 is order - optimal as the example of gaussian @xmath4 and @xmath114 with different variances ( one of them could be zero ) demonstrates .
let @xmath153 and @xmath41 be jointly independent and @xmath154 iid gaussian .
then @xmath155 by talagrand s transportation inequality @xcite : @xmath156 thus , by coupling we also have @xmath157 then , the statement follows from ( [ eq : tpr1 ] ) .
consider the two - user gaussian interference channel ( gic ) : @xmath158 with @xmath159 , @xmath160 and a power constraint on the @xmath7-letter codebooks : either @xmath161 or @xmath162 \le { np_1 } , \quad { \mathbb{e}}[\|x_2\|^2 ] \le np_2 .
\label{eq : power2}\ ] ] denote by @xmath163 the capacity region of the gic . as an application of the results developed in , we prove an outer bound for the capacity region .
[ thm : corner2 ] let @xmath164 .
let @xmath165 and @xmath166 .
assume the almost sure power constraint .
then for any @xmath167 and @xmath168 , any rate pair @xmath169 satisfies @xmath170 where @xmath171 assume the average power constraint . then holds with @xmath172 replaced by @xmath173 consequently , in both cases , @xmath174 implies that @xmath175 where @xmath176 as @xmath177 . without loss of generality ,
assume that all random variables have zero mean .
first of all , setting @xmath178 ( which is equivalent to granting the first user access to @xmath179 ) will not shrink the capacity region of the interference channel . therefore to prove the desired outer bound it suffices to focus on the following z - interference channel henceforth : @xmath180 let @xmath181 be @xmath7-dimensional random variables corresponding to the encoder output of the first and second user , which are uniformly distributed on the respective codebook . for @xmath182 define @xmath183 by fano s inequality there is no difference asymptotically between this definition of rate and the operational one .
define the entropy - power function of the @xmath184-codebook : @xmath185 we know the following general properties of @xmath186 : * @xmath187 is monotonically increasing . *
@xmath188 ( since @xmath184 is uniform over the codebook ) .
* @xmath189 ( since @xmath190 by entropy power inequality ) .
* @xmath186 is concave ( costa s entropy power inequality @xcite ) .
* @xmath191 ( gaussian maximizes differential entropy ) .
we can then express @xmath192 in terms of the entropy power function as @xmath193 it remains to upper bound @xmath194 .
note that @xmath195 and therefore @xmath196 where @xmath197 is defined in .
this in conjunction with the slope property @xmath198 yields @xmath199 which , in view of , yields the first part of the bound . to obtain the second bound ,
let @xmath200 .
using @xmath201 \leq n p_2 $ ] and @xmath202 , we obtain @xmath203 that is , @xmath204 furthermore , @xmath205 note that the second term is precisely @xmath206 .
the first term can be bounded by applying and with @xmath207 , @xmath208 , @xmath209 and @xmath210 : @xmath211 combining yields @xmath212 where @xmath172 is defined in . from the concavity of @xmath186 and ( [ eq : ct2 ] ) @xmath213
where @xmath214 . in view of ,
upper bounding @xmath215 in ( [ eq : ct5 ] ) via ( [ eq : ct3 ] ) we get after some simplifications the second part of ( [ eq : corner2 ] ) .
the outer bound for average power constraint follows analogously with replaced by below : by , the density of @xmath216 is @xmath217-regular . applying to , we have @xmath218 , where @xmath219 again using the fact that @xmath30 distance is non - decreasing under convolutions and invoking talagrand s inequality , we have @xmath220 which yields @xmath221 this yields the outer bound with @xmath222 defined in . finally , in both cases , when @xmath223 , we have @xmath224 and @xmath225 and hence from @xmath226 .
the first part of the bound ( [ eq : corner2 ] ) coincides with sato s outer bound @xcite and ( * ? ? ?
* theorem 2 ) by kramer , which ( * ? ? ?
* theorem 2 ) was obtained by reducing the z - interference channel to the degraded broadcast channel ; the second part of ( [ eq : corner2 ] ) is new , which settles the missing corner point of the capacity region ( see for discussions ) .
note that our estimates on @xmath194 in the proof of theorem [ thm : corner2 ] are tight in the sense that there exists a concave function @xmath186 satisfying the listed general properties , estimates ( [ eq : ct2 ] ) and ( [ eq : ct3 ] ) as well as attaining the minimum of ( [ eq : ct4 ] ) and ( [ eq : ct5 ] ) at @xmath194 . hence
, tightening the bound via this method would require inferring more information about @xmath186 .
the outer bound relies on costa s epi . to establish the second statement about corner point ,
it is sufficient to invoke the concavity of @xmath227 ( * ? ? ?
* corollary 1 ) , which is strictly weaker than costa s epi .
[ rmk : igamma ] the outer bound is evaluated on fig .
[ fig : eval ] for the case of @xmath178 ( z - interference ) , where we also plot ( just for reference ) the simple han - kobayashi inner bound for the z - gic attained by choosing @xmath228 with @xmath229 jointly gaussian .
this achieves rates : @xmath230 for more sophisticated han - kobayashi bounds see @xcite . establishes the location of the upper corner point , as conjectured by costa @xcite .
the bottom corner point has been established by sato @xcite . ]
the two corner points of the capacity region are defined as follows : @xmath231 where @xmath232 . as a corollary , completes the picture of the corner points for the capacity region of gic for all values of @xmath233 under the average power constraint .
we note that the new result here is the proof of @xmath234 for @xmath235 and @xmath167 .
the interpretation is that if one user desires to achieve its own interference - free capacity , then the other user must guarantee that its message is decodable at both receivers .
the achievability of this corner point was previously known , while the converse was previously considered by costa @xcite but with a flawed proof , as pointed out in @xcite .
the high - level difference between our proof and that of @xcite is the replacement of pinsker s inequality by talagrand s and the use of a coupling argument .
below we present a brief account of the corner points in various cases ; for an extensive discussion see @xcite .
we start with a few simple observations about the capacity region @xmath236 : * any rate pair satisfying the following belongs to @xmath236 : @xmath237 which corresponds to the intersection of two gaussian multiple - access ( mac ) capacity regions , namely , @xmath238 and @xmath239 .
these rate pairs correspond to the case when each receiver decodes both messages . * for @xmath240 and @xmath241 ( strong interference ) the capacity region
is known to coincide with @xcite .
so , without loss of generality we assume @xmath242 henceforth .
* replacing either @xmath243 or @xmath244 with zero can only enlarge the region ( genie argument ) . * if @xmath245 then for any @xmath246 we have @xcite @xmath247 this follows from the observation that in this case @xmath248 , since conditioned on @xmath184 , @xmath249 is a noisier observation of @xmath179 than @xmath250 . for the top corner
, we have the following : @xmath251 note that for any @xmath167 , @xmath252 is discontinuous as @xmath253 . to verify ( [ eq : bt2 ] ) we consider each case separately : 1 .
for @xmath254 the converse bound follows from .
for achievability , we consider two cases .
when @xmath255 , we have @xmath256 and therefore treating interference @xmath179 as noise at the first receiver and using a gaussian mac - code for @xmath257 works . for @xmath258 ,
the achievability follows from the mac inner bound .
note that since @xmath259 , a gaussian mac - code that works for @xmath257 will also work for @xmath260 .
alternatively , the achievability also follows from han - kobayashi inner bound ( see , e.g. , ( * ? ? ?
* theorem 6.4 ) with @xmath261 for @xmath262 and @xmath263 for @xmath264 ) .
2 . for @xmath265 and @xmath266
the converse is obvious , while for achievability we have that @xmath267 and therefore @xmath179 is decodable at @xmath250 .
3 . for @xmath265 and @xmath268
the converse is ( [ eq : bt1 ] ) and the achievability is just the mac code @xmath260 with rate @xmath269 .
4 . for @xmath265 and @xmath270
the result follows from the treatment of @xmath271 below by interchanging @xmath272 and @xmath273 .
the bottom corner point is given by the following : @xmath274 which is discontinuous as @xmath275 for any fixed @xmath276 $ ] .
we treat each case separately : 1 .
the case of @xmath277 is due to sato @xcite ( see also ( * ? ? ?
* theorem 2 ) ) .
the converse part also follows from ( for @xmath265 there is nothing to prove ) . for the achievability
, we notice that under @xmath278 we have @xmath279 and thus @xmath179 at rate @xmath277 can be decoded and canceled from @xmath250 by simply treating @xmath184 as gaussian noise ( as usual , we assume gaussian random codebooks ) .
thus the problem reduces to that of @xmath178 .
for @xmath178 , the gaussian random coding achieves the claimed result if the second receiver treats @xmath184 as gaussian noise .
2 . the converse follows from ( [ eq : bt1 ] ) and for the achievability we use the gaussian mac - code @xmath260 and treat @xmath184 as gaussian interference at @xmath249 .
if @xmath280 $ ] , we apply results on @xmath281 in by interchanging @xmath282 and @xmath273 .
fix a finite alphabet @xmath283 and an integer @xmath7 . on the product space
@xmath284 we define the hamming distance @xmath285 and consider the corresponding wasserstein distance @xmath24 .
in fact , @xmath286 is known as ornstein s @xmath3-distance @xcite , namely , @xmath287 , \label{eq : dbar}\ ] ] where the infimum is taken over all couplings @xmath21 of @xmath19 and @xmath20 . for @xmath288 ,
this coincides with the total variation , which is also expressible as @xmath289 for @xmath14 on @xmath290 . for a pair of distributions @xmath14 on @xmath284 we may ask the following questions : 1 .
does @xmath291 control the entropy difference @xmath292 ?
2 . does @xmath293 control the entropy difference @xmath292 ?
recall that in the euclidean space the answer to both questions was negative unless the distributions satisfy certain regularity conditions . for discrete alphabets
the answer to the first question is still negative in general ( see for a counterexample ) ; nevertheless , the answer to the second one turns out to be positive : [ prop : dew ] let @xmath19 and @xmath20 be distributions on @xmath284 and let @xmath294 then @xmath295 in fact , the statement holds for any translation - invariant distance @xmath296 on @xmath290 extended additively to @xmath297 , i.e. , @xmath298 for any @xmath299 . indeed ,
define @xmath300 \le n s \right\}\,,\ ] ] where @xmath301 is an arbitrary fixed string .
it is easy to see that @xmath302 is concave since @xmath303 is .
furthermore , writing @xmath304 and applying chain - rule for entropy we get @xmath305 thus , letting @xmath306 be distributed according to the @xmath3-optimal coupling of @xmath19 and @xmath20 , we get @xmath307)\right ] \label{eq : jpt1}\\ & \le n f_n(\bar d(p , q))\,,\label{eq : jpt2}\end{aligned}\ ] ] where ( [ eq : jpt1 ] ) is by definition of @xmath308 and ( [ eq : jpt2 ] ) is by jensen s inequality .
finally , for the hamming distance we have @xmath309 by fano s inequality .
notice that the right - hand side of ( [ eq : jpt ] ) behaves like @xmath310 when @xmath293 is small .
this super - linear dependence is in fact sharp . and choose @xmath19 to be the output distribution of the optimal lossy compressor for @xmath20 at average distortion @xmath311 . by definition , @xmath312 . on the other hand , @xmath313 as @xmath314 and hence @xmath315 , which asymptotically meets the upper bound with equality . ] nevertheless , if certain regularity of distributions is assumed , the estimate can be improved to be linear in @xmath293 .
the next result is the analog of in the discrete space .
we formulate it in a form convenient for applications in multi - user information theory .
[ prop : dbar ] let @xmath316 be a two - input blocklength-@xmath7 memoryless channel , namely @xmath317 where @xmath318 is a stochastic matrix and @xmath319 .
let @xmath320 be independent @xmath7-dimensional discrete random vectors .
let @xmath51 and @xmath321 be the outputs generated by @xmath322 and @xmath323 , respectively
. then @xmath324 \label{eq : dbari}\end{aligned}\ ] ] where @xmath325 & ~\eqdef \sum_{x\in { \ensuremath{\mathcal{x}}}^n } p_x(x ) \bar d(p_{y|x = x } , p_{{{\tilde{y}}}|x = x } ) . \ ] ] given any stochastic matrix @xmath139 , define @xmath326 .
recall the following fact from ( * ? ? ?
( 58 ) ) about mixtures of product distributions : let @xmath59 and @xmath60 be @xmath7-dimensional discrete random vector connected by a product channel , that is , @xmath327 . then the mapping @xmath328 is @xmath329-lipschitz with respect to the hamming distance , where @xmath330 } l(p_{v_i|u_i})$ ] . consider another pair @xmath331 connected by the same channel , i.e. , @xmath332 .
then lipschitz continuity implies that @xmath333 \le l \ee[d_h(v,{{\tilde{v}}})]$ ] for any coupling @xmath334 .
optimizing over the coupling and in view of , we obtain @xmath335 \le
l n \bar d(p_{v } , p_{{{\tilde{v}}}})\,.\ ] ] repeating the proof of ( [ eq : ppr2]) ( [ eq : ppr4 ] ) , we have @xmath336 applying and to @xmath51 and @xmath337 gives and with @xmath338 defined in .
to bound the mutual information , we first notice @xmath339 applying ( [ eq : hstab ] ) conditioned on @xmath340 we get @xmath341 where @xmath342 } \max_{y , y',a } \log \frac{w(y|x_j , a)}{w(y'|x_j , a)}$ ] . note that @xmath343 for any @xmath344 , averaging over @xmath345 gives @xmath346\,.\ ] ] from the convexity of @xmath347 , which holds for any wasserstein distance , we have @xmath348 $ ] and so the left - hand side of ( [ eq : rgf1 ] ) also bounds @xmath349 from above . in this section we discuss how previous bounds ( and [ prop : dbar ] ) in terms of the @xmath350-distance can be converted to bounds in terms of kl divergence .
this is possible when @xmath20 is a product distribution , thanks to marton s transportation inequality ( * ? ? ?
* lemma 1 ) .
we formulate this together with a few other properties of the @xmath351-distance in the following lemma proved in .
[ lmm : dbar ] 1 .
( marton s transportation inequality @xcite ) : for any pair of distributions @xmath19 and @xmath352 on @xmath284 , @xmath353 2 .
( tensorization ) @xmath354 .
3 . ( contraction ) for @xmath21 and @xmath355 such that @xmath356 , @xmath357 } { \eta_{\rm tv}}(p_{y_i|x_i } ) \bar{d}(p_x , q_x ) .
\label{eq : dbar - contract}\ ] ] where @xmath358 is dobrushin s contraction coefficient of a markov kernel @xmath97 defined as @xmath359 .
if we assume that @xmath360 for some small @xmath361 , then combining and gives @xmath362 where the right - hand side behaves as @xmath363 when @xmath177 . this estimate has a one - sided improvement ( here again @xmath20 must be a product distribution ) : @xmath364 ( see @xcite for @xmath288 and ( * ? ? ?
* appendix h ) for the general case ) . switching to the setting in ,
let us consider the case where @xmath365 has i.i.d .
components , i.e. , @xmath366 . define @xmath367 which is the maximal dobrushin contraction coefficients among all channels @xmath368 indexed by @xmath369
. then @xmath370 \leq { \eta_{\rm tv}}\bar d(p_a , p_{{\tilde{a } } } ) \le { \eta_{\rm tv}}\sqrt{\frac{d(p_{a } \| p_{{{\tilde{a}}}})}{2 n \log e}}\ , , \label{eq : dbar - etatv}\ ] ] where the left inequality is by convexity of the @xmath3-distance as a wasserstein distance , the middle inequality is by , and the right inequality is via ( [ eq : marton1 ] ) .
an alternative to the estimate is the following : @xmath371 \label{eq : gb1}\\ & \le { \mathbb{e}}{\left [ \sqrt{{1 \over2n\log e } d(p_{y|x } \| p_{{{\tilde{y}}}|x } ) } \right ] } \label{eq : gb2}\\ & \le \sqrt{{1\over2n\log e } d(p_{y|x } \| p_{{{\tilde{y}}}|x}|p_x ) } \label{eq : gb3}\\ & \le \sqrt{{1\over2n\log e } { \eta_{\rm kl}}d(p_{a } \| p_{{{\tilde{a } } } } ) } \label{eq : gb4}\end{aligned}\ ] ] where ( [ eq : gb2 ] ) is by ( [ eq : marton1 ] ) since @xmath372 is a product distribution as @xmath365 has a product distribution , ( [ eq : gb3 ] ) is by jensen s inequality , and ( [ eq : gb4 ] ) is by the tensorization property of the strong data - processing constant for divergence @xcite : @xmath373 to conclude , in the regime of @xmath374 for some small @xmath361 our main yields @xmath375 matching the behavior of ( [ eq : chang ] ) .
however , the estimate is stronger , because a ) it is two - sided and b ) @xmath376 can be a mixture of product distributions ( since @xmath4 in may be arbitrary ) . in order to apply to determine corner points of capacity regions of discrete memoryless interference channels ( dmic )
we will need an auxiliary tensorization result .
this result appears to be a rather standard exercise for degraded channels and so we defer the proof to . [
prop : tensoh ] given channels @xmath377 and @xmath378 on finite alphabets , define @xmath379 then the following hold : 1 .
( property of @xmath380 ) the function @xmath381 is concave , non - decreasing and @xmath382 . furthermore , @xmath383 for all @xmath384 , provided that @xmath378 and @xmath377 satisfy @xmath385 and @xmath386 respectively .
2 . ( tensorization ) for any blocklength-@xmath7 markov chain @xmath387 , where @xmath388 and @xmath389 are @xmath7-letter memoryless channels , we have @xmath390 neither of the sufficient condition and for strict inequality is superfluous , as can be seen from the example @xmath391 and @xmath392 , respectively ; in both cases @xmath393 .
the important consequence of is the following implication : and @xmath394 we have @xmath395 this also follows from the concavity of @xmath396 . ] [ cor : tensox ] let @xmath397 , where the memoryless channels @xmath377 and @xmath378 of blocklength @xmath7 satisfy the conditions and .
then there exists a continuous function @xmath398 satisfying @xmath399 , such that for all @xmath7 @xmath400 by , we have @xmath383 for all @xmath384 .
this together with the concavity of @xmath380 implies that @xmath401 is convex , strictly increasing and strictly positive on @xmath402 .
define @xmath403 as the inverse of @xmath401 , which is increasing and concave and satisfies @xmath399 . since @xmath404 , the tensorization result yields @xmath405 i.e. , @xmath406 , where @xmath407
. then @xmath408 by definition , completing the proof .
we are now ready to state a non - trivial example of corner points for the capacity region of dmic .
the proof strategy mirrors that of , with and costa s epi replaced by and , respectively .
[ thm : corner - discrete ] consider the two - user dmic : @xmath409 where @xmath410 , @xmath411 are independent and @xmath412 is i.i.d . for some non - uniform @xmath413 containing no zeros .
the maximal rate achievable by user 2 is @xmath414 at this rate the maximal rate of user 1 is @xmath415 as an example , consider @xmath416 $ ] where @xmath417 .
then the maximum in is achieved by @xmath418 $ ] .
therefore @xmath419 and @xmath420 , where @xmath421 $ ] .
note that in the case of @xmath422 , where is not applicable , we simply have @xmath423 and @xmath424 since @xmath425 .
therefore the corner point is discontinuous in @xmath172
. continues to hold even if cost constraints are imposed .
indeed , if @xmath426 is required to satisfy @xmath427 for some cost function @xmath428 .
then the maximum in and is taken over all @xmath20 such that @xmath429 \leq b$ ] . note that taking @xmath430 is equivalent to dropping the constraint @xmath431 in . in this case , @xmath432 which can be shown by a simpler argument not involving .
we start with the converse . given a sequence of codes with vanishing probability of error and rate pairs @xmath433 , where @xmath434
, we show that @xmath435 , where @xmath436 as @xmath177 .
let @xmath437 be the maximizer of , i.e. , the capacity - achieving distribution of the channel @xmath438 .
let @xmath439 be distributed according to @xmath440 .
then @xmath441 , where @xmath442 . by fano
s inequality , @xmath443 that is , @xmath444 since @xmath445 is a product distribution , marton s inequality yields @xmath446 applying in and in view of the translation invariance of the @xmath350-distance , we obtain @xmath447 \nonumber \\ = & ~ 2 c n \bar{d}(p_{x_2+z_2 } , p_{{{\tilde{x}}}_2 + z_2 } ) \nonumber \\ \leq & ~ ( \alpha \sqrt{\epsilon } + o(1 ) ) n , \nonumber \end{aligned}\ ] ] where @xmath448 and @xmath449 are finite since @xmath413 contains no zeros by assumption . on the other hand , @xmath450 where @xmath451 by fano s inequality . combining the last two displays
, we have @xmath452 next we apply , with @xmath453 . to verify the conditions , note that the channel @xmath377 is memoryless and additive with non - uniform noise distribution @xmath413 , which satisfies the condition .
similar , the channel @xmath378 is memoryless and additive with noise distribution @xmath437 , which is the maximizer of .
since @xmath413 is not uniform , @xmath437 is not a point mass
. therefore @xmath378 satisfies . then yields @xmath454 where the last inequality follows from the fact that @xmath455 attained by @xmath184 uniform on @xmath456 .
finally , note that the rate pair @xmath457 is achievable by a random mac - code for @xmath239 , with @xmath184 uniform on @xmath456 and @xmath458 .
explaining that the `` missing corner point '' requires proving of ( [ eq : costa_req ] ) , as well as the majority of our knowledge on interference channels were provided by prof .
chandra nair .
we acknowledge his scholarship and patience deeply . the research of y.p . has been supported in part by the center for science of information ( csoi ) , an nsf science and technology center , under grant agreement ccf-09 - 39370 and by the nsf career award under grant agreement ccf-12 - 53205 .
the research of y.w . has been supported in part by nsf grants iis-14 - 47879 , ccf-14 - 23088 and ccf-15 - 27105 .
this work would not be possible without the generous support of the simons institute for the theory of computing and california sb-420 .
to prove the tensorization inequality , let @xmath459 be independent and individually distributed as the optimal coupling of @xmath460
. then @xmath461=\sum_{i=1}^n { \mathbb{p}\left[x_i\neq y_i\right ] } = \sum_{i=1}^n { d_{\rm tv}}(p_i , q_i)$ ] . to show ,
let @xmath462 be an arbitrary coupling of @xmath21 and @xmath355 so that @xmath463 is distributed according to the optimal coupling of @xmath464 , that is , @xmath465=n{\bar{d}}(p_x , q_x)$ ] . by the first inequality we just proved , for any @xmath466 , @xmath467 where @xmath468 } { \eta_{\rm tv}}(p_{y_i|x_i})$ ] and the middle inequality follows from dobrushin s contraction coefficient . applying dobrushin s contractoin @xcite ( see ( * ? ? ?
* proposition 18 ) , with @xmath469 and @xmath470 ) , there exists a coupling @xmath471 of @xmath21 and @xmath355 , so that @xmath472 and @xmath473 \leq \eta { \mathbb{e}}_{\pi}[d_h(x,{{\tilde{x } } } ) ] = n \eta { \bar{d}}(p_x , q_x)$ ] , concluding the proof .
basic properties of @xmath380 follow from standard arguments .
to show the strict inequality @xmath383 under the conditions and , we first notice that @xmath380 is simply the concave envelope of the set of achievable pairs @xmath474 obtained by iterating over all @xmath345 . by caratheodory
s theorem , it is sufficient to consider a ternary - valued @xmath59 in the optimization defining @xmath475
. then the set of achievable pairs @xmath476 is convex and compact ( as the continuous image of the compact set of distributions @xmath477 ) . consequently , to have @xmath393 there must exist a distribution @xmath477 , such that @xmath478 we next show that under the extra conditions on @xmath378 and @xmath377 we must have @xmath479 .
indeed , guarantees the channel @xmath378 satisfies the strong data processing inequality ( see , e.g. , ( * ? ? ?
* exercise 15.12 ( b ) ) and ( * ? ? ? * section 1.2 ) for a survey ) that there exists @xmath480 such that @xmath481 from ( [ eq : th0 ] ) and ( [ eq : th1 ] ) we infer that @xmath482 , or equivalently @xmath483 on the other hand , the condition ensures that then we must have @xmath484 .
clearly , this implies @xmath479 in ( [ eq : th0 ] ) . to show the single - letterization statement ( [ eq : sl ] ) , we only consider the case of @xmath485 since the generalization is straightforward by induction .
let @xmath486 be a markov chain with blocklength-@xmath487 memoryless channel in between .
we have @xmath488 where ( [ eq : th2 ] ) is because @xmath489 and hence @xmath490 , and ( [ eq : th3 ] ) is because @xmath491 .
next consider the chain @xmath492 where ( [ eq : th4 ] ) is by @xmath493 and hence @xmath494 , ( [ eq : th5 ] ) is by the definition of @xmath380 and since we have both @xmath495 and @xmath496 , ( [ eq : th6 ] ) is by the concavity of @xmath380 , and finally ( [ eq : th7 ] ) is by the monotonicity of @xmath380 and ( [ eq : th3 ] ) .
ronit bustin , h vincent poor , and shlomo shamai .
the effect of maximal rate codes on the interfering message rate . in _ proc .
2014 ieee int .
inf . theory ( isit ) _ , pages 9195 , honolulu , hi , usa , july 2014 .
max h.m .
costa and olivier rioul . from almost gaussian to gaussian : bounding differences of differential entropies . in _ proc .
information theory and applications workshop ( ita ) _ , san diego , ca , february 2015 . | it is shown that under suitable regularity conditions , differential entropy is @xmath0-lipschitz as a function of probability distributions on @xmath1 with respect to the quadratic wasserstein distance . under similar conditions ,
( discrete ) shannon entropy is shown to be @xmath2-lipschitz in distributions over the product space with respect to ornstein s @xmath3-distance ( wasserstein distance corresponding to the hamming distance ) .
these results together with talagrand s and marton s transportation - information inequalities allow one to replace the unknown multi - user interference with its i.i.d .
approximations . as an application , a new outer bound for the two - user gaussian interference channel is proved , which , in particular , settles the `` missing corner point '' problem of costa ( 1985 ) . |
concentration of the primordial solid material is a fundamental requirement for planet formation .
the interstellar medium contains about 1% solids by mass in the form of dust grains , and the sun has a similar abundance of heavy elements , while the terrestrial planets consist almost entirely of refractory material .
even the most gas - rich solar system planet , jupiter , has between 1.5 and 6 times the solar abundance of heavy elements @xcite .
a basic concentration process operating in the solar nebula and other protostellar disks is the settling of dust particles due to the component of the stellar gravity perpendicular to the disk plane .
the grains settle at different speeds depending on their ratio of mass to area , leading to mutual collisions .
if the disk gas is laminar and the initially sub - micron particles readily stick together , then most of the solid material settles into a thin midplane layer within a few tens of thousands of years at 1 to 10 au @xcite .
such rapid loss of the grains conflicts with observations of scattered starlight showing some dust remains suspended in the atmospheres of million - year - old protostellar disks @xcite .
however mid - infrared colors suggest the dust abundance is reduced near the disk photosphere , consistent with the incorporation of some material into larger bodies in the interior @xcite .
the smaller particles can remain aloft if stirred by turbulence in the gas @xcite .
turbulence is also commonly invoked to drive the observed flow of material onto the central star . in particular , the turbulence resulting from the magneto - rotational instability ( mri ; * ? ? ?
* ; * ? ? ?
* ) can provide the necessary outward transfer of orbital angular momentum .
however , the mri requires a sufficient level of ionization for coupling the disk gas to the magnetic fields .
the weak internal heating means thermal ionization is effective only very near the star , while the interstellar cosmic rays and stellar x - rays ionize only the top and bottom surfaces of the disk , failing to penetrate the interior @xcite . furthermore , the recombination of the free charges on grain surfaces is so efficient that a small mass fraction of sub - micron particles is sufficient to shut off mri turbulence in much of the region where the planets formed @xcite . as a result ,
protostellar disks consist of a laminar dead zone sandwiched between two turbulent active layers .
the dead zone extends from 0.1 to 15 au in the minimum - mass solar nebula , given micron - sized grains and ionization by the median stellar x - ray luminosity @xcite .
this paper deals with the consequences of the active and dead layers for dust settling and disk evolution .
we investigate how the gas flows govern the distribution of the grains , and how the grains in turn affect the dead zone size , using the methods outlined in
[ sec : methods ] to make ionization balance and 3-d mhd calculations of small patches of the disk around a t tauri star .
since a wealth of data is now available on the abundance , size and composition of dust in the surface layers of protostellar disks overlying the regions where planets are likely to form @xcite , we examine the observational signatures of the movements of the grains , seeking to understand the following issues . 1 .
what is the cause of the large measured range in accretion rate at a given stellar mass ( [ sec : massflow ] ) ?
solar - mass t tauri stars grow at rates between about @xmath1 and @xmath2 m@xmath3 ( compiled by * ? ? ?
* ) , while simple magnetic accretion models have a fixed column of accreting gas set by the penetration depth of the ionizing radiation , implying a unique mass flow rate @xcite .
why do t tauri stars show such diverse mid - infrared spectra ( [ sec : movements ] ) ?
the variety in the silicate band strengths and shapes indicates wide ranges in the abundances of small and crystalline particles .
high crystalline mass fractions appear more often in systems having colors consistent with grains concentrated near the midplane @xcite .
3 . what are the origins of the variability over intervals of days to years in the 10-@xmath0 m silicate feature @xcite and nearby continuum @xcite ( [ sec : variable ] ) ? the shorter variation timescales are substantially faster than the orbital periods at the au radii where the feature forms .
a summary , discussion and conclusions are in [ sec : conc ] .
we consider a patch of disk lying 5 au from a solar - mass star , where the orbital period is 11.2 yr . the surface density @xmath4 g @xmath5 and temperature @xmath6 k are taken from the minimum - mass solar nebula @xcite .
we solve the standard equations of isothermal resistive mhd @xcite in the frame co - rotating with the orbital speed of the domain center , using the stratified shearing - box approximation @xcite with the zeus code @xcite .
the density in the initial hydrostatic equilibrium varies with the height @xmath7 as a gaussian @xmath8 having scale height @xmath9 au equal to the ratio of the sound speed @xmath10 with the orbital frequency @xmath11 .
the mean molecular weight of the gas is @xmath12 .
the domain size along the local cartesian axes @xmath13 , corresponding to the radial , azimuthal and vertical directions , is @xmath14 au and the box is divided into @xmath15 grid cells .
the @xmath16-boundaries are shearing - periodic , the @xmath17-boundaries are periodic and the @xmath7-boundaries allow outflow but no inflow . a major factor in fixing the strength of magneto - rotational turbulence is the net vertical flux of the magnetic field @xcite .
the periodic horizontal boundaries in our calculations mean that the initial vertical flux is preserved over time , while any spatially - varying fields are readily destroyed through reconnection .
we choose a starting field that is the sum of a uniform vertical part @xmath18 and a zero - net - flux part with strength @xmath19 , having a sinusoidal variation in @xmath16 with one period in the box width .
the initial magnetic field thus has components @xmath20 , @xmath21 and @xmath22 .
the uniform part is chosen on the basis of the solar nebula fields of 0.1 to several gauss , recorded in the remanent magnetization of meteorites from the asteroid belt @xcite .
the fields at 5 au were more likely near the low end of this range .
magneto - rotational turbulence typically produces fields with an rms vertical component that is 10 to 20% of the total field strength ( e.g. * ? ? ?
* ) . from these considerations ,
we choose @xmath23 mg . the spatially - varying part of the initial magnetic field is given an amplitude @xmath24 mg similar to the rms field strength found in ideal - mhd test calculations .
the pressure in the uniform part of the field is @xmath25 times less than the initial gas pressure at the midplane , and @xmath26 times less than the gas pressure at four scale heights .
the resistivity in each grid cell is computed by solving a time - dependent ionization network including recombination in the gas phase and on grains ( [ sec : resistivity ] ) .
we carry out three resistive mhd calculations , each involving grains of one size , either 1 , 10 or 100 @xmath0 m , modeled as a fluid settling through the gas at the terminal speed ( [ sec : laminar ] ) .
initially the grains are well - mixed with a mass fraction of 1% and the reaction network is in chemical equilibrium .
the largest grains settle two orders of magnitude faster than the smallest .
in addition their cross section per unit gas mass is two orders of magnitude less , producing slower recombination and a smaller dead zone .
the electrical resistivity of the disk gas @xmath27 is inversely proportional to the electron fraction @xmath28 , where @xmath29 is the electron number density and @xmath30 the total number density of neutrals @xcite .
we compute the electron fraction by solving the reduced gas - phase recombination reaction network of @xcite , augmented with reactions involving grains as described by @xcite .
the electrons recombine with the ions through grain charging and neutralization , and in the gas phase through dissociative recombination and through charge transfer to metal atoms followed by radiative recombination .
the fraction of the metal atoms free to leave the grain surfaces and enter the gas phase is chosen to be 1% .
the advection of the reacting species across the grid is treated in an operator - split fashion @xcite .
the ionization is due to the x - rays from the young star and the radioactive decay of @xmath31al nuclei within the disk .
the x - ray ionization rate is interpolated in monte carlo radiative transfer results including scattering , computed by @xcite . at depths
not shown on their plot we attenuate the x - rays using an @xmath32-folding column of 8 g @xmath5 .
the x - rays have a 5 kev thermal spectrum .
where not otherwise specified we use a fiducial stellar x - ray luminosity of @xmath33 erg s@xmath34 , matching young solar - mass stars observed in the orion nebula @xcite .
the ionization rate is proportional to the luminosity .
the @xmath31al radioactive decay ionization rate is @xmath35 s@xmath34 at the initial dust - to - gas mass ratio of 0.01 , and varies in proportion to the local dust abundance .
the x - rays dominate the ionization in the outermost 30 g @xmath5 , or the top and bottom 20% of the mass column ( figure [ fig : ionization ] ) .
we neglect the changes in the x - ray ionization rate caused by the variation of the opacity with the dust abundance @xcite , since no detailed radiative transfer calculations are available . in line with the absence of heavy - element opacities in the monte carlo calculations
, we also neglect the absorption of x - rays in the grains .
the drag force between the gas and dust follows the epstein law , since the mean free path of the molecules in the gas is much larger than the grain size , and the particles generally move through the gas at subsonic speeds .
the acceleration felt by dust particles of radius @xmath36 and internal density @xmath37 moving at speed @xmath38 through gas with density @xmath39 and sound speed @xmath10 is @xmath40 the particles come to rest if no other forces act after the drag time @xmath41 figure [ fig : tstop ] shows the variation of the drag time with height and grain size in the hydrostatic equilibrium density profile . for grains
100 @xmath0 m and smaller , the drag time is less than the orbital period within four scale heights of the midplane . we have chosen an internal density for the particles @xmath42 g @xmath43 about twice that of rock , so that the grains will settle significantly within the 60-orbit time span of the mhd calculations .
the geometric cross - section per unit gas mass , a key to determining the recombination rate , is thus equal to that of particles with a more typical internal density @xmath44 g @xmath43 and a lower dust - to - gas mass ratio 0.5% .
particles falling through the gas under gravity after one drag time reach a steady - state in which the two forces are equal and opposite . balancing the drag force @xmath45 with the vertical acceleration due to gravity in the disk , @xmath46 ,
gives the terminal speed @xmath47 this amounts to @xmath48 times the sound speed for each scale height of distance from the midplane .
the time to settle to a height @xmath7 is approximately @xmath49 or , in words , the settling time in orbits is @xmath50 or about one - fortieth the inverse of the drag time in orbits .
we wish to model the grain settling in magneto - rotational turbulence .
since the drag time is less than the correlation time @xmath51 of the gas motions in the turbulence @xcite , the particles settle at their terminal speeds almost always .
an exception occurs for 100-@xmath0 m particles near @xmath52 , which have drag times comparable to the correlation time , but as shown by the estimates above and confirmed in
[ sec : movements ] , these particles quickly settle into the interior .
we treat the dust as a fluid moving with a velocity equal to the sum of the gas velocity @xmath53 and the terminal velocity @xmath54 , with the latter directed toward the midplane . when the gas is near hydrostatic equilibrium , this is equivalent to the short friction time approximation used by @xcite , and has the advantage that the time step can be longer than the drag time , reducing the computing load by several orders of magnitude compared with the full dust equation of motion .
the back - reaction of the dust on the gas is neglected .
we solve the conservation equation for the grain number density @xmath55 , @xmath56 using the same standard second - order accurate van leer transport algorithm applied to the gas . the courant condition for the dust is satisfied in our mhd calculations because the terminal speed is less than the greatest alfvn speed on the grid , typically about @xmath57 . to test the method , we compute the time evolution of a layer of 100 @xmath0 m dust particles initially well - mixed with the gas .
the domain extends from the midplane to four scale heights and is divided into 64 grid cells .
the gas is stationary while the dusty layer contracts over time through settling . to gauge the accuracy of the solution
, we also compute the paths followed by individual grains using a fourth - order runge - kutta integration scheme @xcite .
the results are shown in figure [ fig : zsett ] .
the settling is fastest in the low - density disk atmosphere , so that most of the grains that have settled significantly at any given time @xmath58 lie bunched - up near the top of the dusty layer .
the thickness of the layer varies roughly as the square root of the logarithm of @xmath59 , in agreement with eq .
[ eq : tsett ] given the gaussian gas density profile .
equation [ eq : tsett ] overestimates the thickness at early times , when the distance the particles have fallen is less than their height , and underestimates the thickness after a few hundred years , when the edge of the dust layer lies near the midplane , the gas density is roughly uniform , and the terminal speed no longer decreases sharply as the particles fall .
the paths of four individual grains from the trajectory integrator starting at 1 , 2 , 3 and 4 scale heights illustrate the pile - up near the top of the dusty layer and show that the treatment of the settling in zeus has satisfactory accuracy .
the uppermost particle ( red curve ) lies near to the highest grid cell in the zeus calculation having a grain abundance more than half the initial value ( gray line ) .
after a few hundred years , when the dusty layer is just a few cells thick , the zeus calculation is less accurate as numerical diffusion becomes important and spreads the layer .
similar results are obtained in test calculations with 10- and 1-@xmath0 m grains where settling is 10 and 100 times slower .
conservation of orbital angular momentum dictates that the rate at which mass flows through the disk to the star @xmath60 is proportional to the height - integrated accretion stress @xmath61 when winds are negligible .
since most of the mass flow passes through the active layers , the accretion rate is the product of the active layer thickness with the volume - averaged stress there .
the thickness of the active layer depends on the ionization profile of the disk material .
magneto - rotational turbulence can be shut off by the ohmic resistivity when charged and neutral particles collide frequently @xcite and by ambipolar diffusion when collisions are rare @xcite . the third non - ideal term in the induction equation
, the hall term , by contrast has little effect on the saturation amplitude of the turbulence over the range of parameters that has been explored in non - linear calculations @xcite .
the active layer thickness also depends on the field strength .
the linear magneto - rotational instability grows slowly if the magnetic pressure exceeds the gas pressure @xcite , on fields having the substantial toroidal component expected given the rapid orbital shear .
numerical ideal - mhd calculations , in general agreement with the linear results , show the stress declines with height in the strongly - magnetized disk corona @xcite .
to estimate the active layer thickness , we compare the limits imposed by the ohmic and ambipolar terms and the magnetic pressure .
we bring the ionization - recombination reaction network to local equilibrium in the presence of well - mixed grains . the electron density approaches within 10% of its equilibrium value in less than one orbit , at all locations below 4 scale heights when grains are present ( figure [ fig : teqm ] ) . consequently , turbulent mixing occurring on timescales longer than the orbital period is unlikely to change the ionization significantly .
the active layer separates the dead zone inside the disk from the corona outside .
rapid ohmic dissipation prevents magneto - rotational turbulence at the high densities found in the dead zone , and rapid ambipolar diffusion prevents turbulence at the lower densities found in the corona @xcite .
the minimum ohmic and ambipolar diffusion rates needed to shut off the turbulence correspond to the field drifting across the characteristic vertical wavelength of the instability , @xmath62 , within the characteristic growth time @xmath63 , where the alfvn speed @xmath64 .
the resulting ohmic dead zone criterion is @xmath65 @xcite . the elsasser number criterion eq .
[ eq : elsasser ] yields more accurate estimates of the dead zone boundary than a version with the alfvn speed replaced by the sound speed @xcite .
the dimensionless number governing whether ambipolar diffusion prevents turbulence in the neutrals @xmath66 @xcite is proportional to the ion - neutral drag coefficient @xmath67 and the ion density @xmath68 , and independent of the magnetic field strength . using the drag coefficient described by @xcite
, we find that eq .
[ eq : ambipolar ] holds above about @xmath69 ( figure [ fig : dimensionless ] ) .
the height where @xmath70 reaches unity is almost independent of the grain properties , because most recombination occurs in the gas phase at the low ambient densities and high ionization fractions @xcite .
the height where the magnetic and gas pressures match depends on the strength of the magnetic fields generated in the turbulence . in our three resistive mhd calculations , the sinusoidal part of the magnetic field quickly dissipates throughout , leaving in the dead zone just the net vertical part @xmath18 , and in the active layers the vertical part plus a newly - regenerated tangled field . the horizontally - averaged maxwell stress outside
the dead zone is a few times the reynolds stress , as expected in mri turbulence .
the magnetic pressure averaged over the time interval from 10 to 60 orbits exceeds the gas pressure above about @xmath71 , and the stress declines steeply with height near the domain boundary at @xmath52 .
ideal - mhd calculations in a taller shearing - box @xcite show an exponential decline in the stress through @xmath72 .
these results indicate that the upper boundary of the active layer is determined by the strong magnetization , rather than the ambipolar diffusion , and lies near @xmath52 . at the base of the active layer the ohmic resistivity given by eq .
[ eq : resistivity ] shuts off the turbulence according to eq .
[ eq : elsasser ] . combining these relations under the conditions found at 5 au in the minimum - mass solar nebula yields a critical electron fraction @xmath73 .
the ratio @xmath74 of the gas pressure to the mean pressure in the vertical component of the field is about @xmath75 near the base of the active layer in our three mhd calculations , so the ionization fraction is a few times @xmath76
. the allowed range of heights for the base of the active layer under more general magnetic field conditions can be found from eq .
[ eq : elsasser ] using the known resistivity profile .
a maximum height for the base and thus a minimum total thickness for the active layer comes from considering only the uniform part of the magnetic field @xmath23 mg in the mhd calculations .
a minimum height for the base and an upper bound on the active layer thickness come from using the field strength @xmath24 mg found in the ideal - mhd tests .
the resulting limits on the active layer thickness are shown in figure [ fig : activeht ] as functions of the grain size and x - ray luminosity .
the x - ray luminosity range plotted corresponds to the spread found in chandra x - ray observatory studies of the orion nebula that detected most of the lightly - obscured young solar - mass stars @xcite .
the active layer thickness varies from @xmath77 , when the field takes the weaker value , the x - ray luminosity is low and the grains are small , to the full @xmath52 extent of the disk , when the field is strong and the grains are 10 @xmath0 m or larger .
the range in active layer thickness is thus a factor 3.2 . determining the range of possible mass flow rates involves estimating the mean stress in the active layers .
the turbulence regenerates the fields , so that the saturated field strength is determined jointly by the imposed magnetic flux and , through the resistivity of the gas , by the x - ray illumination and the distribution of the grains .
the mass flow and the extent of the active layers in the three mhd calculations are shown in figure [ fig : hdmdotdz ] .
the quantity plotted @xmath78 is proportional to the total accretion stress resulting from magnetic and hydrodynamic forces .
the overall mass flow rate in the run with 100-@xmath0 m grains is 3.55 times that in the run with 1-@xmath0 m grains .
this total is made up of a factor 1.49 due to the range in active layer thicknesses , and a remaining factor 2.38 due to the stronger magnetic fields found in the thicker active layers .
if the stresses have the same proportionality to the active layer thickness over the whole of figure [ fig : activeht ] , then variations in the x - ray luminosity and grain size can change the mass flow rate by about a factor 16 .
magneto - rotational turbulence can explain the two - decade range in the observed accretion rates if the remaining factor of six comes from other effects .
possibilities include ( 1 ) some disks having an abundance of very small grains , yielding a very thin active layer @xcite ; ( 2 ) a range of magnetic fluxes among t tauri systems [ the stress increases with the vertical magnetic flux @xcite ] ; and ( 3 ) changes in the mass flow rate with radius and time @xcite .
it is of interest to compare the mass column of the active layer with the x - ray attenuation column , 8 g @xmath5 .
the active column is less than the attenuation column when the grains are small and recombination is fast , and greater when the grains are large .
the mean active column in the run with 1-@xmath0 m grains is 5.18 and 4.82 g @xmath5 on the disk top and bottom , respectively .
the numbers are 13.3 and 12.7 g @xmath5 in the run with 10-@xmath0 m grains , and 32.3 and 33.6 g @xmath5 in the run with 100-@xmath0 m grains .
the average values of 5.0 , 13.0 and 32.9 g @xmath5 span a factor 6.6 .
the dust particles in our three mhd calculations move by settling and by advection with the gas .
the gas is near magneto - hydrostatic equilibrium throughout .
its velocity field consists of magneto - rotational turbulence superimposed on a gradual expansion due to outflow from the disk surfaces .
wave motions replace the turbulence inside the dead zone .
the three following sections outline the effects on the dust of the settling , the outflow and the turbulent stirring , while the fourth section deals with the movements of the particles in the dead zone .
dust settles fastest in the upper atmosphere , where the gravity is strong and the low density means weak gas drag . in the absence of turbulence , settling is quick . according to the rough estimate of eq .
[ eq : tsett ] , particles 1 , 10 and 100 @xmath0 m in size in 100 orbits reach heights of 3.2 , 2.4 and @xmath79 , while the more accurate numerical integrations shown in figure [ fig : zsett ] yield 2.5 , 1.7 and @xmath80 .
coagulation can decrease the cross - sectional area per unit mass , making settling faster still , but we have assumed here that the particles do not stick to one another . the settling speed is inversely proportional to the gas density ( eqs .
[ eq : tdrag ] and [ eq : vterminal ] ) .
the magnetic fields cause substantial departures from the hydrostatic equilibrium gaussian density profile outside about @xmath81 , where the magnetic pressure approaches or exceeds the gas pressure .
magnetic support allows higher densities in the upper atmosphere , with the extra material coming from the deeper layers where the density is correspondingly reduced ( figure [ fig : densitiez ] ) .
the mean gas density at @xmath52 is up to eight times greater than the hydrostatic value .
horizontal density fluctuations in the turbulence however yield an average settling speed @xmath82 greater than the speed predicted from the mean density @xmath83 ( figure [ fig : vsettratio ] ) .
the fluctuations make settling faster than the mean density profile by about 10% at @xmath84 and 80% at @xmath52 .
density fluctuations in the dead zone are small and have little effect on the settling .
the ionization remains near its time - varying equilibrium value in the mhd calculations as the grains settle out .
chemical equilibration occurs faster than settling except with 100 @xmath0 m grains above 3.5 scale heights ( figures [ fig : tstop ] and [ fig : teqm ] ) . at the low densities found above this height , reactions on the surfaces of the large grains in any case are negligible and the ionization state is determined by gas - phase processes .
gas intermittently escapes from the top and bottom of the domain as the turbulence carries material across the boundaries .
the greatest speeds are a few times the sound speed , and thus the lost gas remains gravitationally bound to the star and may return to the disk .
since the calculations do not reach high enough nor include the radial gradients that may be important for any acceleration through escape speed , we here describe the outflows only to the extent needed to subtract their effects .
the outflows result in the surface density declining between 10 and 60 orbits by 6.9 , 8.3 and 11% in the calculations with 1- , 10- and 100-@xmath0 m grains .
the average mass flow rate in the 1-@xmath0 m case is @xmath85 @xmath86 yr@xmath34 au@xmath87 .
the remaining gas stays near hydrostatic equilibrium by expanding slightly , and the expansion carries gas and in some cases dust from the interior into the disk atmosphere . for continuity of mass ,
the expansion speed is approximately proportional to the inverse density .
the speeds of gas upwelling and grain settling thus have similar functional increases with height . at @xmath81 ,
the gas flows away from the midplane at about 0.3% of the sound speed , while grains of 1 , 10 and 100 @xmath0 m settle at 0.012 , 0.12 and 1.2% of the sound speed in the initial gaussian density profile according to eq .
[ eq : vterminal ] and figure [ fig : tstop ] .
the settling is faster than the outflow at all heights in the run with 100-@xmath0 m grains , so that the dust distribution results mostly from the competition between settling and turbulent stirring .
by contrast the outflow is faster than the settling everywhere in the run with 1-@xmath0 m grains , but at @xmath52 the grains settle at 30% of the gas upwelling speed . as a result ,
the grains are lost at a slower rate than the gas and their mass fraction increases over time in the upper atmosphere .
the run with 10-@xmath0 m grains is an intermediate case , and the outflow carries the particles to about @xmath88 ( figure [ fig : settlez ] ) . above this height
the settling is faster than the outflow .
turbulence causes mixing , making the dust - to - gas ratio more uniform .
the timescale for mixing through one density scale height is @xmath89 where the denominator is the turbulent diffusion coefficient and is proportional to the correlation time of the turbulence @xmath90 and to the mean squared turbulent vertical velocity @xcite .
the turbulent speed is obtained by subtracting the horizontally - averaged outflow speed @xmath91 from the total .
the mixing timescale is tens of orbits in much of the active layer . in particular , at the dead zone boundary the average mixing time is 24 , 53 and 46 orbits in the calculations with 1 , 10 and 100 @xmath0 m grains , respectively . by comparison ,
grains initially in the upper atmosphere settle to the dead zone boundary in 900 , 270 and 112 orbits , respectively . also , the outflowing gas crosses one scale height near the dead zone boundary in approximately 43 , 162 and 288 orbits .
the turbulent mixing is thus the fastest of the three processes at the edge of the dead zone .
the results of mixing the 1-@xmath0 m grains appear in figure [ fig : xgvsz1um ] , showing the dust mass fraction versus height .
the dust settles in one orbit to @xmath92 . by 60 orbits ,
settling without turbulence produces a spike in the dust abundance near @xmath71 . in the mhd calculation where outflow and turbulence occur alongside the settling ,
the spike is smoothed out across the active layer as follows .
the settling enhances the dust abundance at 2 to @xmath84 .
the outflow carries some of the dust - enriched material upward , and the turbulence spreads the dust - enriched material both up into the atmosphere and down into the dead zone .
the downward mixing overshoots the dead zone boundary by about one scale height . settling and turbulence together
thus speed up the concentration of the solids in the deeper layers compared with the laminar case .
this contradicts the usual wisdom that turbulence counteracts settling .
similar processes operate on the 10-@xmath0 m grains ( figure [ fig : xgvsz10um ] ) .
few of the particles remain by 60 orbits above @xmath93 , where the settling time is less than an orbit and settling overwhelms both the turbulent mixing and the entrainment in the outflow .
the turbulence mixes the dust almost uniformly through the dead zone , as the movements of the gas overshoot the dead zone boundary all the way to the midplane .
the 100-@xmath0 m grains ( figure [ fig : xgvsz100um ] ) have mostly settled below @xmath94 by 1 orbit and remain there at 60 orbits . nearer
the midplane , the stirring and settling timescales are comparable .
the dead zone extends just @xmath95 either side of the midplane , and overshooting turbulent motions reach all parts of the interior .
the settling is fast enough that the dust abundance has become greatest close to the midplane .
the turbulent mixing thus on balance moves grains from the dead zone to the active layers , preventing further concentration .
the movements of the dust have little effect on the dead zone depth over the time interval spanned by the calculations . in figure
[ fig : deadzonebdy ] we plot three different measures of the dead zone boundary height versus time , in the case with 1-@xmath0 m grains . the elsasser number from eq .
[ eq : elsasser ] is computed ( 1 ) holding the resistivity profile fixed at its time - averaged level , ( 2 ) holding the squared alfvn speed profile fixed , and ( 3 ) allowing both the resistivity and the alfvn speed to vary with time .
the first and third cases generally coincide , showing that it is the magnetic field , rather than the rate of recombination on grains , that controls the changes in the dead zone size over timescales comparable to the orbital period .
to check whether neglecting particle growth is valid , we compare the expected grain - grain collision speeds with the threshold of about 1000 cm s@xmath34 for the disruption of silicate aggregates established by laboratory measurements @xcite . adopting the expressions for the relative velocities of particles in isotropic turbulence derived by @xcite
, we find that aggregates of 10 @xmath0 m or larger can be disrupted by collisions with 1-@xmath0 m particles in the three mhd calculations above about @xmath84 .
compact spheres as large as 100 @xmath0 m settle fast enough that they rarely reach such heights , judging from figure [ fig : xgvsz100um ] .
meanwhile near the midplane , collisions are too slow for disruption , and growth to larger sizes is likely . in this section we examine the movements of the dust particles in and near the dead zone .
we focus first on the calculation with the 1-@xmath0 m grains , where the dead zone is deep enough that turbulent mixing fails to reach the midplane . to isolate the contributions of settling and outflow we compare the mhd results against ( 1 ) the hydrostatic settling case and ( 2 ) a numerical solution of the dust continuity equation in a time - constant flow @xmath96 that approximates the outflow found in the mhd calculation .
the expansion of the gas carries less - settled and therefore less - dust - enhanced interior material into the surface layers , leading to slower increases in the dust abundance over time than in the hydrostatic case .
the evolution of the dust abundance at several different heights is shown in figure [ fig : deadsett ] . outside the dead zone boundary , at @xmath94 , the abundance in the mhd calculation increases and decreases intermittently due to the turbulence , while overall tracking slightly below the outflow solution .
the deficit results from the loss of dust to the deeper layers through turbulent mixing ( figure [ fig : xgvsz1um ] ) . just inside the dead zone boundary , at @xmath97 , the abundance is higher than the outflow solution owing to the dust mixed down from above .
meanwhile at the midplane , the dust abundance follows the laminar solution with small oscillations .
deep in the dead zone , the settling proceeds largely as if there were no turbulence in the active layers . to see how turbulence overshooting the dead zone boundary
affects the motions of individual grains , we have also computed the paths followed by a large number of particles in the time - dependent gas density and velocity field of the mhd calculation using the runge - kutta scheme described in section [ sec : laminar ] .
selected particles are shown in figure [ fig : trajctry ] .
the grains found in the active layers follow looping paths in the turbulence , as on parts of the blue and orange curves .
the radial and vertical speeds are comparable , and the vertical motions plotted in the bottom panel are irregular .
grains found in the dead zone in contrast show mostly slower periodic vertical oscillations and have little radial motion , as on the red , green , cyan and magenta curves . particles cross the dead zone boundary moving both toward and away from the midplane , while particles near the midplane remain close to their initial positions .
the non - zero hydrodynamic accretion stress in the dead zone results from propagating waves , which have little impact on the particle settling .
on the other hand , the particles near the midplane oscillate with amplitudes a few tenths of a scale height and phases that drift over time relative to one another , suggesting that the settling will not yield a thin flat midplane dust layer of the kind typically assumed for the formation of planetesimals by gravitational instability @xcite .
the thickness of the settled layer in an otherwise stationary nebula is just 1% of the gas scale height , if limited by the onset of the kelvin - helmholtz instability between the dust - rich keplerian component and the overlying gas that orbits at sub - keplerian speeds owing to its partial support against the stellar gravity by the radial gas pressure gradient @xcite .
time variations in the strength and shape of the silicate band at wavelength 10 @xmath0 m have now been seen by several groups using instruments on the ground and in space .
about a quarter of the two dozen low- and intermediate - mass young stars monitored by @xcite and @xcite from 1992 to 2002 showed significant variations in the silicate bands but not in the continuum .
the observations were separated by months to years and the variations showed no clear period .
the generally poor correlations with changes in optical brightness and extinction appear to rule out an origin in the reprocessing of the starlight by a static envelope , but may be consistent with the features arising from dust in the disk atmosphere @xcite . judging from static models ,
the 10-@xmath0 m emission would come mostly from dust near 1 au @xcite . a much larger census of the chameleon i star - forming region with the spitzer space telescope detected photometric variations at 3.6 to 8 @xmath0 m in about half of all disk - bearing sources observed at intervals of months to years @xcite . the silicate band varied on comparable timescales in two of 11 accreting t tauri stars observed with spitzer by @xcite and in the ex lupi system observed by @xcite .
a further system , the transitional disk lrll 31 , exhibits variability in the silicate feature and mid - infrared continuum in observations separated by as little as a week , which might be due to changing shadows cast on the disk by material near the star @xcite . in several examples ,
spatially resolved optical and near - infrared images taken a few days to months apart show changes in shape and brightness at distances of 50 to 350 au @xcite .
dynamical processes operating at these radii appear incapable of producing such rapid changes .
photopolarimetry of one of the objects , hh30 , shows periodic modulation on timescales of about a week , explained by a lighthouse picture in which a beam or shadow from a central source sweeps across the disk @xcite .
altogether , the evidence is consistent with the widespread infrared variability arising at least in part from changes in the distribution or illumination of the dust in the disk atmospheres . to help evaluate whether magnetic activity
moves the dust around enough to cause the variability , we use the results of the mhd calculation with 1-@xmath0 m particles in this section to compute in turn the height of the photosphere , the appearance of the disk patch at mid - infrared wavelengths , and the obscuring column along lines of sight grazing the disk surface .
the photosphere height is plotted versus time in figure [ fig : tau1 ] .
optical depths are calculated by integrating the dust column along the ray passing vertically down through the center of the box .
the results are shown for opacities of 1000 , 100 and 10 @xmath98 per gram of dust .
the highest of these is comparable to the geometric cross - section of 1500 @xmath98 g@xmath34 for the 1-@xmath0 m grains and to the 982 @xmath98 g@xmath34 opacity at wavelength 10 @xmath0 m obtained for grains in dense molecular cloud cores by @xcite .
this highest opacity places the photosphere in the most magnetically - active layers , where the fluctuations in the dust density are greatest .
the location of the photosphere varies over time by more than a scale height .
the lower - opacity curves correspond to dust depletions of one and two orders of magnitude . at the lowest opacity ,
the active layers are optically thin and the photosphere lies inside the dead zone .
optically - thin active layers at 1 au have been suggested as a consequence of the low dust abundance needed for magnetic stresses there to yield the measured stellar accretion flows @xcite .
the time variation in the thermal emission from the dust is illustrated in figure [ fig : movie ] .
we calculate the emission at wavelength 10 @xmath0 m using a ray - tracing radiative transfer code based on that of @xcite as described by @xcite , taking an opacity of 100 @xmath98 per gram of dust .
scattering provides a small fraction of the total opacity ( e.g. * ? ? ?
* ) so we treat only absorption and thermal emission . to mimic the warm surface layer produced by starlight @xcite , stellar x - ray irradiation @xcite and photoelectric heating @xcite , we assume when solving the transfer equation that the temperature increases linearly above one scale height , reaching 600 k at @xmath52 . temperature fluctuations due to changes in the location where the stellar radiation is absorbed
are neglected .
the time sequence in figure [ fig : movie ] shows the brightness variations over a three - orbit span .
every few orbits , the magnetic fields along the orbital direction grow strong through the shear , reducing the gas density by one to two orders of magnitude so that the dust quickly settles out ( figure [ fig : episodic ] ) . the fields eventually become buoyant and rise through the boundary .
gas expanding into the space vacated by the fields carries dust particles back into the warm layer where they become bright in the 10-@xmath0 m band .
the eruptions of dusty gas recur episodically , synchronized with the variations in the magnetic activity .
uncorrelated brightness variations in many patches across the disk will produce only slight changes in the emission of the system as a whole .
however the atmosphere can contribute to variability also by casting time - dependent shadows on the more distant parts of the disk , and by obscuring the central star in systems viewed near edge - on .
the shadows sweep across the disk on the orbital period of the shadowing material , while possibly changing in size and strength over shorter timescales determined by the turbulence .
we estimate the magnitude of the faster variations by measuring the mass column along a horizontal ray that is fixed in the rotating frame of the shearing - box , passing @xmath99 above the origin .
the extinction to the central star as seen by an outside observer , on the other hand , changes as the orbital motion carries material through the line of sight .
we estimate the timescales involved by measuring the mass column along a ray that is fixed in the inertial frame and moves across the shearing - box at the orbital speed , making use of the periodic azimuthal boundary condition .
the inertial line of sight crosses the azimuthal extent of the domain in just 0.06 orbits .
the columns of gas and dust are plotted against time in figure [ fig : xcolumn ] . over intervals of several orbits ,
the column decreases by an order of magnitude due to the growth of the magnetic field , before quickly returning to its old level when the field escapes .
the dust settles out fastest when the column is low , reducing the ratio of dust to gas .
faster , erratic column variations on timescales @xmath100 orbit result from the density fluctuations in the turbulence , while still faster variations over about 0.01 orbit , visible only on the inertial line of sight , result from the orbital motion .
as seen on the rotating line of sight , the dust column changes by more than a factor two in 0.1 orbit every 2.3 orbits on average , while similar changes within a shorter interval of 0.01 orbit occur just once in 8.8 orbits . on the inertial line of sight , dust column measurements separated by 0.1 orbit differ by more than a factor two every 1.4 orbits , and those separated by 0.01 orbit show a similar change every 0.22 orbits .
these results indicate that the turbulence will yield changes in the depth of the shadows cast on the disk mostly over timescales longer than a tenth of an orbit , while changes in the extinction to an observer will be substantial on timescales down to at least 0.01 orbit .
note that the patch of disk treated in the mhd calculation has radial optical depth near unity at @xmath99 and thus can cast a strong shadow , if the opacity is a few hundred @xmath98 per gram of dust . in the calculation with 10-@xmath0 m grains ,
the settling is so effective at times of low column that the upper atmosphere becomes almost dust - free .
the gas expands owing to the magnetic pressure making the densities less than the initial hydrostatic profile .
the particles then settle below @xmath93 in less than an orbit ( figure [ fig : xgvsz10um ] ) . the case with 100-@xmath0 m grains
is dust - free at @xmath99 almost always .
we have made resistive mhd calculations of a patch of disk in orbit around a young star , treating simultaneously for the first time three processes that together regulate the distribution of magneto - rotational turbulence .
first , the dust drifts through the gas under the gravity and gas drag forces .
second , the dust alters the resistivity of the gas , and third , the resistivity controls the evolution of the magnetic fields driving the gas flows .
the three processes combine in quite different ways in the disk atmosphere and interior . in the atmosphere ,
the high flux of ionizing radiation from the star allows good coupling between gas and magnetic fields even when the dust abundance is high .
intermittency in the turbulence and in the distribution of dust result from fluctuations in the magnetic field strength . during intervals of weak fields , the magnetic forces drive turbulence that carries dust particles high in the atmosphere .
the turbulent mixing of the grains is faster than either the settling or the outflows resulting from the open boundaries .
when the fields are stronger , their large pressure prevents magneto - rotational turbulence and reduces the gas density , so that the particles settle quickly . at the same time in the dead zone near the disk midplane , recombination on grain surfaces reduces the ionization fraction below the level needed to sustain turbulence .
grains settle at the laminar rate , with superimposed vertical oscillations due to the propagation of waves excited in the turbulent layers .
particles falling into the laminar layer do not return to the disk atmosphere and will eventually concentrate near the midplane , in a layer whose thickness is not less than the minimum imposed by the vertical oscillations . a dead zone enriched in solids
may provide a favorable environment for the growth of larger bodies .
many t tauri systems show a combination of 10-@xmath0 m silicate emission and shallow millimeter spectral slope that is consistent with micron - sized grains in the disk atmosphere and millimeter - sized or larger particles in the interior , though it should be noted that the two wavelengths probe different distances from the star @xcite .
the turbulent and dead zones exchange dust and gas when turbulent motions overshoot the mutual boundary .
the mixing extends about one scale height into the dead zone , as shown by the penetration of the dust - enriched material in figures [ fig : xgvsz1um ] to [ fig : xgvsz100um ] .
the motions can perhaps be described in terms of alfvn waves driven by the breakup of channel flows in the disk atmosphere @xcite .
the net effect is a transfer of dust to the dead zone , because settling proceeds fastest in the disk atmosphere . during the early stages of the concentration of the solid material , turbulence
thus speeds up settling .
the turbulent layer will eventually become dust - free unless replenished through radial transport within the disk or rain - out from an outflow or from the circumstellar envelope .
well - mixed micron or sub - micron grains with the interstellar mass fraction produce a dead zone reaching more than two scale heights from the midplane at 5 au in the minimum - mass solar nebula .
however , grains 10 @xmath0 m or larger yield a dead zone ending below one scale height .
turbulent mixing then penetrates to the midplane , consistent with the finding by @xcite that diffusion adequately describes the effects of the gas motions on particles in a dead zone with a small vertical extent . while we were able to run the case with 1-@xmath0 m grains only about 60 orbits , the continuing loss of dust into the dead zone without resupply
will ultimately yield a depletion factor in the surface layers exceeding the value of about ten needed for mixing to reach the midplane . the rate at which mass flows through the patch of disk toward the star is the product of the turbulent layer depth with the mean accretion stress .
both these quantities increase with the grain size , the ionizing flux from the star , and the magnetic flux threading the disk .
depletion of the dust in the surface layers will have a similar effect to an increase in the grain size .
the two - decade spread in accretion rate among t tauri stars can not easily be produced by grain growth and the observed spread in stellar x - ray output , but can be understood if in addition the objects have a modest range of disk magnetic field strengths .
the mass accretion rate resulting from the magnetic stresses will vary with radius , leading to the build - up of material in some annuli @xcite , an effect not captured in our shearing - box calculations .
apparently the stellar x - rays both increase the disk accretion rate within 5 au ( figure [ fig : activeht ] ) and increase the rate at which material escapes from the disk into a photoevaporating wind at 10 - 40 au @xcite .
furthermore the ionization of the inner disk could contribute to a wind driven by magnetic activity @xcite .
all these effects tend to reduce the surface density of gas in the planet - forming region faster around stars with high x - ray luminosities .
whether it is possible in this way to account for the tendency of non - accreting t tauri stars to have higher x - ray luminosities @xcite or for an inverse correlation between x - ray luminosity and stellar accretion rate @xcite remains unclear .
we suggest that the diversity in the mid - infrared silicate bands among classical t tauri stars is a consequence of the concentration of solid material associated with planet formation .
the growth of larger bodies requires a loss of dust , leading generally to a declining recombination cross - section .
the smallest grains provide most of the recombination when the particles have a size spectrum appropriate for the interstellar medium @xcite , and will dominate still more in the disk atmosphere , where settling removes the larger grains .
the disruption of aggregates through collisions ( [ sec : stirring ] ) thus slows the decline . however , unless the grains making up the aggregates also fragment so that the minimum particle size becomes smaller , the loss of cross - section is likely to continue , since even 1-@xmath0 m particles show appreciable settling in the turbulent disk atmosphere .
the decreasing recombination rates will allow the active layers to expand into the disk interior , and if the mass column is not too great , turbulent mixing will eventually reach the midplane , returning the second - generation debris from planet formation to the atmosphere .
questions for future investigation include when and where the mixing reaches the midplane , and whether the observed correlation between the degree of settling and the crystalline mass fraction @xcite can be explained by the return to the atmosphere of minerals formed near the midplane at high temperature and pressure in planetesimal interiors , impacts , or shocks .
any such correlation will be blurred by the fact that the degree of dust depletion required for mixing to reach the midplane depends on the column of gas , the x - ray luminosity of the star and the magnetic flux delivered to the disk from the surroundings .
the mid - infrared variability of protostellar disks has been proposed to arise from the changing illumination of dust in the disk atmosphere .
explaining timescales shorter than a week is a challenge , since the dust in the disks around young solar - mass stars extends inward only to about 0.1 au @xcite , where the orbital period is 12 days .
the results of the mhd calculations demonstrate that magnetic activity moves dust from place to place in the disk atmosphere on several timescales .
the optical depth along rays grazing the surface of the patch of disk frequently varies by a factor two within a tenth of an orbit as regions of enhanced dust density pass by . over longer time intervals of a few orbits ,
the photosphere moves up and down by as much as a scale height due to changes in the balance between settling and stirring .
clearly the effects of the magnetic activity on the dust distribution should be considered when attempting to understand the infrared variability .
properly accounting for the shadows cast on the disk surface requires models of the inner edge of the dust distribution , perhaps informed by the time - steady picture that has been developed for herbig ae stars @xcite .
our results suggest that magneto - rotational turbulence near the dust sublimation radius can cast shadows that vary over timescales as short as a day .
we are grateful for discussions with s. desch , c. dullemond , a. glassgold , j. goodman , k. kretke and m. wardle .
the work was carried out in part at the jet propulsion laboratory , california institute of technology with the support of the jpl research and technology development and nasa solar systems origins programs .
copyright 2009 .
all rights reserved .
brahm , p. , juhsz , a. , dullemond , c. p. et al .
2009 , nature , 459 , 224 akeson , r. l. 2008 , journal of physics conf .
131 , 012019 akeson , r. l. , boden , a. f. , monnier , j. d. et al .
2005 , , 635 , 1173 armitage p.
j. , livio m. & pringle j. e. 2001 , , 324 , 705 bai x .-
& goodman j. 2009 , , 701 , 737 balbus , s. a. , & hawley , j. f. 1991 , , 376 , 214 balbus , s. a. , & hawley , j. f. 1998 , rev .
, 70 , 1 bary , j. s. , leisenring , j. m. , & skrutskie , m. f. 2009 , , in press ; arxiv:0910.3454 blaes , o. m. , & balbus , s. a. 1994 , , 421 , 163 blum , j. , & wurm , g. 2008 , , 46 , 21 bouwman , j. , henning , th . ,
hillenbrand , l. a. , et al .
2008 , , 683 , 479 burrows , c. j. , stapelfeldt , k. r. , watson , a. m. et al .
1996 , , 473 , 437 chiang , e. i. , & goldreich , p. 1997
, , 490 , 368 chiang , e. , & murray - clay , r. 2007 , nature physics , 3 , 604 chiang , e. , & youdin , a. 2009 , annu .
rev . earth planetary sci .
, in press ; arxiv:0909.2652v2 cisowski , s. m. , & hood , l. l. 1991 , in `` the sun in time '' , ed . c. p. sonett , m. s. giampapa & m. s. matthews ( tucson : univ . of arizona press ) , 761 cotera , a. s. , schneider , g. , hines , d. c. , whitney , b. a. , stapelfeldt , k. r. , & padgett , d. l. 2005 , in proceedings of the conference `` protostars and planets v '' , lpi contribution no . 1286 . ,
dalessio , p. , calvet , n. , hartmann , l. , franco - hernndez , r. , & servn , h. 2006 , , 638 , 314 drake , j. j. , ercolano , b. , flaccomio , e. , & micela , g. 2009 , , 699 , l35 dullemond , c. p. , & dominik , c. 2004 , , 421 , 1075 dullemond , c. p. , dominik , c. & natta , a. 2001 , , 560 , 957 durn - rojas , m. c. , watson , a. m. , stapelfeldt , k. r. , & hiriart , d. 2009 , , 137 , 4330 flaccomio , e. , damiani , f. , micela , g. , sciortino , s. , harnden , f. r. , jr . , murray , s. s. , & wolk , s. j. 2003 , , 582 , 398 muzerolle , j. , flaherty , k. , balog , z. , et al .
2009 , , 704 , l15 fleming , t. , & stone , j. m. 2003 , , 585 , 908 fromang , s. , & papaloizou , j. 2006 , , 452 , 751 furlan , e. , hartmann , l. , calvet , n. et al .
2006 , , 165 , 568 gammie , c. f. 1996 , , 457 , 355 garmire , g. , feigelson , e. d. , broos , p. , hillenbrand , l. a. , pravdo , s. h. , townsley , l. , & tsuboi , y. 2000 , , 120 , 1426 glassgold , a. e. , najita , j. , & igea , j. 1997 , , 480 , 344 glassgold , a. e. , najita , j. , & igea , j. 2004 , , 615 , 972 goldreich , p. , & ward , w. r. 1973 , , 183 , 1051 hartmann , l. , dalessio , p. , calvet , n. , & muzerolle , j. 2006 , , 648 , 484 hawley , j. f. , gammie , c. f. , & balbus , s. a. 1995 , , 440 , 742 hawley , j. f. , & stone , j. m. 1998 , , 501 , 758 hayashi , c. , nakazawa , k. , & nakagawa , y. 1985 , in : black , d. c. , & matthews , m. s. , editors .
protostars and planets ii ( univ . of arizona press : tucson ) , 1100 igea , j. , & glassgold , a. e. 1999 , , 518 , 848 ilgner , m. , & nelson , r. p. 2006
, , 445 , 205 isella , a. , & natta , a. 2005 , , 438 , 899 jin , l. 1996 , , 457 , 798 johansen , a. , & klahr , h. 2005 , , 634 , 1353 kamp , i. , & dullemond , c. p. 2004 , , 615 , 991 kim , w .- t . , & ostriker , e. c. 2000 , , 540 , 372 luhman , k. l. , allen , l. e. , allen , p. r. , gutermuth , r. a. , hartmann , l. , mamajek , e. e. , megeath , s. t. , myers , p. c. , & fazio , g. g. 2008 , , 675 , 1375 luhman , k. l. , mamajek , e. e. , allen , p. r. , & cruz , k. l. 2009 , , 703 , 399 miller , k. a. , & stone , j. m. 2000 , , 534 , 398 muzerolle , j. , calvet , n. , hartmann , l. , & dalessio , p. 2003
, , 597 , l149 muzerolle , j. , luhman , k. l , briceo , c. , hartmann , l. , & calvet , n. 2005 , , 625 , 906 nakagawa , y. , nakazawa , k. , & hayashi , c. 1981 , icarus , 45 , 517 nakagawa , y. , sekiya , m. , & hayashi , c. 1986 , icarus , 67 , 375 neuhuser , r. , sterzik , m. f. , schmitt , j. h. m. m. , wichmann , r. , & krautter , j. 1995 , , 297 , 391 oppenheimer , m. , & dalgarno , a. 1974 , , 192 , 29 ormel , c. w. , & cuzzi , j. n. 2007 , , 466 , 413 pessah , m. e. , chan , c. , & psaltis , d. 2007 , , 668 , l51 pinte , c. , padgett , d. l. , mnard , f. , et al .
2008 , , 489 , 633 preibisch , th . ,
kim , y .- c . , favata , f. , et al .
2005 , , 160 , 401 preibisch , th . , ossenkopf , v. , yorke , h. w. , & henning , th .
1993 , , 279 , 577 sano , t. , & inutsuka , s. 2001 , , 561 , l179 sano , t. , inutsuka , s. , turner , n. j. , & stone , j. m. 2004 , , 605 , 321 sano , t. , miyama , s. m. , umebayashi , t. , & nakano , t. 2000 , , 543 , 486 sano , t. , & stone , j. m. 2002a , , 570 , 314 sano , t. , & stone , j. m. 2002b , , 577 , 534 saumon , d. , & guillot , t. 2004 , , 609 , 1170 schegerer , a. , wolf , s. , voshchinnikov , n. v. , przygodda , f. , & kessler - silacci , j. e. 2006 , , 456 , 535 stapelfeldt , k. r. , mnard , f. , watson , a. m. , krist , j. e. , dougados , c. , padgett , d. l. , & brandner , w. 2003 , , 589 , 410 stelzer , b. , & neuhuser , r. 2001 , , 377 , 538 stone , j. m. , & norman , m. l. 1992a , , 80 , 753 stone , j. m. , & norman , m. l. 1992b , , 80 , 791 suzuki , t. k. , & inutsuka , s. 2009 , , 691 , l49 telleschi , a. , gdel , m. , briggs , k. r. , audard , m. , & palla , f. 2007 , , 468 , 425 turner , n. j. , bodenheimer , p. , & bell , k. r. 1997 , , 480 , 754 turner , n. j. , willacy , k. , bryden , g. , & yorke , h. w. 2006 , , 639 , 1218 turner , n. j. , sano , t. & dziourkevitch , n. 2007 , , 659 , 729 turner , n. j. , & drake , j. f. 2009 , , 703 , 2152 wardle , m. 2007 , , 311 , 35 watson , a. m. , & stapelfeldt , k. r. 2004 , , 602 , 860 watson , a. m. , & stapelfeldt , k. r. 2007 , , 133 , 845 watson , d. m. , leisenring , j. m. , furlan , e. et al .
2009 , , 180 , 84 weidenschilling , s. j. 1980 , icarus , 44 , 172 wooden , d. h. , harker , d. e. , & woodward , c. e. 2000 , in `` thermal emission spectroscopy and analysis of dust , disks and regoliths '' , asp conf .
sitko , m. l. , sprague , a. l. , & lynch , d. k. ( san francisco : astronomical society of the pacific ) , 99 wooden , d. h. , bell , k. r. , harker , d. e. , & woodward , c. e. 2000 , , 32 , 1482 woodward , c. e. , wooden , d. h. , harker , d. e. , bell , k. r. , butner , h. m. , sitko , m. l. , lynch , d. , & russell , r. 2004 , in : `` debris disks and the formation of planets : a symposium in memory of fred gillett '' , asp conf .
caroff , l. , moon , l. j. , backman , d. , & praton , e. ( san francisco : astronomical society of the pacific ) , 224 yorke , h. w. 1986 , in : `` astrophysical radiation hydrodynamics '' , ed .
winkler , k .- h .
a. , & norman , m. l. , ( dordrecht : d. reidel ) , 141 zhu , z. , hartmann , l. , & gammie , c. 2009 , , 694 , 1045 | we apply ionization balance and mhd calculations to investigate whether magnetic activity moderated by recombination on dust grains can account for the mass accretion rates and the mid - infrared spectra and variability of protostellar disks .
the mhd calculations use the stratified shearing - box approach and include grain settling and the feedback from the changing dust abundance on the resistivity of the gas . the two - decade spread in accretion rates among solar - mass t tauri stars
is too large to result solely from variations in the grain size and stellar x - ray luminosity , but can plausibly be produced by varying these parameters together with the disk magnetic flux .
the diverse shapes and strengths of the mid - infrared silicate bands can come from the coupling of grain settling to the distribution of the magneto - rotational turbulence , through the following three effects .
first , recombination on grains 1 @xmath0 m or smaller yields a magnetically - inactive dead zone extending more than two scale heights from the midplane , while turbulent motions in the magnetically - active disk atmosphere overshoot the dead zone boundary by only about one scale height .
second , grains deep in the dead zone oscillate vertically in wave motions driven by the turbulent layer above , but on average settle at the rates found in laminar flow , so that the interior of the dead zone is a particle sink and the disk atmosphere will become dust - depleted unless resupplied from elsewhere .
third , with sufficient depletion , the dead zone is thinner and mixing dredges grains off the midplane .
the last of these processes enables evolutionary signatures such as the degree of settling to sometimes decrease with age .
the mhd results also show that the magnetic activity intermittently lifts clouds of small grains into the atmosphere .
consequently the photosphere height changes by up to one - third over timescales of a few orbits , while the extinction along lines of sight grazing the disk surface varies by factors of two over times down to a tenth of an orbit .
we suggest that the changing shadows cast by the dust clouds on the outer disk are a cause of the daily to monthly mid - infrared variability found in some young stars . |
interacting galaxies are known to trigger bursts of star formation , although this statement is toned down when considering selection biases and the weak contribution of mergers relative to main sequence galaxies to star formation in the universe @xcite .
nevertheless , a merger event remains a major milestone in the star formation history of a given galaxy .
since mergers provide a wide range of physical conditions for star formation , it is likely that they also account for the diversity of observed dense stellar systems , from associations to globular clusters and possibly ultra compact dwarf galaxies .
what type of stellar objects can be formed in a merger , as opposed to a more regular formation in an isolated galaxy , and under which conditions , remains poorly understood . for several decades
, simulations have been an efficient tool to study the morphology of interacting galaxies and mergers , and comparing the physics of their star formation with isolated cases .
studies first determined the tidal origin of the tails and the influence of orbital parameters on the morphology and dynamics of the collision ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
hydrodynamical simulations established a link between pericenter passages and bursts of star formation ( lasting a few @xmath0 ) and have began to explore the properties of the interstellar medium ( ism ) and the formation of massive star clusters in mergers @xcite . because of technical limitations , previous simulations were not able to fully probe the scale of star forming clouds ( @xmath1 ) .
however , the coupling between the kpc - scale dynamics and the physics of star formation has been established using high resolution simulations of isolated galaxies @xcite .
using simulations of mergers , @xcite underlined the dependance of the simulated star formation rate ( sfr ) with resolution , and thus the importance of resolving parsec - scale star forming structures .
however , the computational cost of considering two galaxies , their tidal features spanning huge volumes and the highly fragmented nature of the ism prevented them to reach numerical convergence in their estimate of the sfr .
( their simulation reached a resolution of @xmath2 , and did not include stellar feedback . ) other numerical works focussed on the formation of star clusters in mergers , highlighting the possible formation of very massive clusters in the nuclear regions of remnants or during the collision stages @xcite , emphasising relations between the radial position of clusters and their internal properties like metallicity @xcite and age @xcite , and proposing mergers to be one formation channel of massive globulars @xcite .
these pioneer simulations faced the technical challenges of resolving star clusters and probing their intermediate mass regime ( @xmath3 ) . with the present day computational ressources , we are able to push these limits to the parsec scale , smaller masses and account for more detailed physics . the antennae galaxies ( ngc 4038/39 , arp 244 ) have received a particular attention , both from the observational and the numerical sides , being the closest major merger in the starburst phase ( @xmath4 , @xcite , see also @xcite ) . several regions , and not only the galactic nuclei , host the formation of very massive stellar objects ( @xmath5 , @xmath6 , @xcite ) . in total , more than 1000 star clusters have been detected in the central @xmath7 of the antennae .
the richness of this system and the profusion of observational constraints make the antennae an excellent benchmark for theories and simulations on the role of merger in ( re-)shaping the structural properties of galaxies and of their stellar populations . in this paper
, we present a hydrodynamical simulation of the antennae galaxies , at @xmath8 resolution , including star formation and stellar feedback ( photo - ionisation , radiative pressure and supernovae ) .
the gain from previous works in term of resolution and physics implemented allows us to better describe the formation and early evolution of stellar systems by probing smaller objects and resolving the inner structure of the most massive cases . @xcite
used this simulation to establish a relation between the compressive nature of the tides and the interstellar turbulence in mergers and the enhancement of star formation . here , we focus our analysis on the properties of the stellar populations formed during the several stages of the interaction .
we first present the numerical techniques adopted ( section [ sec : numerics ] ) and the model of the antennae ( section [ sec : model ] ) .
section [ sec : classification ] details the classification of the stellar objects detected in the simulation .
their formation and early evolution is presented in section [ sec : formation ] .
the cluster mass function is discussed in section [ sec : cmf ] .
section [ sec : variety ] compares the properties of the stellar objects with observations and the formation history of a few objects is followed across time to provide possible scenarios for the formation of equivalent systems in the reality .
many aspects of this simulation are being analysed to address questions about the nature of the interstellar medium and to compare the properties of the modelled systems to the real antennae galaxies adopting an observational perspective .
these will be presented in forthcoming contributions ( bournaud et al .
, in prep .
, duc et al . , in prep . ) .
the simulation has been run using the adaptive mesh refinement ( amr ) code ramses @xcite . the overall method and
parameters are comparable to those used in the simulation of the milky way by @xcite .
the main difference is the use of a thermodynamical model including heating by ultraviolet radiation of cosmic origin , and atomic cooling tabulated at solar metallicity by @xcite , as in @xcite .
the minimum allowed temperature is set to 50 k to prevent numerical artefacts @xcite .
the use of such heating / cooling scheme allows for deviations from the otherwise simpler polytropic equation of state of @xcite , in particular at low densities and high temperatures , a common situation in galaxy mergers where tidal interactions can eject a significant fraction of the ism from the galactic discs .
gas denser than @xmath9 is converted into stellar particles ( called stars hereafter , for simplicity ) of minimum mass of @xmath10 , with an efficiency of 2 per cent per free - fall time .
this sets the star formation rate to @xmath11 before the interaction .
the newly formed stars inject feedback energy in the form of photo - ionisation in hii regions , radiation pressure @xcite and type - ii supernova thermal blasts @xcite .
to speed - up the computation , only one out of ten young stellar particles ionises its surrounding ism , but with an energy ten times higher than a single source . since stars form in a clustered way in the dense gas regions , each of these regions hosts at least one ionising source , such that our simplification does not alter the global energy budget at the scale of molecular clouds .
the refinement strategy adopted is `` quasi - lagrangian '' ( as in @xcite ) : an amr cell is refined when it contains more than a given number ( @xmath12 ) of particles ( dark matter and stars from the initial conditions ) , or when its baryonic density ( stars and gas ) exceeds a certain value ( which obviously depends on the refinement level of the cell ) .
furthermore , the code ensures that the jeans length is always resolved by at least four cells .
the maximum resolution is @xmath8 in the most refined volumes of the @xmath13 computational domain .
the dark matter and stellar components from the initial setup ( as opposed to the stars formed during the simulation ) of each of the two galaxies are rendered with @xmath14 millions particles and evolved using a particle - mesh technique at the resolution of @xmath15 .
these components provide a `` live '' background potential on top of which the ism and the young stellar objects can evolve .
the stars formed from the gas are evolved using the resolution of the amr grid , i.e. up to @xmath8 . due to the softening of the gravitational force on the stars ,
the two - body interaction is not properly described in our simulation and related processes could not be probed .
for instance the internal evolution of star clusters driven by two - body relaxation is not accounted for in this work .
we argue later that the relaxation time of such objects is much longer than the time - lapse considered in our analysis , i.e. mainly the duration of the galactic interaction ( see section [ sec : cfr ] ) .
the simulation has been run for eight million core hours on 4096 cores of the supercomputer supermuc hosted at the _ leibniz - rechenzentrum_.
lcc & ngc 4038 & ngc 4039 + + x position [ kpc ] & -15.0 & 10.6 + y position [ kpc ] & 30.4 & -30.3 + z position [ kpc ] & -46.3 & 46.7 + x velocity [ km / s ] & 26.0 & -26.9 + y velocity [ km / s ] & -23.3 & 23.2 + z velocity [ km / s ] & 71.3 & -71.8 + x spin axis & 0.65 & 0.67 + y spin axis & 0.65 & -0.71 + z spin axis & -0.40 & 0.20 + total mass [ @xmath16 & 3.10 & 3.10 + + mass [ @xmath17 & 6.5 & 5.0 + characteristic radius [ kpc ] & 6.5 & 3.9 + truncation radius [ kpc ] & 16.0 & 10.5 + characteristic height [ kpc ] & 0.4 & 0.4 + truncation height [ kpc ] & 1.5 & 1.2 + + number of particles & 1,006,000 & 1,006,000 + mass [ @xmath17 & 47.8 & 47.8 + characteristic radius [ kpc ] & 3.5 & 4.0 + truncation radius [ kpc ] & 11.0 & 12.0 + characteristic height [ kpc ] & 0.35 & 0.35 + truncation height [ kpc ] & 1.0 & 1.0 + + number of particles & 372,000 & 372,000 + mass [ @xmath17 & 17.6 & 17.6 + core radius [ kpc ] & 0.8 & 0.8 + truncation radius [ kpc ] & 2.5 & 2.5 + + number of particles & 1,000,000 & 1,000,000 + mass [ @xmath17 & 237.6 & 237.6 + core radius [ kpc ] & 10.0 & 10.0 + truncation radius [ kpc ] & 35.0 & 35.0 + the present simulation has been tailored to reproduce the morphology and kinematics of the antennae galaxies ( ngc 4038/39 , arp 244 ) .
it is based on the initial parameters from @xcite , found with a try - and - see approach where the quality of the model is estimated by eye , and not relying on numerical techniques to optimise the match with observational data ( for such studies , see e.g. @xcite , @xcite ) .
this model can be used to analyse the physical processes of similar prograde - prograde major mergers , as opposed to other configurations producing different features and hosting different populations ( see e.g. @xcite ) . for simplicity
, we adopt the naming of the real antennae when referring to our simulated galaxies : ngc 4038 is the northernmost galaxy at present time , associated with the southern tidal tail . since we observe the antennae between their second and third pericenter passages , and that the two galaxies switch from north to south at each passage , ngc 4038 is also north initially ( i.e. before the first collision ) .
the initial parameters for the progenitors and their orbit are given in table [ tab : init ] .
the two progenitor galaxies have been initialized as disc galaxies , with almost exactly the same mass .
the dark matter halo follows a @xcite profile truncated at the radius when its reaches five times the stellar disc mass .
the total baryonic fraction is eight per cent .
the stellar component is made of a @xcite bulge which represents 27 per cent of the stellar mass , and an exponential disc .
the initial gas fraction is @xmath18 per cent .
we ensure that no non - axisymetric structure like a bar or spirals is formed in the discs before the first collision .
such structures may create a phase reference point which would play an important role in the shaping of the tidal features , and the transfer of angular momentum within the central regions of the discs .
we note however that the collision itself accelerates the formation of a bar in both our discs , as often seen in simulations of interacting galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the orbit is comparable to that of @xcite and @xcite , although the progenitors are initially separated from a larger distance than in these previous works .
our simulation is stopped at the beginning of the final coalescence of the galaxies .
following the evolution for a longer time would be numerically costly and may require to account more precisely for cosmological aspects like the environment of the pair @xcite . in the following , stellar objects in the galaxy merger are compared to those detected , using the same methods , in the simulation of the milky way ( mw ) presented in @xcite .
the maximum resolution of this simulation is @xmath19 , i.e. 32 times higher than that of the antennae .
however , only a few myr have been simulated at this resolution . to ease the comparison with the antennae , we use a slightly earlier snapshot from the mw , when the resolution is set to @xmath8 ( i.e. @xmath20 before the instant pictured in the figure 4 of @xcite ) , but already when the large - scale structures ( spirals and bar ) of the galaxy are in place and host star formation .
the two simulations differ in term of the cooling scheme adopted : the milky way simulation uses a piecewise polytropic equation of state while ultraviolet heating and atomic cooling are invoked for the antennae run .
@xcite demonstrated that the difference between the two schemes is negligible , at least in the dense , star - forming gas .
therefore , the physical properties of the stellar objects from both simulations are directly comparable . the reference time @xmath21 corresponds to the first pericenter passage , occurring @xmath22 after the start of the simulation .
a long pre - collision phase is required to eliminate possible artefacts in the initial setup of the galactic discs , and for the galaxies to start separated by a sufficient distance so that they would not experience strong tidal effects at setup .
[ fig : morphology ] shows a sequence of snapshots extracted at various time along the collision .
contrary to the tidal tails which do not host much star formation , the gaseous component of the tidal bridge ( i.e. the elongated structures of tidal origin , connecting the two galaxies ) rapidly fragments into dense clumps .
self - gravity takes over the strong extensive ( i.e. destructive ) tidal field , the clouds collapse and further form stars in a compact , clustered fashion .
the clustered nature of star formation , followed by the elongation of these structures by tidal and orbital effects is visible in the outer parts of the progenitors .
some of the gas clouds and star clusters remain close to the gravitational lagrange point of equilibrium between the two galaxies , but most of them fall back at high speed ( up to @xmath23 ) toward one or the other progenitor .
note that this process allows for material from one disc to be accreted onto the other .
although this has little importance in our simulation , it may be a key aspect for mixing the gaseous and stellar material when very different galaxies are involved in the merger ( e.g. non - coeval or with different metallicities , sfrs etc . ) .
[ fig : morpho_gas ] shows the large scale morphology of the gas component when it best matches the observations of the antennae .
this instant is reached at @xmath24 , i.e. @xmath25 after the second pericenter passage and the moment when the two discs start to overlap ( duc et al .
, in prep . ) , and
a few myr before the third ( and last ) passage .
the simulation is run until the beginning of the final coalescence
. the much higher level of fragmentation of the ism at this late stage , compared to the pre - interaction epoch , increases significantly the numerical cost of the simulation , such that it becomes very expensive to pursue the integration of the merger remnant . because of the difficulty to distinguish density centers just before coalescence ) , and sfr during the interaction ( @xmath21 marks the first pericenter passage ) .
bursts of star formation are associated to the pericenter passages ( vertical lines ) and the beginning of the final coalescence ( @xmath26 ) .
this curve has been smoothed with a @xmath27 wide gaussian window , to eliminate the high frequencies , for the sake of visibility .
the letters a , b , c and d label several epochs used later in our analysis , as in fig .
[ fig : morphology ] . the best match with the observations
is reached at @xmath28 , and yields a sfr of @xmath29 . ]
[ fig : sfr ] displays the evolution of the sfr along the interaction . as noted by numerous previous studies ,
episodes of intense star formation occur at the times of pericenter passages and at the beginning of the final coalescence of the two galaxies @xcite .
the details of the bursts ( shape , intensity ) depend on the physics resolved by the simulation , in particular the ability to capture the turbulence cascade @xcite . using the simulation presented here and the same setup at lower resolutions , we have shown in @xcite that the global sfr becomes independent of the numerical resolution when it reaches @xmath30 .
we also explained that compressive tides generated by the galactic encounters inject turbulent energy into the ism .
in such turbulence , the compressive component overtakes the solenoidal one , which creates an excess of dense gas , leading to an intense episode of star formation .
this process triggers star formation not only in the central regions of galaxies but also over extended volumes , several kpc away from the galactic nuclei .
after the first starburst , the sfr drops by a factor of 2.5 ( @xmath31 ) before rising again within @xmath6 , and slowly declining as the progenitors separate .
this drop corresponds to the end of the star formation in the tidal debris , in particular in the tidal bridges between the two galaxies .
this results from the lowering of the gas density in these regions because of both consumption to form stars , and dynamical evolution of the cloud leading to its elongation and dilution of gas .
large volumes within the discs still host star formation , which maintains a global sfr @xmath32 . a few myr later , tidal debris from the bridges and the tails fall back into the galaxies , either collide with other gaseous structures or simply get compressed along their orbit .
this fuels a secondary episode of star formation , at a lower rate than the initial starburst : @xmath33 on average , with short and localized bursts reaching @xmath34 .
this process continues during the entire separation of the galaxies and the global sfr decreases to @xmath35 just before the second pericenter passage .
the second encounter also triggers a rapid rise of the sfr , but at this stage , it becomes involved to identify the peaks of the global sfr and connect them to their physical cause .
dynamical friction prevents the galaxies to separate significantly after the second passage : the maximum distance between the nuclei is @xmath36 , i.e. less than the radii of the stellar discs .
in other words , the two gas reservoirs continuously overlap for @xmath37 .
gravitational torques drive inflows of gas toward the nuclei @xcite . at the same time , collisions between gas structures ( spiral arms , clumps ) occur more frequently than at the first passage because of a smaller impact parameter @xcite . these processes , in addition to the compression by tides and turbulence @xcite , trigger a new starburst episode .
the peak of the sfr corresponds to the merger of the two nuclei , leading to a short ( @xmath6 ) but intense ( up to @xmath38 ) star formation episode .
this point is further discussed in section [ sec : cfr ] .
the sfr curve in fig . [ fig : sfr ] already indicates the very different nature of the two main bursts . in our setup , both bursts of star formation yield , at first order , an exponential decline with characteristic timescales of @xmath39 and @xmath40 respectively .
furthermore , the full burst episode , i.e. the rather sharp rise of the sfr followed by a slower decline can be roughly fitted by a log - normal functional form . in that case
, the skewness of the function is about three times larger for the first burst than for the second .
this shows the differences between ( i ) the long - range effect of the tidal compression at the first burst , ( ii ) the slow decrease of the sfr during the separation of the progenitor , and ( iii ) the direct collision between the discs and their nuclei at the second and third encounters . after the third passage ,
the nuclei start to coalesce .
the damped modulation in the peaks of the sfr for @xmath41 corresponds to the oscillations of the two nuclei around each other , while dynamical friction makes them loose orbital energy , making each nucleus - nucleus passage less and less efficient at triggering a central starburst .
note that the same behaviour is commonly found in the mass fraction in compressive tides @xcite .
although the numerical cost of the simulation prevents us from evolving the system for a long post - coalescence period , a complementary run at the resolution of @xmath42 indicates that the sfr resumes its pre - interaction level at @xmath43 , i.e. @xmath44 after the last starburst .
therefore , when including the rise of sfr just before the first passage , the full episode of enhanced star formation lasts @xmath45 .
the best match with observations being reached at @xmath46 , we note that the sfr has declined and increased again before reaching the present day , by a factor up to @xmath47 during the last @xmath48 , which contradicts the assumption of a constant sfr made by @xcite to interpret the observed age distribution of clusters in the antennae .
a longer time - lapse between the pericenter passages would allow the sfr to reach a relatively constant value after the first starburst and would thus be more consistent with the assumption of @xcite . in our simulation ,
the sfr averaged over the last @xmath48 is higher than the present - day value .
therefore , assuming a constant sfr based on the observed value leads to an underestimate of the dissolution rate of clusters in that age range .
this would then support the idea of @xcite and @xcite that the age function of star clusters is mainly due to their dissolution and not their formation rate , but not for the reasons these authors claim .
however , the lack of observed older clusters ( @xmath49 ) in the antennae is explained by all models by the low sfr ( @xmath50 ) before the first encounter , i.e. by a formation process rather than a disruption one . table [ tab : sfr ] indicates the stellar mass formed during the three major phases of the merger . despite their physical differences , the main two starburst episodes produce the same amount of stars , in comparable time - lapses .
furthermore , the phase of separation hosts a non - constant yet continuous formation of stars , producing 1.5 times more stellar mass than each burst , although it does not count any major starburst episode itself .
in other words , the interaction produces more stars during the separation than during each individual starburst episode . ) , compared to the global sfr . extended star formation exists throughout the merger , except at the second starburst episode ( associated with the third pericenter passage and the final coalescence , @xmath51 ) when it is almost exclusively central , due to gas inflows .
] fig .
[ fig : extended_sf ] shows the spatial extent of star formation during the merger .
gravitational torques generate inflow of gas toward the central region throughout the merger .
however , as suggested in @xcite , extended star formation exists before the pericenter passages themselves ( @xmath52 and @xmath53 ) , being triggered by compressive tides and compressive turbulence , in addition to the nuclear inflows .
the main starburst occurring at the beginning of final coalescence ( @xmath54 ) is almost exclusively central .
such variations in the spatial distribution of star formation reveal once again the different nature of the two starburst episodes .
we explore this point in term of formation of star clusters in section [ sec : cfr ] .
.stellar mass formed along the merger [ cols="<,^,^",options="header " , ]
and @xmath55 . the average of the two means ( dashed line , @xmath56 ) is the velocity dispersion criterion used to tell appart the two populations . ] for the analysis presented here , the dense , cluster - like objects formed during the simulation are detected thanks to the friend - of - friend algorithm hop @xcite , around peaks of local stellar density @xmath57 .
sub - clumps are merged if the saddle density between them exceeds @xmath58 and an outer boundary of @xmath59 is set to tell apart clump members from field stars and tidal debris . for simplicity
the stars present in the initial conditions and the gas are not taken into account in the computation of the densities .
the output of the detection method provides the membership of all stars to either an overdensity or the field .
the properties of the objects , like the mass and the half - mass radius are then derived . using a different set of density criteria
slightly changes the estimate of the properties of the detected objects , but does not affect our conclusions . modifying these by an order of magnitude would lead to split some clusters or , on the contrary , to a too severe truncation of viable objects . by using a criterion based on the density only , and not involving kinematical information ,
our method allows us to detect unbound objects . in order to make a difference between genuine star clusters and unbound structures which might be transient over - densities , we adopt additional selection criteria based on the velocity dispersion of the stellar object . fig . [
fig : veldisp ] shows the distribution of internal velocity dispersion of all the stellar objects detected by the friend - of - friend algorithm at the final coalescence ( epoch d ) .
this velocity dispersion @xmath60 is the standard deviation of the three - dimensional velocities of the stars contained within the half - mass radius of the object .
two populations appear : the majority ( 80 per cent ) of the objects have a high velocity dispersion , of the order of the galactic - scale motions like the rotation speed of the disc or the relative velocity of the two galaxies , i.e. @xmath61 . the other objects have much lower velocity dispersions .
we tell apart the two populations by fitting the global distribution with a sum of two gaussian functions .
the average of the means of the two gaussians ( @xmath62 ) is then used as a selection criterion . in the mass - half mass radius plane ( @xmath63 ) shown in fig .
[ fig : selection ] , almost all the objects with a high velocity dispersion ( @xmath64 ) lie on a sequence ( @xmath65 ) running over the full mass range we probe .
these objects are transient structures in which the stars enter and leave within a few myr , in a similar fashion as density waves . because of the mixing of their content
, these structures count stars with a wide diversity of ages , that have formed under different conditions and environments , populate the field , and eventually are temporarily gathered in such over - densities . )
plotted against velocity dispersion ( @xmath60 ) for all the stellar objects detected in the simulation . after removing the objects with high velocity dispersions ,
we split the remaining into two classes and use this identification in the mass - size diagram of the bottom panel where they are still well - separated.,title="fig : " ] ) plotted against velocity dispersion ( @xmath60 ) for all the stellar objects detected in the simulation . after removing the objects with high velocity dispersions ,
we split the remaining into two classes and use this identification in the mass - size diagram of the bottom panel where they are still well - separated.,title="fig : " ] by estimating the dimensionless virial ratio ( @xmath66 ) , we can split the objects with low velocity dispersion into two families ( fig .
[ fig : virial ] ) .
this also corresponds to the separation visible in the mass - size diagram , on both sides of the sequence , as noted above . note that the softening of the gravitational potential at small scale ( on the resolution of the grid , i.e. @xmath8 ) introduces a bias tending to stabilize small stellar structures against collapse . for these reasons
, we over - estimate the potential energy and thus , underestimate the virial ratio of the small objects . nevertheless , the separation made in fig .
[ fig : virial ] tells apart bound objects ( class c1 , low virial ratio ) from unbound ones ( class c2 ) , even at small radii , on top of the separation from structures with a high velocity dispersion ( class c3 ) already made . :
low velocity dispersion and low virial ratio ( bottom left , class c1 ) , low velocity dispersion and high virial ratio ( top left , c2 ) and high velocity dispersion ( right , c3 ) .
the mass and half - mass radius of the object are indicated in each panel .
the scalings ( size , bin size and color ) are the same for all three panels . ]
but for three objects with the same half - mass radius , and comparable masses .
the object from the c2 class is the same as the one shown in fig .
[ fig : clus_morpho ] . ] fig .
[ fig : clus_morpho ] and fig .
[ fig : clus_morpho2 ] illustrate the differences in morphology between these three classes of objects . even if all are over - densities of young stars , only the c1 objects ressemble star clusters , with a clear central peak density and a spherical shape .
the c2 class seems equivalent to ( massive ) associations , while the c3 structures are transient over - densities of field stars , with no particular features or sub - structures .
the classes c1 and c2 represent well the diversity of binding of stellar objects observed by e.g. @xcite . a more detailed comparison to observed objects
is made in section [ sec : obs ] . in the following
, we will refer to c1 objects as star clusters and to c2 objects as associations .
the bi - modality in velocity dispersion does not exist throughout the interaction : during the earliest phases ( before and during the first encounter ) , almost all the objects detected have a velocity dispersion smaller than @xmath67 .
the class c3 does not exist at these epochs .
after the first starburst , the bi - modality appears and the high velocity dispersion mode slowly shifts towards higher and higher values .
therefore , we adjust the value of our criterion @xmath68 as a function of time to separate the c3 structures form the c1 and c2 objects . .
the colour represents the velocity dispersion . ]
[ fig : mr_evol ] shows the distribution of star clusters and associations in the mass - size plane at the four epochs highlighted in fig .
[ fig : morphology ] and [ fig : sfr ] , with colour indicating the velocity dispersion .
at all epochs , we detect sub- and super - virial objects ( classes c1 and c2 ) . these two classes are well separated most of the time ( using the same criterion than in fig .
[ fig : virial ] , at the epoch d ) , but the distinction is less clear during the first starburst ( pictured here at epoch b ) . at this epoch ,
the extreme conditions of the galactic collision in term of sfr and tidal effects probably create objects with intermediate virial ratios and/or shift existing objects to this regime .
evolution processes make them collapse , expand or even dissolve and thus re - creating the clear cut between the c1 and c2 classes .
since the criterion on the virial ratio adopted at final coalescence applies also during earlier stages , we choose to use it for all snapshots .
in the rest of the paper , we focus the analysis and discussion on star clusters ( c1 ) .
, which explains the differences in absolute values with that of the fig .
[ fig : sfr ] . only c1 objects are considered . ] using the detection and diagnostics presented in section [ sec : classification ] , we identify the star clusters ( c1 only ) on snapshots every @xmath69 .
the time distribution of the first detections of clusters then gives us the cluster formation rate ( cfr ) , shown in fig .
[ fig : cfr ] , and compared to the sfr recomputed with the same time binning ( i.e. @xmath20 ) .
obviously the cfr only traces the mass of clusters at the epoch of their formation , ignoring their future growth or evaporation . as discussed in section [ sec : sfr ] , after @xmath31 , star formation is not active anymore in the tidal debris , but limited to the central parts of the discs .
the cfr suddenly drops by a factor @xmath70 showing indeed that star formation only proceeds in already formed clusters , and in the galactic nuclei .
this remains valid during the separation of the galaxies , and even for the second major starburst episode ( @xmath71 ) meaning that the vast majority of clusters are formed by the first galactic encounter . ) .
the colour map represents the strength of the tidal field ( computed as the density weighted average of the largest eigenvalue of the tidal tensor at the scale of @xmath72 , see @xcite ) , running from no tides ( white ) to strongly compressive tides ( red ) .
contours mark the regions experiencing converging compressive turbulent flows , and the blue dots indicate the position of the young star clusters ( @xmath73 ) . ]
@xcite established a relation between compressive tides the importance of which significantly rises in mergers , compressive turbulence and star formation .
[ fig : compressive ] illustrates the spatial correspondance between the three aspects , short after the first pericenter passage .
although a delay of @xmath74 exists between the initial tidal trigger and its effect on the star formation rate , we confirm that almost all young star clusters are found inside tidally and turbulently compressive regions .
such large - scale effects trigger star ( cluster ) formation over spatially extended volumes , and not only in the galactic nuclei . however , the shorter impact parameters of the second and third encounters imply a more important role of gas inflow onto the nuclei
. therefore , star formation occurs mostly in the nuclei and not in a number of star clusters at various distances from the centre as it does during the first pericenter passages . at @xmath75 , 82 per cent of the young stars ( @xmath76 )
form in the central @xmath77 of the merger , in the galactic nuclei , and thus do not form new clusters : the cfr remains low despite a burst of the sfr .
note that the non - equality between the cfr and sfr does not invalidate the fact that all stars form in clusters : in our simulation , stars can form in pre - existing clusters ( or in the galactic nuclei ) , such that the sfr remains high despite a zero or small cfr .
to further investigate the formation of clusters , we define three quantities : ( 1 ) the epoch when a cluster has formed 10 per cent of its final mass , i.e. a proxy for the beginning of its star forming episode , ( 2 ) the epoch when it has formed 90 per cent of its mass , i.e. a proxy for the end of its star forming episode , and ( 3 ) the difference of the two , i.e. the duration of the star forming episode . setting our thresholds to 10 and 90 per cent instead of 0 and 100 avoids the contamination by isolated star forming episode in the lowest mass clusters , where the star formation history is sensible to poisson s noise .
[ fig : mr_duration ] displays these three quantities for the clusters detected at the epoch d. the absence of relation between the mass and the epoch of formation reveals that massive objects are formed throughout the interaction process .
it however exists an age - half mass radius relation : the most extended object formed earlier .
this could indicate either an evolution pattern , namely the expansion of the clusters , or general properties of the cluster they maintain since their birth .
[ fig : mr_evol ] reveal that in the early stages of the merger , young clusters can be detected with a relatively large radius ( @xmath78 ) before any internally - driven evolution takes place .
therefore , the extended nature of the clusters formed early is a signature of the conditions of their formation .
the interaction and the associated compression by tides and turbulence @xcite tend to form dense but small gas clumps .
the star clusters formed in these clumps inherit their physical properties .
clusters formed before a pericenter passage are born in a less fragmented ism than those born after : the fraction of dense gas ( @xmath79 ) is multiplied by a factor 5 to 10 by the interaction , as visible in the density probability distribution functions of ( * ? ? ?
* their fig .
the mean duration of the star forming episode is @xmath80 , but the distribution peaks at @xmath81 indicating that most stellar objects have formed during a single , short episode and are thus coeval .
a few clusters however yields longer star formation durations , up to @xmath48 .
such distribution of star formation histories is compatible with the estimates of @xcite who observed the central regions and the tidal features of the interacting galactic pair arp 284 and found a majority of short formation episodes and only a few @xmath82-long cases .
our findings however contradicts the observations of ymcs in the merger remnant ngc 343 by ( * ? ? ?
* see also @xcite ) who advocate a single , short ( @xmath83 ) formation event . as proposed by @xcite , the difference of galactic environment , which in these cases is mostly due to different merger stages after the first pericenter passage and before the second one @xcite , while ngc 34 is in the coalescence phase . ]
could play a role on the star formation histories of clusters . in our simulation ,
long , extended star formation histories are split into several star formation episodes separated by @xmath84 . the episodes producing most of the stars in a given cluster are associated with the first and second galactic pericenter passages .
( the details of this timing vary depending on the exact position of the clusters in the galaxies when they collide : clusters in the outermost regions of the galaxies form their second stellar population before the galactic pericenter passage itself . ) therefore , the interaction acts as a global trigger to form star clusters , but also their secondary stellar populations .
we also note that the most massive clusters gain a fraction of their mass by accreting smaller objects . since these smaller objects also preferentially formed during the global starburst , the resulting stellar population in the final cluster yields a small spread in age , shorter than the duration of the starburst , but not two distinct populations .
an analysis of the chemical abundances of such objects ( impossible in our simulation ) may unveil the complexity of their star formation histories .
as mentioned before , computing the gravitational potential on a grid ( particle - mesh method ) and softening the gravitational potential implies that the collisional aspect of the star clusters and thus , the full dynamical evolution can not be probed accurately .
simulations accounting for this do exist but the price for accuracy is a strong limitation in the number of bodies ( i.e. stars ) included . simulating the @xmath85-body driven evolution of several @xmath86 collisional objects remains therefore out of reach .
one can argue that such evolution for clusters in the mass range we consider is very slow compared to the duration of the merger , and thus can be neglected in a first approach .
( core - collapse for such clusters would take @xmath87 , see @xcite . ) other drivers , related to @xmath88 long evolution , are accounted for in our simulation , at least to some extent ( gas accretion , tides , relaxation ) .
the same detection method and analysis have been applied to the mw simulation @xcite . at the instant chosen ,
the mw counts 341 star clusters ( c1 ) , and 21 associations ( c2 ) .
all objects yield velocity dispersion below @xmath89 ( the mean value being @xmath90 ) .
[ fig : mr_mw ] compares the distributions of the star clusters in the simulations of the mw and the antennae in the mass - size diagram .
we found clusters spanning several decades in mass and @xmath91 dex in half - mass radius , with no correlation between the two quantities , and mostly overlapping with the loci of the observed milky way clusters ( @xmath92 , @xmath93 , see section [ sec : obs ] ) .
furthermore , no global correlation is found between the mass ( or size ) of the clusters and their position within the simulated disc .
however , the objects with the highest velocity dispersion are found close to the tip of the bar , and in the portion of spiral arm undergoing fragmentation .
overall , although the cluster populations of the two galaxy models are comparable in many aspects , we note several divergences .
first , contrary to the antennae , the mw simulation does not contain very massive clusters ( @xmath94 ) .
furthermore , the average mass of clusters in the antennae is 15 times that in the mw .
this point is further discussed in section [ sec : cmf ] .
second , the antennae clusters are generally more extended than those of the mw ( by a factor 2 , in average ) . in the end , the cluster in the merger are @xmath95 times denser than in the mw .
third , none of the mw clusters is found in massive and extended regime ( @xmath96 , @xmath97 ) whereas an handful of cases exists in the antennae , clearly separated from the bulk of the distributions .
this reflects the effects of the rapidly varying environment of these clusters ( see section [ sec : w3 ] ) .
vertical markers indicate the average mass of the clusters for each distribution .
( the histograms are slightly shifted horizontally with respect to each other , for the sake of clarity . ) ] fig .
[ fig : cmf ] shows the cluster mass function ( cmf ) at the four stages marked in fig .
[ fig : sfr ] . in the antennae , before the first sfr peak ( epoch a ) , the long - range interaction has already triggered the formation of a handful of clusters . at the first post - starburst epoch ( b ) , following the peak of cfr ( recall fig .
[ fig : cfr ] ) , the number of clusters has been multiplied by @xmath70 .
the tidally triggered fragmentation of the ism leads to the formation of many @xmath98 clusters and also very massive cases , up to @xmath99 .
the average mass of clusters doubles with respect to the isolation stage , as noted observationally by @xcite . at the end of the separation of the galaxies ( epoch c ) ,
the cmf is comparable to that of the previous stage for @xmath100 but the number of low - mass objects decreases .
a deficit of intermediate mass clusters ( @xmath101 ) appears .
finally , during the second starburst and at the final coalescence ( epoch d ) , a few very massive ( up to @xmath102 ) objects are detected in the central region of the merger , increasing the average cluster mass by a factor 1.8 .
the deficit of intermediate - mass clusters is conserved .
such deficit leads to a major difference between bell - shaped simulated cmf and the observed one ( see e.g. * ? ? ?
* who detected a power - law shape down to @xmath103 ) .
since our numerical method allows for the formation of intermediate mass clusters ( see the next section ) , we do not think that the difference principally arises from a numerical artefact , but rather from physical conditions .
it is however possible that interaction with dark matter particles could destroy small clusters .
the mass ratio between dark matter and stellar particles is @xmath104 in our simulation .
such a high ratio could induce strong effects due to two - body interaction @xcite , in a comparable fashion as the destruction of low - mass clusters by collisions with giant molecular clouds @xcite .
unfortunately , decreasing this mass ratio would require a very large number of dark matter particles ( @xmath105 ) and thus a much higher numerical cost , which prevents us to quantity this effect .
the cmf in the milky way model yields a log - normal shape ( see fig . [
fig : cmf_mw ] ) .
the most massive cluster in the mw simulation has a mass of @xmath106 and is found at the tip of the galactic bar , where gas clouds in the bar interact with those from the spirals arms to form giant molecular associations and massive young clusters .
cloud - cloud collisions are known to trigger intense , localized episodes of star formation in such environment ( @xcite , see also @xcite about collisions in other environments ) .
cloud - cloud collisions are also expected in galaxy mergers , in particular at the coalescence phase .
unfortunately , the eulerian nature of our simulation makes it difficult to follow the trajectories of gas clouds and estimate their collision rate , in particular in a fast - evolving systems like the antennae . without being able to quantify precisely their contribution to the formation of clusters , we note that the cloud - cloud collisions provide sufficient pressure to trigger the formation of massive clusters ( @xcite , @xcite , see also @xcite ) , and such conditions are found in some specific areas like the overlap region of the merger .
other factors like clouds being shocked by a density wave created by the galactic interaction and the compressive trigger introduced in @xcite also favor the formation of massive clouds and subsequently massive clusters . all together
, these processes explain the difference in the maximum mass of star clusters between an isolated galaxy and a merger .
a full comparison between the simulation and the real antennae galaxies is presented in duc et al.(in prep . ) . in this section ,
we use the simulation as a prototype of antennae - like mergers , without seeking a one - to - one correspondance with the antennae themselves .
the analysis is made using the data at the end of the simulation , at coalescence ( epoch d ) , i.e in the post - merger phase when it is most probable to observe interacting galaxies ( owing the short duration of the collision phases ) . during the close encounters of the galaxies , where the observational classification of systems as mergers is the most robust ,
star forming regions are generally strongly obscured by the overlap of the ism of two galaxies .
therefore , only dense , young and bright clusters are observationally detected . keeping this issue in mind , we analyse the resemblances and differences between our simulated objects and the compilation of observations from @xcite , to which we add published data specific to the antennae @xcite and to ngc 7252 ( a merger remnant , * ? ? ?
[ fig : mr_obs ] shows the positions of the observed and simulated stellar objects in the mass - size plane , assuming that the observed effective radius corresponds to the simulated half - mass radius . assuming that each stellar particle of the simulation fully samples the imf and represents , by definition , a coeval population , we evaluate its luminosity in the v band as a function of its mass @xmath107 and age @xmath108 : @xmath109 following @xcite .
we compute the intrinsic surface brightness of a cluster @xmath110 by summing the luminosities of its members and using the surface given by the half - mass radius : @xmath111 the simulated star clusters lie in the same loci in the mass - size diagram than the young massive clusters observed in the antennae ( fig . [
fig : mr_obs ] ) .
the most massive cases are comparable to clusters detected in the overlap region ( @xmath96 , @xmath112 , @xcite ) .
[ fig : mr_evol ] also tells us these clusters have been the last ones to form , in agreement with observational data .
the least massive end of the distribution of simulated clusters is comparable to the clusters detected in more external regions of the galaxies ( @xmath113 , @xmath114 , @xcite ) , compatible with an earlier formation .
the distribution is also compatible with that of the clusters found in another merger , ngc 7252 , by @xcite .
our clusters are offset from the bulk of the milky way and extragalactic globular clusters ( @xmath113 , @xmath42 ) , in both the observations and the simulations .
the simulation reproduces the global effect of a merger in shifting the distribution of star clusters toward more massive and slightly larger objects for the reasons discussed in section [ sec : cmf ] .
furthermore , the different physical conditions between the two starbursts ( tidal compression for the first one , and gas inflow for the second , see section [ sec : cfr ] and @xcite ) explain the spread of the distributions , from intermediate - mass , extended clusters formed early , and massive compact object formed late .
the most extended objects in the high - mass half of our distribution are in the regime of ultra - compact dwarf galaxies ( ucds ) and nuclear star clusters . the formation scenario and the properties of two of them
are discussed in details in section [ sec : w3 ] .
we note that the maximum mass of objects found in the simulation ( @xmath102 ) is higher than that detected in the real antennae galaxies , but is in good agreement with objects observed in a wider sample of environments @xcite .
the difference between our model and the antennae is therefore a lack of realism either in the physics implemented , and/or in the realism of the model itself ( initial conditions , orbit , etc . ) .
we expect the sub - grid physics and the neglected physical phenomena like magnetic field and two - body relaxation to affect mostly the objects made of a few stellar particles but to have a milder impact on the global properties ( mass , size ) of the largest cases over the timescales considered here .
interestingly enough , the physical setup leading an antennae - like system allows to produce most of the @xmath115 stellar objects observed , including an handful of the most massive ones ( see next section ) .
this suggests that real systems showing a small variation in their parameters with respect to the antennae galaxies could host the formation of massive objects , comparable to those seen in our simulation . among the dense stellar objects , ultra - compact dwarf galaxies ( ucds ) are at the frontier between star clusters and galaxies , such that their formation scenario is still debated
it has been proposed that ucds could be the remnant of a tidally - stripped dwarf galaxy ( see e.g. * ? ? ?
* ; * ? ? ?
however , @xcite argued that these objects lie in the most luminous and massive ends of the distributions of globular clusters which indicate the two families of objects share the same formation mechanism ( see also @xcite , zhang et al . in prep and liu et al . in prep . which propose opposite conclusions ) . the w3 stellar object , observed in the galaxy merger ncg 7252 and marked in fig . [ fig : mr_obs ]
is found between the regimes of the massive star clusters and that of the ultra - compact dwarf ( ucd ) galaxies ( @xmath116 , @xmath117 , @xcite ) . using numerical simulations
, @xcite proposed that w3 could form through merging of star clusters formed during a starburst episode .
these authors successfully found a set of parameters ( structural and orbital ) for a star cluster complex made of sub - clusters to reproduce the observed characteristics of w3 ( with however a discrepancy between the modelled luminosity profile and the observations , see @xcite ) , arguing that such configurations are commonly found in interacting galaxies .
our galaxy simulation can be used to complement their star cluster simulation and discuss the likelihood of their scenario . at the end of our simulation ,
several objects share the properties of observed ucds , and two in particular are comparable to w3 .
the first has a similar mass but is less extended ( @xmath118 , @xmath119 ) .
the gas cloud leading to this cluster forms in the outer region of ngc 4039 facing ngc 4038 , along with many clumps of comparable mass distributed along a @xmath120 long gaseous arm which fragments under the effects of the tidal interaction .
because of the rotation of the galactic disc , this region rapidly moves to the other side of the system ( within @xmath81 , at @xmath121 in the global reference frame ) , close to but not within the tidal tail .
the gas clumps remain relatively close to the galaxy center ( @xmath122 ) . during this process ,
the gas clouds gather in a small volume , in a comparable fashion as the piling - up scenario proposed in @xcite , and rapidly merge into a handful of clumps . our cloud of interest ( that will form the w3 analogue ) accretes gas from its host arm and form stars at @xmath123 during @xmath124 .
star formation stops when the surrounding interstellar medium becomes too diffuse to continue to fuel gas accretion onto the cloud .
the young cluster then keeps a constant mass , and its orbit keeps it close to the galactic center ( within @xmath125 ) during the entire simulation .
our second analogue is slightly less massive but more extended ( @xmath126 , @xmath127 ) .
it forms in a comparable environment than the first candidate , i.e. in between the galaxies , but in the other progenitor .
it also moves to the other side and remains fairly close to the galactic centre ( @xmath128 ) .
its gaseous nursery travels alternatively through dense and diffuse medium and experiences sequentially ram pressure and accretion of gas . through this sequence of events ,
the stellar object builds up during @xmath129 ( with a sfr peaking twice at @xmath130 during the first global starburst ) .
then , its mass remains approximatively constant until the end of the simulation .
this object is compact during most of its life ( @xmath131 ) . during the third pericenter passage
however , it travels through the central kpc of the merger .
the outer layers of the associated gas cloud ( i.e. between @xmath132 and @xmath133 ) are significantly altered : ram pressure and ( at a smaller extend ) tidal effects strip half of the gas mass within @xmath81 ( see fig . [
fig : w3_c5 ] ) .
however , the inner part ( within the half - mass mass radius of @xmath134 ) is dense enough ( @xmath135 ) , not dynamically affected and conserves its mass and density . after the crossing of the galaxy ,
the gravitational potential of the cloud is twice as shallow than before , and since the stellar object keeps approximately the same velocity dispersion , it becomes super - virialized ( see @xcite for comparable effects due to gas expulsion by stellar feedback ) .
thus , the cluster grows shortly after the crossing of the merger and its half - mass radius jumps from @xmath136 to @xmath137 within a half - mass relaxation time ( @xmath138 ) .
despite this rapid change , the cluster remains bound and survives . ) of the gas cloud associated with our second w3 analogue ,
a few @xmath139 before ( left ) and after ( right ) it flies through the galaxy merger .
arrows indicate the direction of the galactic centre toward which ( left ) and from which ( right ) the cloud is moving . within the @xmath140 between these snapshots , the outer envelope of the cloud has been stripped by ram pressure and tides .
the inner region has not been strongly affected , but the stellar counterpart starts to re - adjust to the new potential well with an increase of its half - mass radius by a factor 7 within the next @xmath40 . ] both our w3 analogues are free of dark matter and form in the outskirts to the discs , almost in the tidal bridges , during the first starburst , in a somewhat similar fashion than the hinge clumps detected by @xcite . with the analysis of the evolution of the cmf presented in section [ sec : cmf ] , we conclude that the conditions to form such massive and compact objects are only fulfilled during the starburst phases of a galaxy merger .
the two formation histories we have highlighted above ( @xmath40-long accretion of star forming clouds and violent relaxation of a dense object flying through the center of a galaxy ) favors the idea of w3-like objects in particular and ucds in general being the massive end of the distribution of star clusters and not tidally - stripped dwarf .
stripping is however advocated in the second scenario , but on the gaseous left - overs of a young massive cluster and not on a dwarf galaxy .
the early merger of clouds is comparable to the scenario proposed by @xcite , but we note that the clouds are not part of a single structure or cloud - complex .
the hierarchical flavor of their scenario ( following @xcite ) is not found in our simulation .
we note that the numerical dissipation of turbulence in our simulation at the parsec - scale could modify the way the ism fragments at these scales and bias our findings .
however , the hierarchical formation of a massive cluster has been detected elsewhere in the system ( see section [ sec : tails ] ) , indicating that the simulation allows such process to happen . in conclusion ,
antennae - like mergers provide the physical condition for the formation of young stellar objects as massive as w3 and the most compact cases of ucds .
their formation channel is comparable to that of massive clusters .
the age distribution of their stars is extended ( @xmath141 or @xmath129 ) but continuous . yet
, an additional process is needed to eject the udcs in intergalactic regions to account for their observed spatial distribution .
future galactic collisions , which are likely to happen in galaxy cluster environment but much less common in galactic pairs , could play this role . in our simulation
, the southern tidal tail does not contain any clustering of newly formed stars , but an elongated , rather diffuse distribution of a few field stars along the tail , over @xmath142 .
all these stars formed several @xmath143 before the first collision , i.e. before the tidal triggering of star formation and the formation of the tail itself .
the cluster in which they have been formed has been stripped and a fraction of its stars has been tidally ejected into the tail .
the gas content of the tail is too diffuse to be converted into stars and thus , it does not host _ in situ _ star formation .
oppositely , the northern tail does not show spare young stars but a single concentration into a dense stellar object , situated @xmath144 away from the merger remnant ( at epoch d ) .
this massive object ( @xmath145 ) is much more compact ( @xmath146 ) than classical tidal dwarf galaxies ( tdgs , @xcite ) , although one would expect that the weak tidal field in this region would allow for a higher radial expansion before reaching tidal truncation .
however , the duration of the simulation is too short compared to the expansion timescale of such a stellar system , and the tidal field is not strong enough , or has not acted long enough to have significantly heated the stars of this object and impacted its structure ( see * ? ? ?
the stellar population of this object is @xmath147 ( @xmath148 ) younger than the tidal tail itself , indicating an _ in situ _ formation .
furthermore , the spin velocity ( @xmath149 ) of this object is about five times higher than its velocity dispersion , indicating a net rotation , while other objects in the simulation show much less clear signatures of rotation , if any .
the presence of such object is compatible with the observations of the antennae , showing that , although a tidal dwarf galaxy could not form , parsec - size stellar object formed _ in situ _ exists in the tidal tails @xcite .
in the two tidal tails . the compressive tidal field
is shown in orange - red , and the extensive field is in cyan - blue .
the maps have been filtered to show only the dense regions , and hide the intergalactic medium . ]
although neither the real antennae nor our simulation contain any tdg , we can assume the key features of the scenarios accounting for their _ in situ _ formations also apply to the smaller object we detect @xcite . among them
, @xcite proposed that gas accumulates into a massive and dense clump near the tip of the tail and then collapses to form a young stellar object .
a slightly different scenario happens in our simulation : about @xmath150 after the first pericenter passage , the bend in the northern tail ( visible at a later stage in fig .
[ fig : morpho_gas ] ) is @xmath151 away from the center of ngc 4039 and is moving away from the galaxy .
the complex overlap of the potentials of the dark matter halos , the galaxy , and the tail itself creates a region of compressive tides over @xmath152 ( fig .
[ fig : tdg_tides ] ) .
such tidal field helps the increase of the gas density and form a gas clump .
the same process also takes place in the southern tail , but the tidal field there is about 100 times weaker and spans several small regions , instead of a large one .
the morphology of the tails ( and thus the relative inclination of the discs at the pericenter passage which rules the efficiency of the spin - orbit coupling ) seems to make most of the differences between the two tidal fields . in both tails , additional phenomena like low shear and the absence of kpc - scale streaming motions in this particular region
further help the gathering of gas , but a seed @xmath153 times denser than the surrounding tail material only form in the northern tail . further away from the disc than our seed
, the tip of the tail unfolds northward rapidly ( @xmath154 , because of its initial rotation speed from its progenitor disc ) . at the same time , the rest of the tail tends to fall back toward the galactic center , at comparable speeds but in the oposite direction , as visible in fig .
[ fig : tail_velocity].a . in between
, the seed lies in a region of low velocity and accretes more and more gas ( @xmath155 ) , helped by an increasing self - gravity . because of the geometry of the tail
, the accretion becomes axisymetric . as a result ,
the tail narrows at this specific location , and the pinching transforms the seed into an elongated gas cloud .
two dense filaments ( @xmath156 long , @xmath157 ) are formed on each side of the tail , both parallel to its axis ( fig .
[ fig : tail_velocity].b ) .
such elongated over - densities could be associated to caustics on both edges of tidal tail ( see e.g. * ? ? ? * on caustics propagating along tails ) . then , the filaments collapse along their major axis under the effect of self - gravity . during the process , each structure keeps accreting surrounding gas and further collapses to finally form clouds ( fig . [
fig : tail_velocity].c ) .
this fragmentation stage lasts about @xmath158 , i.e. of the order of the free - fall time of the initial filaments . at the end
, one of the clumps does not show any sign of rotation ( the western one , on the right - hand side in fig .
[ fig : tail_velocity].c ) while the other spins at about @xmath159 .
most probably , this difference originates in the slight asymmetry of the collapse of the second clump which thus accrete material with non - negligible angular momentum . because of its fast rotation , the second cloud form small gaseous spiral arms . in the mean time , the first cloud , lacking the support from rotation , further collapses and gets dense enough to form stars ( @xmath160 ) .
interestingly enough , the two clouds , about @xmath161 from each other , with comparable masses ( @xmath96 ) , formed at the same epoch and under comparable conditions ( to some extent ) , show completely different star formation histories . the second , rotating cloud starts to form its own stars @xmath162 after the first cloud . during this timelapse
, a third cloud appears along a spiral arm of the second one .
the three clumps interact with each other , keep accreting and form sub - clouds ( fig .
[ fig : tail_velocity].d ) , following the idea of @xcite and similar to the formation scenario of w3 by ( * ? ? ? * see section [ sec : w3 ] ) although the object formed here is 300 times more compact that w3 .
@xmath163 after the creation of the tidal tail at the galactic pericenter passage , the three gas clumps host star formation , with no sign of any recent external trigger , contrarily to the clouds in the central regions of the galaxies . at the end of the simulation ,
the stellar objects have merged into a single cluster while the gas clouds are still separated , as shown in fig .
[ fig : tdg_cluster ] .
the full star formation episode produces @xmath145 over @xmath164 , with a peak at @xmath165 . ) of the gas , and the position of the young stars formed in the northern tidal tail , before the clusters merge ( left ) , and after ( right ) .
the giant molecular association host about ten dense sub - clouds , while only one cluster can be detected . ]
this stellar object appears to be much more compact than observed cases of tdgs @xcite .
it has , in fact , the characteristics of a globular cluster ( despite its strong rotation ) .
we note that , at @xmath166 , i.e. when the separation between the galaxies is maximum , this cluster started to fall back toward the galactic centers .
its high orbital energy ( about twice larger than the other clusters in the simulation ) suggests that , despite the energy loss it will experience when entering regions of high density close to the galactic center , it could maintain a large apocentric distance .
the strong tidal field it would eventually experience when going close to the galactic center would significantly accelerate its mass - loss @xcite , and would slow down its rotation ( by tidally stripping stars with high angular momentum , see @xcite and references therein ) , making its properties more in line with the bulk of observed globular clusters .
however , our simulation does not allow us to follow it for a sufficient time , nor to properly account for mass - loss effects to determine the precise fate of this object .
note that only one star cluster is formed in our tidal tails , contrarily to the observations of several tens of clusters in tails of the mice or the tadpole galaxies by @xcite , and of `` beads on a string '' seen by @xcite .
@xcite argued that tails could only host the formation of many small structures along them , or a massive tdg - like object , but not both .
the presence in our simulated tails of a star cluster as massive as a tdg and the absence of any other cluster supports this conclusion .
it is difficult to foresee to what extend the scenario presented here is specific to our simulation , or could be extended to other setups , but this object demonstrates that globular clusters can be formed in mergers , even at large distances from the galactic center .
other globular clusters with comparable properties ( size , mass ) are also found in the central @xmath7 , indicating that the formation of a massive , dense stellar object in merger is not limited to the environment of tidal tails .
we analyse the properties of star clusters in a hydrodynamical simulation of the antennae galaxy merger at parsec - resolution , including star formation and stellar feedback ( photoionisation , radiative pressure and supernovae ) .
as shown in @xcite , the star formation rate becomes independent of the numerical resolution of the simulation at the scale of a few parsecs .
our main findings are as follows .
* although each galactic encounter triggers a burst of star formation , the physics of the bursts strongly varies along the merger .
in particular , the last collision enhances star formation mostly in the galactic center while earlier passages generate starburst over extended volumes . *
more stars are formed during the ( @xmath82-long ) separation of the galaxy progenitors , i.e. in between the encounters , than in each ( @xmath6-long ) starburst episode of the merger sequence . *
the different physical condition of each collision strongly impacts the formation of star clusters : most of the clusters are formed in the first galactic passages while the final coalescence does not trigger the formation of additional stellar systems at the parsec resolution . *
the galactic interaction modifies the nature of the tidal field and the turbulence of the interstellar medium which translates into a starburst activity and the formation of clusters and dense stellar systems 30 times more massive than in isolated galaxies . *
the typical star formation episode of star clusters lasts about @xmath81 .
some objects contain stars with a wider age spread , either because of the accretion of smaller clusters , or because of a secondary burst of star formation triggered by the kpc - scale dynamics of the merger .
* a few objects formed have the properties of ultra - compact dwarf galaxies ( several @xmath167 , @xmath168 ) .
they share the formation mechanism of massive clusters .
their large half - mass radius originate either from their formation via a merger of gas clouds , or a violent expansion after flying through the galactic center .
the hierarchical formation in sub - clouds being part of a giant complex , as proposed by @xcite is not a unique way to form such objects .
* one tidal tail shows the formation of a young massive cluster , too compact to be considered as a tidal dwarf galaxy .
its _ in situ _ formation is favored by the kpc - scale dynamics of the system at a specific position along the tails .
compressive tides and the absence of shearing motions allow for the formation of a dense gaseous seed , its further accretion of surrounding material and finally star formation .
this object is likely the progenitor of a globular cluster . despite the convergence of the star formation rate reached by our simulation
, several important physical aspects are still missing .
in particular , the collisional treatment of the gravitational force which rules the evolution of dense stellar systems through two - body relaxation is yet too numerically costly to be included at galactic scales .
thus , resolving the inner structure and internal dynamics of star clusters remains out - of - reach in present - day galaxy simulations .
pioneer steps have been taken to bridge this gap , e.g. by including time - dependent galactic tidal fields in @xmath85-body simulations of star clusters @xcite , but a full coupling of the two scale is yet out - of - reach .
the simulation presented here shows the way for a more realistic description of the initial conditions of the long - term evolution of star clusters .
as mentioned above , our simulation suggests that galaxy mergers could play a role in the formation of secondary populations of star in massive clusters .
this topic deserves to be further explored .
a more detailed description of stellar evolution and stellar feedback , together with a resolution of the sub - parsec structure of star clusters could provide interesting constraints on the effects of self - enrichment and the possible formation of multiple stellar populations .
we acknowledge fruitful and constructive comments from nate bastian , sara ellison , bruce elmegreen , eric emsellem , mark gieles , richard de grijs , simon karl , doug lin , thorsten naab , rory smith , curt struck and the anonymous referee .
this work was supported by genci and prace resources ( 2013,2014-gen2192 and pr86di allocations ) .
fr and fb acknowledge support from the european research council through grant erc - stg-257720 .
f. , burns c. r. , madore b. f. , mager v. a. , phillips m. m. , freedman w. l. , boldt l. , contreras c. , folatelli g. , gonzlez s. , hamuy m. , krzeminski w. , morrell n. i. , persson s. e. , roth m. r. , stritzinger m. d. , 2008 , aj , 136 , 1482 c. , bell e. f. , mcintosh d. h. , rix h .- w .
, barden m. , beckwith s. v. w. , borch a. , caldwell j. a. r. , hussler b. , heymans c. , jahnke k. , jogee s. , meisenheimer k. , peng c. y. , snchez s. f. , somerville r. s. , wisotzki l. , 2005 , apj , 630 , 771 h. , keable c. j. , dunlop j. s. , cannon r. d. , griffiths w. k. , 1988 , in grindlay j. e. , philip a. g. d. , eds , the harlow - shapley symposium on globular cluster systems in galaxies vol .
126 of iau symposium , the nuclei of nucleated dwarf elliptical galaxies - are they globular clusters ? . | we present a hydrodynamical simulation of an antennae - like galaxy merger at parsec resolution , including a multi - component model for stellar feedback and reaching numerical convergence in the global star formation rate for the first time .
we analyse the properties of the dense stellar objects formed during the different stages of the interaction .
each galactic encounter triggers a starburst activity , but the varying physical conditions change the triggering mechanism of each starburst . during the first two pericenter passages ,
the starburst is spatially extended and forms many star clusters .
however , the starburst associated to the third , final passage is more centrally concentrated : stars form almost exclusively in the galactic nucleus and no new star cluster is formed .
the maximum mass of stars clusters in this merger is more than 30 times higher than those in a simulation of an isolated milky way - like galaxy .
antennae - like mergers are therefore a formation channel of young massive clusters possibly leading to globular clusters .
monitoring the evolution of a few clusters reveals the diversity of formation scenarios including the gathering and merger of gas clumps , the monolithic formation and the hierarchical formation in sub - structures inside a single cloud .
two stellar objects formed in the simulation yield the same properties as ultra - compact dwarf galaxies .
they share the same formation scenario than the most massive clusters , but have a larger radius either since birth , or get it after a violent interaction with the galactic center .
the diversity of environments across space and time in a galaxy merger can account for the diversity of the stellar objects formed , both in terms of mass and size .
galaxies : interactions galaxies : starburst galaxies : star clusters : general galaxies : star formation ism : structure methods : numerical |
the constant terms of powers of laurent polynomials appear in different areas of mathematics and theoretical physics @xcite , @xcite .
they show up prominently in the expansion of the _ fundamental period _
@xmath0 of a laurent polynomial in @xmath1 , and @xmath2 is the @xmath3-dimensional torus cycle @xmath4 . in the theory of mirror symmetry for calabi - yau hypersurfaces in a toric variety ,
this fundamental period is of great importance , @xcite , @xcite . to compute the taylor expansion of @xmath5 in the neighborhood of @xmath6 we write @xmath7 and integrate using cauchy s formula .
the @xmath8-th coefficient in the power series expansion of @xmath5 is therefore the constant term of the @xmath8-th power of the laurent polynomial @xmath9 .
for example the constant term of the 151st power of @xmath10 is + @xmath11 a number with 200 digits . in this paper
we consider the following general problem : + _ how to compute for a multivariate ( laurent ) polynomial @xmath9 the coefficient of the monomial @xmath12 of the @xmath13-th power , @xmath14 of @xmath9 . _
+ at first glance this problem is trivial .
why cantt we just split the problem in smaller ones and compute @xmath15 and @xmath16 s.t . @xmath17 and get the coefficient we are interested in by cauchys product formula @xmath18 a general polynomial in @xmath8 variables in which the degree of each individual variable is not greater than @xmath19 has @xmath20 monomials . in our example
@xmath21 after multiplication with the common denominator @xmath22 .
in order to use ( [ cauchy - naive ] ) we could compute @xmath23 and @xmath24 which will require to store about @xmath25 coefficients of intermediate polynomials @xmath26 and @xmath27 . while this number is not really big , we must keep in mind that , in general , the absolute value of the coefficients grow exponentially as @xmath13 increases .
therefore , the amount of memory needed to store one particular coefficient will , typically , linearly grow with @xmath13 .
another possibility is to use a multivariate discrete fourier transform : evaluate the polynomial @xmath9 on sufficently many points of the @xmath8-dimensional grid of roots of unity of sufficient high degree , compute the @xmath13-th power of the values and apply the inverse discrete fourier transform to get the single coefficient we are interested in . here
, again , the number of interpolation nodes and therefore of the intermediate values which have to be stored is of the same order as the number of monomials in @xmath28 i.e. @xmath29 . in this paper
, we present an efficient and , we believe , simple algorithm which adresses this problem .
our approach is similar to the dft method in that it based on evaluation and iterpolation .
however , the amount of memory needed is much lower and the running speed can be improved by a trivial parallelization of the evaluation .
we also describe the implementation of the parallel version of the algorithm for the cuda - enabled graphics hardware .
finally , an application to the problem of finding the picard - fuchs differential equation of the family of toric calabi - yau varieties is given .
this problem was the main motivation for this work .
+ i like to thank d. van straten for suggesting the problem and e. schmer for useful discussions .
this work was realised with support from the dfg priority program _
algorithmic and experimental methods in algebra , geometry and number theory _ , as part of the project _
monodromy algorithms in singular_.
let @xmath9 be a multivariate polynomial in variables @xmath30 with rational coefficients . using multiindex notation we write @xmath31 @xmath32 and @xmath33 a finite index set .
let @xmath34 denote the degree of @xmath9 in the variable @xmath35 .
for a given @xmath36-dimensional multiindex @xmath37 we denote by @xmath38_j$ ] the coefficient of @xmath9 in front of @xmath39 i.e. @xmath40_j:=c_j.\ ] ] furthermore let @xmath41 denote the @xmath42-dimensional multindex obtained by truncating @xmath37 after the @xmath42-th component and for an @xmath43-dimensional multiindex @xmath44 let @xmath45 denote the concatenation of both i.e. for @xmath46 and @xmath47 @xmath48 given @xmath49 pairs @xmath50 such that @xmath51 for @xmath52 , there is a unique _ univariate interpolation polynomial _ of degree @xmath3 such that @xmath53 @xmath54 where @xmath55 is the lagrange basis polynomial for @xmath56 .
@xmath57 has rational coeffiients and @xmath58 .
let @xmath59 denote the polynomial @xmath9 with @xmath60 substituted for the last variable i.e. @xmath61 we have the following observation about the coefficients of the interpolation polynomial and those of @xmath59 .
[ lem1 ] let @xmath62 $ ] , @xmath63 , @xmath64 as above , and @xmath65 .
then , we have @xmath66_{i}=\left[\mathcal{i}^{n}(u_{0},\dots , u_{n};[f_{u_0}]_{i'},\dots,[f_{u_n}]_{i'})\right]_{i_{n}}\label{eq:1}.\ ] ] let @xmath67 and @xmath68_{i'},\dots,[f_{u_n}]_{i'})(y)=\sum_{k=0}^{n}a_{k}y^{k}.\label{eq : interpol}\ ] ] then , the right hand side of ( [ eq:1 ] ) equals @xmath69 and , therefore , we have to show that @xmath70_{i}=a_{i_{n}}$ ] .
the coefficients @xmath71 of the interpolation polynomial are found by multiplying the vector @xmath72_{i'},[f_{u_1}]_{i'},\dots,[f_{u_n}]_{i'})^{t}$ ] by the inverse of the vandermonde matrix for @xmath73 .
this inverse exists since @xmath73 are pairwise distinct .
therefore , we have @xmath74_{i'}\\ \left[f_{u_1}\right]_{i'}\\ \vdots\\ \left[f_{u_n}\right]_{i'}\end{array}\right)\label{zurueck}\ ] ] by ( [ polynom ] ) we have @xmath75_{i ' } & = & \left[\sum_{\nu_{1},\nu_{2},\dots,\nu_{n-1},\tau}c_{\nu_{1}\nu_{2}\dots\tau}u_{j}^{\tau}x_{1}^{\nu_{1}}x_{2}^{\nu_{2}}\dots x_{n-1}^{\nu_{n-1}}\right]_{i'}\\ & = & \sum_{\tau}c_{i_{1}i_{2}\dots i_{n-1}\tau}u_{j}^{\tau}\\ & = & \sum_{\tau}c_{i'\tau}u_{i}^{\tau},\end{aligned}\ ] ] where in the last expression @xmath76 is an abbriviration for the multiindex @xmath77 ) . using this ,
rewrite ( [ zurueck ] ) as @xmath78 at the same time @xmath79 from ( [ hin ] ) and ( [ zurueck - again ] ) we get @xmath80 and with ( [ polynom ] ) @xmath81_{i}\ ] ] follows . since @xmath82 we have @xmath83_{i}=\left[\mathcal{i}^{n}(u_{0},\dots , u_{n};[f_{u_0}^p]_{i'},\dots,[f^p_{u_n}]_{i'})\right]_{i_{n}},\ ] ] where @xmath84 .
let @xmath85 $ ] be a laurent polynomial .
we want to calculate the first coefficients of its constant terms series @xmath86_0)_{p\in{\mathbb{n}}}$ ] .
let @xmath87 be the common denominator of the laurent monomials of @xmath27 .
then @xmath88 $ ] and @xmath89_0=[s^pf^p]_{p\cdot s},\ ] ] where @xmath90 .
we want to apply ( [ cor_1 ] ) to @xmath91 and @xmath92 .
the simple idea now is to compute @xmath93_{i'},\ , j=0,\dots , n$ ] which appear in ( [ cor_1 ] ) recursively i.e. @xmath94_{i'}=\left[\mathcal{i}^{n}(u_{0},\dots , u_{n};[f_{u_j u_0}^p]_{i''},\dots,[f^p_{u_j u_n}]_{i''})\right]_{i_{n-1}}\ ] ] and so on for @xmath95 . with each level of recursion
the number of variables decreases by one , therefore the depth of the recursion equals @xmath8 . for simplicity
assume @xmath96 the computation of @xmath97_i$ ] proceeds as follows : 1 . for @xmath98 fix @xmath99 as in lemma and
compute @xmath100 , where @xmath101 is the vandermonde matrix of @xmath102 .
2 . invoke the recursive procedure @xmath103 with initial parameters @xmath104 and @xmath105coefficient matrix of @xmath9 .
the pseudo code of the procedure @xmath106 is given in the algorithm 1 .
@xmath107 , @xmath108 , @xmath109 , @xmath110 @xmath97_i$ ] @xmath111do it using horner s rule @xmath112
@xmath114 compute the @xmath115-th component of @xmath116 @xmath117 to improve the performane we apply some well known computational tricks . the evaluation in the line 2
is done using the horner s rule which for a polynomial @xmath118 at @xmath119 says @xmath120 in this way the expensive computation of powers of @xmath121 is avoided .
the vandermonde matrix @xmath101 can be inverted using classical gauss elimination algorithm in @xmath122 time . since on each level of recursion
we need only one row of @xmath123 ( line 9 in the code above ) precomputing the whole inverse seems too expensive .
fortunately , if the nodes @xmath124 are choosen to be equidistant there is a recursive algorithm which computes one single row of @xmath123 in @xmath125 @xcite .
the computation of the @xmath13-th power in the line 4 can be done using binary powering algorithm as described in @xcite .
we also note that @xmath126 s in line 6 do not depend on each other and thus the order in which they are computed does not matter .
in particular they can be computed parallely .
finally , the number of nodes @xmath3 need not to be the same on each recursion level but depend on the degree of @xmath9 in a particular variable i.e. we have to choose @xmath127 on the @xmath128-th recursion level .
for example if @xmath9 has degree @xmath129 in one variable and degree @xmath130 in the remaining variables we could improve speed by considering @xmath131 interpolation nodes only on one level of recursion and @xmath13 interpolation nodes on the ramaining .
if @xmath9 happens to have special form @xmath132 with @xmath133 $ ] a trick can be applied to slightly speed up the computation .
if ( [ decomp ] ) holds , then so is for @xmath59 , @xmath134 with @xmath135 $ ] .
consequently , this decomposition holds on all levels of recursion .
we note that in this case @xmath136 and @xmath137_p & = & \left[\sum_{i=0}^{p}{p \choose
i , i}a^{i}c^{i}b^{p-2i}\right]_{p}\label{split}\\ & = & \sum_{i=0}^{p}{p \choose i , i}\left[a^{i}c^{i}b^{p-2i}\right]_{p}\nonumber.\end{aligned}\ ] ] since we need only the terms with @xmath138 i.e. @xmath139 .
we apply this
shortcut on the last but one level of recursion to compute @xmath140_p$ ] when we have to deal with polynomials in two variables .
we use the modified version of coeff and a new procedure split2 for this .
thus the depth of recursion is reduced by one . for the case @xmath141 the call tree of the algorithm is depicted in figure [ tree ] .
@xmath107 , @xmath108 , @xmath109 , @xmath110 @xmath97_i$ ] @xmath111 @xmath142 @xmath113 @xmath143 @xmath117 @xmath144 , @xmath108 , @xmath109 , @xmath110 @xmath97_i$ ] @xmath145 @xmath146 @xmath147 @xmath148 @xmath149 @xmath150 @xmath151 @xmath152 @xmath153 @xmath154 @xmath155 @xmath156 @xmath157 @xmath155 @xmath158 @xmath159 @xmath160 @xmath161 the bottom procedure split2 is called exactely @xmath162 times . inside the split2 procedure
we have two nested loops of depth 2 .
the bodies of both inner loops ( lines 12 - 14 and 25 - 28 in the listing [ alg - split2 ] ) are executed both at most @xmath163 times .
therefore the time complexity of the algorithm is of order @xmath164 .
thanks to split2 we gain a factor @xmath165 compared to the version which does not use split2 .
if we take into consideration the linear growth of the length of the numbers involved into computation we arrive at assymtotical running time @xmath166 or @xmath167 , since @xmath3 depends linearly on @xmath13 and @xmath19 . as we can see from the description of the algorithm , the memory consumption is @xmath168 , which is the space needed to store the @xmath169 rows of the inverse vandermonde matrix . in particular , for fixed @xmath19 it is independent from the number of monomials @xmath9 has . [ tree ] .,width=377 ] when @xmath19 is fixed , the running time depends on @xmath13 and the number of monomials in @xmath9 .
as was mentioned before , the algorithm can be parallelised in a very straight - forward way . as a target platform
we choose a graphics card built around nvidia geforce gtx 480 gpu .
this gpu is capable of running up to 480 parallel execution threads .
the gpu is programmed using an extension of c / c++ programming language called _ nvida cuda c _ @xcite .
a typical cuda program is executed on both the cpu and the gpu .
the parts of the code that run on the graphics hardware are called _ kernels _ and each kernel can be run @xmath3 times in parallel by @xmath3 different cuda threads .
the threads are organized in one- , two- or three - dimensional _ thread blocks _ while blocks are organized in one- or two - dimensional _ grid_. the graphics hardware was origianlly created for the purpose of low - precision floating point operations it uveils its full computing power only when dealing with 32-bit floating point numbers .
therefore we can perform exact computations only on numbers within the range @xmath170 or equivalently @xmath171 .
that is because only 23 bits of 32 are used to store the significant digits ( 8-bits being reserved for the exponent and one remaining bit for the sign ) .
this range is far too small to be useful due to the rapid growth of the numbers @xmath97_0 $ ] .
therefore we used a _ residue number system _ ( rns ) to represent large integers . given a ( fixed ) set of pairwise coprime natural numbers @xmath172 , the integer @xmath173 is represented by the system of residues @xmath174 @xmath175 is called the _ dynamic range _ of the system . by the well - known _
chinese reminder theorem _ the number @xmath176 can be reconstructed from @xmath177 . in practice ,
the conversion from an rns representation of @xmath176 to a decimal representation is done with the _ mixed radix conversion ( mrc ) _
i.e. @xmath176 is represented in the form @xmath178 where @xmath179 s are called _ mixed radix digits_. once @xmath179 s are known the decimal value can be computed easily .
one classical algorithm for finding @xmath179 s from rns representation of @xmath176 is given in @xcite .
in our implementation the program branches in several execution threads in the procedure coeff on the top level of recursion .
the threads are organized in @xmath180 blocks with @xmath181 threads per block .
all the threads within the @xmath36-th block performs computation modulo prime number @xmath182 .
the @xmath44-th thread of @xmath36-th block computes @xmath183 for @xmath184 .
the arrangement of threads is illustrated in figure 2 .
[ threads ]
in the theory of toric calabi - yau varieties the constant terms of powers of laurent polynomials occur in the following way .
consider a reflexive lattice polytope @xmath185 .
every vertex @xmath186 of @xmath187 corresponds to a monomial @xmath188 .
then the laurent polynomial corresponding to @xmath187 is @xmath189 this polynomial defines a family of hypersurfaces @xmath190 .
the function @xmath191_0 z^i\ ] ] is the period of the holomorphic 3-form @xmath192 on @xmath193 and is a solution of the picard - fuchs equation @xmath194 where @xmath195 @xmath196 and @xmath197 are polynomials with integral coefficients @xmath198 the numbers @xmath199 @xmath200_0 , & i\geq 0\\ 0 , & \text{otherwise } \end{cases}\ ] ] then satisfy the following recurrence relation @xmath201 for all @xmath202 .
we say that the recurrence has length @xmath203 and degree @xmath204 .
conversely , if a sequence satisfies ( 15 ) then its generating function is anihilated by the differential operator ( 13 ) . given @xmath187
we want to find the corresponding differential operator @xmath205 . to do this ,
we compute first @xmath3 coefficients of the constant terms series ( 12 ) of @xmath206 and try to find a reccurence of the form ( 15 ) .
assume there is one with lenght @xmath203 and degree @xmath19 .
then it is determined by @xmath207 coefficients @xmath208 of @xmath209 .
we check if there is a solution to system of @xmath49 linear equations for the @xmath207 unknowns @xmath210 @xmath211 we have to keep this system overdetermined i.e. to have @xmath212 .
if there is a solution which does not change when we add more equations , we hope that we found the correct operator . since , a priori , we do not know the length and the degree of the reccurence we make the ansatz ( 16 ) for all @xmath213 which are allowed by the requirement that ( 16 ) be overdetermined . thus it is essential to be able to compute @xmath214_0 $ ] for as many @xmath44 s as possible in reasonable time .
this method was used by batyrev and van straten @xcite and has been polular ever since .
more recently , batyrev and kreuzer used this approach in @xcite to determine the picard - fuchs operators for some new families of calabi - yau threefolds with picard number 1 .
they succeeded in 28 cases out of 68 . using our algorithm were able to compute 4 more calabi - yau operators . although we knew up to 300 constant terms
, in many cases the recurrence did not show up .
the operators and their corresponding polynomials are : + * # 24 * : 99 , _ statistical theory of the energy levels of complex systems _ , journal of mathematical physics , vol . 3 , 1962 . , _
constructing new calabi - yau 3-folds and their mirrors via conifold transitions _
arxiv:0802.3376v2 , 3rd ed .
, addison - wesley , 1998 .
, nasa technical note , 1969 , indagationes math . * 9 * , ( 1998 ) , 221 - 231 .
, _ residue arithmetic and its application to computer technology , 1967 _ | we present an algorithm for efficient computation of the constant term of a power of a multivariate laurent polynomial .
the algorithm is based on univariate interpolation , does not require the storage of intermediate data and can be easily parallelized . as an application we compute the power series expansion of the principal period of some toric calabi - yau varieties and
find previously unknown differential operators of calabi - yau type . |
we consider the solution @xmath4 to the * graphical mean curvature flow ( mcf ) * , @xmath5 where @xmath6 $ ] , and compare it to the solution @xmath7 to the heat equation with the same initial graph , @xmath8 when @xmath2 , the mean curvature flow is called curve shortening flow and equation reduces to @xmath9 for this dimension , nara and taniguchi @xcite showed that if the initial graph @xmath10 is in @xmath3 , the solution to the curve shortening flow converges to the solution to the heat equation with the same initial data .
the proof relies heavily on the existence of a gauss kernel and the fact that the right - hand side is a potential . in higher dimensions ( @xmath11 ) ,
a couple of results suggest a similar behavior .
if the initial graph @xmath10 is compactly supported , an argument by huisken using large balls as barriers and the maximum principle shows that @xmath12 goes to zero uniformly as @xmath13 , which is also the asymptotic behavior of @xmath14 ( see @xcite for the mean curvature flow and @xcite for the heat equation ) . for bounded and radially symmetric entire initial graphs , nara @xcite proved the convergence of solutions to the mean curvature flow to solutions to the heat equation again .
the object of this note is to show that the result of nara taniguchi does not extend to higher dimensions in general .
[ thm : main ] for @xmath1 , for every constant @xmath15 , there is a function @xmath16 , @xmath17 $ ] , for which the solution @xmath12 to the mean curvature flow with initial data @xmath10 stabilizes to a constant @xmath18 , while the solution @xmath14 to the heat equation with initial data @xmath10 stabilizes to the constant 1 .
more precisely , we have @xmath19 and @xmath20 [ thm : mainosc ] for @xmath1 , for every @xmath15 , there is a function @xmath21 , @xmath17 $ ] , for which the solution @xmath22 to the mean curvature flow with initial data @xmath23 oscillates @xmath24 and the solution @xmath25 to the heat flow with initial data @xmath23 tends to @xmath26 uniformly on @xmath27 as @xmath13 .
these theorems are reminiscent of the different behaviors of the mean curvature flow of embedded initial data when @xmath1 or when @xmath2 .
indeed , grayson proved that embedded curves remain embedded under mean curvature flow @xcite , but embedded surfaces can pinch off @xcite .
the long time existence of smooth solutions @xmath28 for entire lipschitz graphs @xmath10 was established by ecker
huisken @xcite .
looking at interior estimates for the mean curvature flow from the same authors @xcite and standard theory for the heat equation , one can see that both @xmath28 and @xmath29 move more slowly as time goes to infinity .
therefore , the flows can not separate much as time gets large and what happens right at the start is critical .
both our functions @xmath10 and @xmath23 illustrate how the mean curvature flow can pull away from the heat flow instantly .
the idea for @xmath10 is to build tall thin spikes that enclose a large volume .
the solution to the heat equation converges to the average of the volume enclosed between the graph of @xmath10 and the base @xmath30-plane , while the evolution by mean curvature loses a lot of volume at the start and will tend to a smaller constant .
we use a similar idea for @xmath23 . starting with an oscillating graph
, we add spikes in the lower regions , thereby adding volume .
this forces the heat flow to stabilize but does not affect the mean curvature flow much .
we use shrinking doughnuts and spheres as barriers
. * acknowledgment . *
the authors would like to thank bruce kleiner for suggesting the first example during the workshop
geometric flows and riemannian geometry " at the american institute of mathematics ( aim ) in september 2015 .
we would like to thank the aim for the hospitality and the organizers of the workshop , lei ni and burkhard wilking , for the invitation .
in @xcite , angenent constructed an embedded rotationally symmetric torus in @xmath31 that shrinks self - similarly under mean curvature flow to a point in finite time .
we denote both this torus and its parametrization by @xmath32 .
the standard scaling is such that @xmath33 is a solution to the self - shrinker equation @xmath34 where @xmath35 is the mean curvature and @xmath36 is the projection onto the normal space . for what follows ,
we require only the existence of such a shrinking torus ; the uniqueness of @xmath32 is still an open problem . for applications in the following sections
, we will work with a rescaling of @xmath32 , which we denote by @xmath37 .
the scale is chosen so that @xmath37 is in the cylinder of radius @xmath38 around its axis of rotation and its height is less than @xmath39 .
notice that any rescaling of @xmath32 also shrinks self - similarly under mean curvature flow to the origin in finite time , although it no longer satisfies the self - shrinker equation . to determine the scaling for @xmath37
, we consider its profile curve in the half - plane @xmath40 , which we denote by @xmath41 .
it follows from the construction in @xcite that @xmath41 is a simple closed convex curve ; it intersects the @xmath42-axis perpendicularly at two points ; and it is symmetric with respect to reflections about the @xmath42-axis .
let @xmath43 and @xmath44 be the two points where @xmath41 intersects the @xmath42-axis , and choose @xmath45 .
also , let @xmath46 denote the maximum @xmath47-coordinate of @xmath41 .
we say that @xmath48 is the * inner radius * , @xmath49 is the * outer radius * , and @xmath46 is the * maximum height * of the torus @xmath37 ( see figure [ fig : doughnut ] ) . by our choice of scaling , we have @xmath50 and @xmath51 . picturing @xmath37 in @xmath31 , with coordinates @xmath52 , so that it has rotational symmetry about the @xmath47-axis and reflection symmetry about the base @xmath30-plane @xmath53 , we can interpret @xmath48 , @xmath49 , and @xmath46 in the following way .
the projection of @xmath37 into the base @xmath30-plane is the annulus @xmath54 , and @xmath37 is sandwiched between the planes @xmath55 and @xmath56 .
.,height=192 ] ( 0,0 ) ( -82,55)@xmath48 ( 0,38)@xmath57 ( 0 , 140)@xmath46 ( 68,155)@xmath41
in this section , we construct the initial graph @xmath58 , which has tall thin spikes on the integer lattice , then study its evolution under mean curvature flow and heat flow .
consider the torus @xmath37 described in the previous section and translate it so that it is centered at the point @xmath59 .
this translated torus shrinks self - similarly under mean curvature flow to its center in finite time . using the property that @xmath50 , we see that this torus is contained in the box @xmath60^n \times [ 0,2\delta_0]$ ] .
we also note that the projection of this torus onto the base @xmath30-plane is the annulus @xmath61 .
for each point @xmath62 , we place a copy of @xmath37 centered at the point @xmath63 and denote it by @xmath64 . because @xmath50 , these tori are disjoint .
we introduce two subsets of @xmath65 , which we identify with the base @xmath30-plane in @xmath31 .
first , we let @xmath66 be the region in the base @xmath30-plane obtained by removing the union of discs @xmath67 : @xmath68 second , we let @xmath69 be the union of the discs @xmath70 : @xmath71 now , we are ready to define an entire function @xmath10 supported in the region @xmath69 .
we first define @xmath10 on the disc @xmath72 to be a smooth non - negative function such that @xmath73 next , we extend @xmath10 to be zero on the remainder of the square @xmath60^n$ ] : @xmath74^n \ , \backslash b_{\ell_0}(0,0).\ ] ] finally , we define @xmath10 on all of @xmath65 by making it periodic : @xmath75 the graph of @xmath10 may be viewed as a collection of tall spikes .
each spike is concentrated around some point @xmath76 , and assuming @xmath77 , the spike passes through the hole of @xmath64 . in the next section ,
we describe the information these carefully placed tori give us on @xmath12 .
we begin this section by noting that for each @xmath78 , the torus @xmath64 sits on or above the graph of @xmath10 .
next , we run mean curvature flow . on one hand , because @xmath10 is a smooth entire function , the mean curvature flow with initial data @xmath79 has a smooth solution @xmath80 that exists for all @xmath81 ( see @xcite,@xcite ) , and an application of the strong maximum principle shows @xmath82 for @xmath83 . on the other hand , the torus @xmath64 shrinks to its center under the mean curvature flow in finite time @xmath84 .
it follows from the comparison principle for mean curvature flow that for @xmath83 , the evolution of the torus @xmath64 is disjoint from the evolution of the entire graph @xmath85 . in particular , by the time @xmath86 , each tall spike of the original entire graph @xmath85 must pass through the hole of the surrounding torus .
we will show that by the time @xmath87 , the maximum height of @xmath12 is less than or equal to @xmath46 .
the torus @xmath37 evolves under mean curvature flow as a family of self - similar tori @xmath88 that is shrinking to its center as @xmath89 approaches @xmath90 .
let @xmath91 , @xmath92 , and @xmath93 denote the inner radius , outer radius , and maximum height of the torus @xmath94 , respectively , and set @xmath95 @xmath96 [ prop:2 ] for @xmath97 , we have @xmath98 on @xmath99 .
this is true when @xmath100 by definition of @xmath10 , and by comparison with the tori barriers , we have @xmath98 on @xmath101 , for @xmath102 .
an application of the maximum principle shows that @xmath98 on @xmath99 when @xmath97 . because the domain we are working in is unbounded , we provide the details for the reader s convenience . using the continuity of @xmath28 and the periodicity of @xmath103
, we see that the supremum of @xmath103 over the closed set @xmath104 is achieved and depends continuously on @xmath89 .
if the proposition does not hold , then there will be a first time @xmath105 where the supremum of @xmath106 over @xmath107 equals @xmath46 . given the height estimate on the boundary @xmath108 , this supremum will be achieved at an interior point @xmath109 .
consequently , we have @xmath110 , @xmath111 in @xmath112 , and @xmath113 in @xmath99 for @xmath114 , which contradicts the maximum principle ( * ? ? ?
* chapter 2 ) .
[ prop:3 ] for @xmath115 , we have @xmath116 on @xmath65 . using the previous proposition , we know that @xmath98 on @xmath99 for @xmath97 .
in addition , as @xmath117 , we know that @xmath99 converges to @xmath65 . then , from the continuity of the solution , we deduce that @xmath118 on @xmath65 . by the maximum principle
, we conclude that @xmath119 on @xmath65 , for @xmath115 .
we are now ready to show that the solution to the mean curvature flow converges to a constant .
the solution @xmath12 to the mean curvature flow with initial condition @xmath120 tends to a constant , denoted by @xmath121 , uniformly in @xmath27 as @xmath122 .
it follows from the maximum principle that if such a constant @xmath121 exists , it can not be bigger than @xmath123 .
the proof follows from results of ecker
huisken @xcite and the fact that @xmath124 is periodic in the spacial variables .
here we use the notation @xmath125 for the euclidean gradient and @xmath126 for the tangential gradient on the graph of @xmath28 . in this notation
we have the estimate @xmath127 where @xmath128 is the second fundamental form and @xmath129 is the norm on the graph of @xmath28 ( * ? ? ?
* proof of corollary 4.2 ) .
now , using the periodicity of @xmath10 , we know that @xmath130 is bounded , and consequently @xmath131 is bounded for all time by ( * ? ? ?
* corollary 3.2 ) . then applying ( * ? ? ?
* proposition 4.4 ) we have that @xmath132 as @xmath122 . because @xmath133 is bounded for all time , the right - hand side of goes to @xmath134 , so that @xmath135 as @xmath136 for some constant @xmath137 .
the periodicity implies that @xmath138 and gives the uniform convergence of @xmath28 to a constant as @xmath136 .
the volume under the graph of @xmath10 restricted to the square @xmath60^n$ ] is @xmath139^n } u_0 \ , dvol = 1 $ ] .
a simple estimate shows that @xmath140^n ) \subset b_{r+\sqrt{n}/2}(0).\ ] ] using these inclusions to estimate the integral of @xmath10 over unit cubes centered at points in the integer lattice , we have @xmath141 and it follows that @xmath142 the statement of theorem [ thm : main ] follows from the result of repnikov eidelman @xcite for the heat equation .
in this section , we expand upon the previous example to construct an example of an initial entire graph over @xmath65 , @xmath1 , that oscillates under the mean curvature flow and stabilizes to a constant under the heat flow . in the one - dimensional case , nara
taniguchi ( * ? ? ?
* proposition 3.4 ) proved that there is a function @xmath143 such that the solution @xmath144 of ( [ eq : mcf - nonparam ] ) with @xmath145 does not stabilize .
the heuristic idea for the proof of theorem [ thm : mainosc ] is to first take the one - dimensional example @xmath144 and extend it to higher dimensions , simply by noticing that @xmath146 is also a solution to the graphical mean curvature flow .
then by placing spikes with sufficient volume in the regions where the initial function @xmath147 is zero , we can construct an initial function @xmath148 that will stabilize to the constant 1 under the heat flow .
however , since those spikes shrink very quickly under the mean curvature flow , the evolution of @xmath23 under the mean curvature flow mimics the oscillatory behavior of @xmath144 .
[ proposition 3.4 @xcite ] [ prop : nt ] fix constants @xmath149 , @xmath150 , and @xmath151 with @xmath152 and @xmath153 . set @xmath154 $ ] and @xmath155 $ ] , for @xmath156 .
let @xmath157 be a function that satisfies for later purpose , let us consider the two families of slabs for @xmath166 @xmath167\ } , \\
s_{2k+1 } & = \ { x \in \mathbb r^n : |x_1| \in [ ( 2k+1 ) !
+ \tfrac{1}{4 } , ( 2k+2 ) ! -\tfrac{1}{4}]\}. \end{aligned}\ ] ] note that @xmath168 and @xmath169 if @xmath170 .
let @xmath180 be the example from theorem [ thm : main ] . recall that @xmath139^n } u_0 = 1 $ ] .
we define the function @xmath181 by @xmath182 where @xmath183 is the characteristic function of the set @xmath128 ( see picture ) .
( 162.5,68)(0,0 ) ( -0,30)(1.48,30)(1.56,30 ) ( 1.56,30)(1.64,30)(5,30 ) ( 5,30)(8.59,29.69)(9.38,31.25 ) ( 9.38,31.25)(10.16,32.81)(10,40 ) ( 10,40)(9.84,47.19)(10.62,48.75 ) ( 10.62,48.75)(11.41,50.31)(15,50 ) ( 15,50)(19.38,50)(27.5,50 ) ( 27.5,50)(35.62,50)(40,50 ) ( 40,50)(43.59,50.31)(44.38,48.75 ) ( 44.38,48.75)(45.16,47.19)(45,40 ) ( 45,40)(44.84,32.81)(45.62,31.25 ) ( 45.62,31.25)(46.41,29.69)(50,30 ) ( 50,30)(57.19,30)(85,30 ) ( 85,30)(112.81,30)(120,30 ) ( 120,30)(123.59,29.69)(124.38,31.25 ) ( 124.38,31.25)(125.16,32.81)(125,40 ) ( 125,40)(124.84,47.19)(125.62,48.75 ) ( 125.62,48.75)(126.41,50.31)(130,50 ) ( 130,50)(134.61,50)(143.44,50 ) ( 143.44,50)(159.27,50)(162,50 ) ( 10,12.5)(0,2)23(0,1)1 ( 45,12.5)(0,2)23(0,1)1 ( 125,12.5)(0,2)23(0,1)1 ( 6.25,16.25)(0,3.93)11(0,1)1.96 ( 13.75,25)(0,3.82)9(0,1)1.91 ( 41.25,25)(0,3.82)9(0,1)1.91 ( 48.75,16.25)(0,3.93)11(0,1)1.96 ( 121.25,16.25)(0,3.93)11(0,1)1.96 ( 128.75,25)(0,3.82)9(0,1)1.91 ( 48.75,25)(1,0)72.5 ( 121.25,25)(1,0)0.12 ( 48.75,25)(-1,0)0.12 ( 6.25,20)(1,0)42.5 ( 48.75,20)(1,0)0.12 ( 6.25,20)(-1,0)0.12 ( 0,30)(1,0)162.5 ( 162.5,30)(1,0)0.12 ( 121.25,55)(1,0)3.75 ( 125,55)(1,0)0.12 ( 121.25,55)(-1,0)0.12 ( 86.25,21.25)(0,0)[cc]@xmath184 ( 125,55)(1,0)3.75 ( 128.75,55)(1,0)0.12 ( 125,55)(-1,0)0.12 ( 45,55)(1,0)3.75 ( 48.75,55)(1,0)0.12 ( 45,55)(-1,0)0.12 ( 41.25,55)(1,0)3.75 ( 45,55)(1,0)0.12 ( 41.25,55)(-1,0)0.12 ( 10,55)(1,0)3.75 ( 13.75,55)(1,0)0.12 ( 10,55)(-1,0)0.12 ( 6.25,55)(1,0)3.75 ( 10,55)(1,0)0.12 ( 6.25,55)(-1,0)0.12 ( 127.5,60)(0,0)[cc]@xmath186 we note that @xmath193 with uniform convergence for @xmath194 by construction .
note that the existence of the limit at @xmath195 is a necessary and sufficient condition for the convergence of @xmath14 on compact subsets of @xmath27 as @xmath13 @xcite .
the uniform convergence of the limit is equivalent to the uniform convergence of @xmath14 to 1 as @xmath136 @xcite .
let @xmath196 be the solution of ( [ eq : mcf - nonparam ] ) with @xmath197 .
also , let @xmath198 be the shrinking sphere with extinction time @xmath90 .
the radius of such a sphere is denoted by @xmath199 and we know from @xcite that @xmath200 .
we use spheres in @xmath31 centered at points @xmath201 with @xmath202 as barriers to obtain @xmath203 because the solution @xmath196 is periodic in @xmath204 , an argument similar to the one from propositions [ prop:2 ] and [ prop:3 ] for the punctured slabs @xmath205 gives us that @xmath206 by placing spheres above @xmath23 in @xmath185 , i.e. spheres centered at @xmath207 , for @xmath208 , we obtain @xmath209 it follows that at time @xmath90 , we can construct an upper barrier for @xmath210 by taking @xmath211 , where @xmath212 and satisfies @xmath213\\
\phi^+_0(x_1 ) = { \varepsilon } , \quad |x_1| \in [ ( 2k+1 ) ! + \tfrac{1}{2 } , ( 2k+2 ) ! -\tfrac{1}{2 } ] \end{gathered}\ ] ] by proposition [ prop : nt ] , the solution @xmath214 to the curve shortening flow with initial data @xmath215 oscillates and @xmath216 . the function @xmath217 is a solution to the graphical mean curvature flow and an upper barrier for @xmath196 for @xmath218 .
we therefore have @xmath219 the function @xmath220 , which is the solution to with initial data @xmath221 , serves as a lower barrier and we obtain @xmath222 this completes the proof of theorem [ thm : mainosc ] .
, _ shrinking doughnuts _ , in nonlinear diffusion equations and their equilibrium states , 3 ( gregynog , 1989 ) , vol .
7 of progr .
nonlinear differential equations appl .
, birkhuser boston , boston , ma , 1992 , pp . | we present two initial graphs over the entire @xmath0 , @xmath1 for which the mean curvature flow behaves differently from the heat flow . in the first example , the two flows stabilize at different heights . with our second example , the mean curvature flow oscillates indefinitely while the heat flow stabilizes .
these results highlight the difference between dimensions @xmath1 and dimension @xmath2 , where nara
taniguchi proved that entire graphs in @xmath3 evolving under curve shortening flow converge to solutions to the heat equation with the same initial data . |
in 1995 delphi collaboration reported data of the 3rd order bose - einstein correlations ( bec ) and concluded that there is a genuine the 3rd order bec in @xmath5 channel and the effect of the 2nd order bec in @xmath0 channel @xcite . the method used in their analyses
is the formulation in terms of the plane wave amplitude . from theoretical point of view , at almost the same time
, formulations for the 2nd order bec by means of the coulomb wave function had been investigated in refs .
moreover , formulations for the 3rd and 4th order bec by means of the coulomb wave function have been recently proposed in refs .
in particular , the formulation of ref . @xcite contains the degree of coherence ( @xmath6 ) for the magnitude of the be exchange term .
the advantage of this approach is that it can be directly applicable to analyses of the data on @xmath5 channel with the cern - minuit program . ) a priori . ] in the present study we investigate whether this approach using formula derived by means of the coulomb wave function can be extended to an unlike charged combination channel , @xmath0 .
we should stress that : * through our approach , the plane wave amplitude is used as a basic calculation . at the second step the coulomb wave function is utilized . * the interferometry effect for @xmath0 ,
i.e. , the squared amplitude by the coulomb wave function , contains the following neutral current : @xmath7 if the magnitude of this contribution is zero , we can not explain the data on @xmath0 channel by delphi collaboration .
their data ask for the finite magnitude which is expressed by a new parameter @xmath8 in this paper .
this paper is organized in the following way : in 2 , a simple model for @xmath0 channel is investigated . in 3 , a model including full amplitudes ( @xmath9 ) is presented .
a new formula for @xmath0 channel is also derived here . in 4 , analyses of the data on @xmath5 channel are performed .
the estimated parameters are compared with those in @xmath0 channel . in the final section , concluding remarks are given .
authors of ref . @xcite stressed that data on @xmath0 channel are described by the formulas of the 2nd order bec .
we shall study whether this statement is correct or not using the interferometry approach formulated in terms of the coulomb wave function . to analyze data on bec effect in @xmath10 channel , first of all we consider the simplest diagram shown in fig .
[ fig1 ] .
the plane wave amplitudes ( @xmath11 ) for @xmath0 channel are given as [ eq2 ] pwa(1;+ ) & = & 12 e^ i ( k_1^(+ ) x_1 + k_2 x_2 + k_3 x_3 ) , + pwea(1;+ ) & = & 12 e^ i ( k_1^(+ ) x_1 + k_2 x_3 + k_3 x_2 ) , where @xmath12 stands for the positive pion and @xmath13 represents the plane wave exchange amplitude due to the be statistics . when the coulomb wave functions is taken into account because of charged pions ( @xmath0 ) , the above @xmath11 and @xmath13 should be replaced by @xmath14 and @xmath15 , [ eq3 ] a(1;+ ) & = & _ k_1k_2^c+-(x_1 , x_2 ) _
k_2k_3^c(x_2 , x_3 ) _
k_3k_1^c+-(x_3 , x_1 ) , + ea(1;+ ) & = & _ k_1k_2^c+-(x_1 , x_3 ) _
k_2k_3^c(x_3 , x_2 ) _
k_3k_1^c+-(x_2 , x_1 ) , where @xmath16 is defined as @xmath17 \qquad \label{eq4}\end{aligned}\ ] ] with @xmath18 , @xmath19 , @xmath20 , @xmath21 and @xmath22 ( ( @xmath23 ) and ( @xmath24 ) are the like - charge combination and unlike - charge one , respectively ) .
@xmath25 $ ] and @xmath26 are the confluent hypergeometric function and the gamma function , respectively .
the parts of the plane wave of eq .
( [ eq3 ] ) are given as @xcite [ eq5 ] a(1;+ ) & & e^(3/2)i [ k_1^(+)x_1 + k_2x_2 + k_3x_3 ] , + ea(1;+ ) & & e^(3/2)i [ k_1^(+)x_1 + k_3x_2 + k_2x_3 ] , where the factor ( @xmath27 ) is attributed to property of the coulomb wave function and 3-body problem .
the interferometry effect for the ( @xmath0 ) channel is calculated as @xmath28\ , \label{eq6}\end{aligned}\ ] ] where @xmath29 stand for the source functions of particle @xmath30 .
we use the gaussian distribution of the radius @xmath31 , @xmath32 $ ] , and [ eq7 ] & & q_3 = , + & & f_1^(+ ) = 12 [ |a(1;+)|^2 + |ea(1;+)|^2 ] , + & & f_2^(+ ) = re[a(1;+)ea^*(1;+ ) ] . the parameter @xmath6 is introduced in order to estimate the strength of the be effect , where the magnitude of the cross mark ( @xmath33 ) is expressed by @xmath6 .
of course @xmath34 should be less than one ( 1 ) , because it can be interpreted as the degree of coherence in quantum optics [ see also ref .
@xcite ] .
our result is given in fig .
[ fig2 ] and table [ table1 ] . as seen there
, the magnitude @xmath34 is larger than 1 .
this suggests that we have to consider also additional diagrams ( contributions ) to fig .
therefore we should seek other possible schemes for the description of @xmath0 channel .
[ table1 ] .analyses of @xmath0 and @xmath1 bec by delphi collaboration .
the systematic errors for all points are assumed to be @xmath35 .
small normalizations ( @xmath36 ) are attributed to the long range effect @xmath37 . [ cols="^,^,^,^,^,^,^",options="header " , ]
as explained in the previous section , we should consider more complex diagrams than fig .
[ fig1 ] . in this case
the following @xmath11 and @xmath13 in addition to @xmath38 and @xmath39 are necessary , because of the increased number of diagrams ( @xmath9 ) , [ eq8 ] pwa(2;-+- ) & = & e^ i ( k_1^(+ ) x_2 + k_2 x_3 + k_3 x_1 ) , + pwea(2;-+- ) & = & e^ i ( k_1^(+ ) x_2 + k_2 x_1 + k_3 x_3 ) , + pwa(3;+ ) & = & e^ i ( k_1^(+ ) x_3 + k_2 x_1 + k_3 x_2 ) , + pwea(3;+ ) & = & e^ i ( k_1^(+ ) x_3 + k_2 x_2 + k_3 x_1 ) .
these additional amplitudes described by means of the coulomb wave function are given as [ eq9 ] a(2;-+- ) & = & _ k_1k_2^c+-(x_2 , x_3 ) _
k_2k_3^c(x_3 , x_1 ) _
k_3k_1^c+-(x_1 , x_2 ) , + ea(2;-+- ) & = & _ k_1k_2^c+-(x_2 , x_1 ) _
k_2k_3^c(x_1 , x_3 ) _
k_3k_1^c+-(x_3 , x_2 ) , + a(3;+ ) & = & _ k_1k_2^c+-(x_3 , x_1 ) _
k_2k_3^c(x_1 , x_2 ) _
k_3k_1^c+-(x_2 , x_3 ) , + ea(3;+ ) & = & _ k_1k_2^c+-(x_3 , x_2 ) _
k_2k_3^c(x_2 , x_1 ) _
k_3k_1^c+-(x_1 , x_3 ) . the interferometry effect for @xmath0 channel described by eqs .
( [ eq3 ] ) and ( [ eq9 ] ) is given as @xmath40\ , \label{eq10}\end{aligned}\ ] ] where [ eq11 ] f_1^(+ ) & = & 16 \{_i = 1 ^ 3 a(i;c_1c_2c_3)a^*(i;c_1c_2c_3 ) + _ i=1 ^ 3 ea(i;c_1c_2c_3)ea^*(i;c_1c_2c_3 ) } + & & 1 , + f_2^(+ ) & = & 16 [ _ i = 1 ^ 3 a(i;c_1c_2c_3)ea^*(i;c_1c_2c_3 ) + c. c. ] + & & ( typical pw ) e^(3/2)i [ ( k_2-k_3 ) x_2 + ( k_3-k_2 ) x_3 ] , + f_3^(+ ) & = & 16 [ _ i = 1 ^ 3_j i a(i;c_1c_2c_3)ea^*(j;c_1c_2c_3 ) + c. c. ] + & & ( typical pw ) e^(3/2)i [ ( k_1-k_3 ) x_1 + ( k_3-k_1 ) x_3 ] , + f_4^(+ ) & = & 16 [ _ i = 1 ^ 3_j i a(i;c_1c_2c_3)a^*(j;c_1c_2c_3 ) + & & + _ i = 1 ^ 3_j i ea(i;c_1c_2c_3)ea^*(j;c_1c_2c_3 ) ] + & & ( typical pw ) e^(3/2)i [ ( k_1-k_3 ) x_1 + ( k_3-k_2 ) x_2 + ( k_2-k_1 ) x_3 ] , in the above equations for the sake of simplicity we neglect the suffix @xmath12 . in actual analyses
, however we should fix the charge assignment of @xmath12 in @xmath41 .
moreover , we have to introduce a new parameter ( @xmath8 ) to describe the strength of the shadow parts in @xmath42 and @xmath43 , cf . ,
[ fig4 ] . a possible interpretation of the role of the shadow region is a `` @xmath44-meson - like contribution '' occurring here in order to satisfy the conservation of the neutral current , cf . ,
[ fig5 ] .
the results obtained by our new formula , i.e. , eq .
( [ eq10 ] ) , are given in fig .
[ fig6 ] and table [ table1 ] .
as seen in table [ table1 ] , @xmath34 becomes negative for large @xmath8 and exceeds unity for small @xmath8 .
the possibly sets of parameters @xmath45 leading to reasonable results are @xmath46 , @xmath47 , @xmath48 and @xmath49 . to estimate the strength of @xmath34 we shall now analyze bec data on @xmath50 .
to analyze data on 3@xmath51 bec we can use the following formula presented in ref .
@xcite @xmath52\,,\qquad \label{eq12}\end{aligned}\ ] ] where [ eq13 ] f_1^(3 ^ - ) & = & 16 _ i=1 ^ 6 a(i)a^*(i ) , + f_2^(3 ^ - ) & = & 16 [ a(1)a^*(2 ) + a(1)a^*(3 ) + a(1)a^*(4 ) + a(2)a^*(5 ) + & & + a(2)a^*(6 ) + a(3)a^*(5 ) + a(3)a^*(6 ) + a(4)a^*(5 ) + & & + a(4)a^*(6 ) + c. c. ] , + f_3^(3 ^ - ) & = & 16 [ a(1)a^*(5 ) + a(1)a^*(6 ) + a(2)a^*(3 ) + a(2)a^*(4 ) + & & + a(3)a^*(4 ) + a(5)a^*(6 ) + c. c. ] , and [ eq14 ] a(1 ) & = & _ k_1k_2^c(x_1 , x_2 ) _ k_2k_3^c(x_2 ,
x_3 ) _ k_3k_1^c(x_3 , x_1 ) + & & e^
i k_12 r_12 e^
i k_23 r_23 e^ i k_31 r_31 = e^ ( 3/2)i ( k_1 x_1 + k_2 x_2 + k_3 x_3 ) , + a(2 ) & = & _ k_1k_2^c(x_1 , x_3 ) _
k_2k_3^c(x_3 , x_2 ) _
k_3k_1^c(x_2 , x_1 ) + & & e^
i k_12 r_13 e^
i k_23 r_32 e^ i k_31 r_21 = e^ ( 3/2)i ( k_1 x_1 + k_2 x_3 + k_3 x_2 ) , + a(3 ) & = & _ k_1k_2^c(x_2 , x_1 ) _
k_2k_3^c(x_1 , x_3 ) _
k_3k_1^c(x_3 , x_2 ) + & & e^
i k_12 r_21 e^ i k_23 r_13 e^ i k_31 r_32 = e^ ( 3/2)i ( k_1 x_2 + k_2 x_1 + k_3 x_3 ) , + a(4 ) & = & _ k_1k_2^c(x_2 , x_3 ) _
k_2k_3^c(x_3 , x_1 ) _
k_3k_1^c(x_1 , x_2 ) + & & e^
i k_12 r_23 e^
i k_23 r_31 e^ i k_31 r_12 = e^ ( 3/2)i ( k_1 x_2 + k_2 x_3 + k_3 x_1 ) , + a(5 ) & = & _ k_1k_2^c(x_3 , x_1 ) _
k_2k_3^c(x_1 , x_2 ) _
k_3k_1^c(x_2 , x_3 ) + & & e^
i k_12 r_31 e^
i k_23 r_12 e^ i k_31 r_23 = e^ ( 3/2)i ( k_1 x_3 + k_2 x_1 + k_3 x_2 ) , + a(6 ) & = & _ k_1k_2^c(x_3 , x_2 ) _
k_2k_3^c(x_2 , x_1 ) _
k_3k_1^c(x_1 , x_3 ) + & & e^
i k_12 r_32 e^
i k_23 r_21 e^ i k_31 r_13 = e^ ( 3/2)i ( k_1 x_3 + k_2 x_2 + k_3 x_1 ) .
the result obtained by this formula is given in fig .
[ fig7 ] and table [ table2 ] .
notice that fixed @xmath53 results in large @xmath54 .
this means that an additional parameter is necessary here , i.e. , that we should allow the degree of coherence ( @xmath34 ) to vary as well .
as seen in table [ table2 ] we have found that there is a common region @xmath55 for @xmath0 channel and @xmath56 for @xmath50 channel .
[ table2 ] ccccccc formulas & @xmath8 & @xmath36 & @xmath31 [ fm ] & @xmath34 & @xmath57 & @xmath58 + eq .
( [ eq12 ] ) & & [email protected] & [email protected] & 1.0 ( fixed ) & [email protected] & 20.6/31 + eq .
( [ eq12 ] ) & & [email protected] & [email protected] & [email protected] & [email protected] & 4.0/30 + + eq .
( [ eq10 ] ) & [email protected] & [email protected] & [email protected] & 0.7 ( fixed ) & [email protected] & 1.2/29 +
we have obtained the new formula for @xmath0 channel in the unlike-3rd order bec , introducing the degree of coherence ( @xmath34 ) and the effective magnitude of neutral current ( @xmath8 ) .
we have analyzed bec data on @xmath0 and @xmath1 channels in @xmath4 collision at @xmath3 gev , using the new formula .
( notice that to compare the obtained here values of @xmath31 with those obtained by using the plane wave approach , we have to multiply it by the factor @xmath27 , i.e. , @xmath60 fm . ) as seen from table [ table2 ] , the following choice of parameters is possible which leads to good fit to data , @xmath61 provided that the degree of coherence ( @xmath34 ) is almost the same in both channels .
this finite @xmath8 suggests that there is the genuine 3rd order contribution even in @xmath0 channel .
channel would be available , in which case the usefulness of eq .
( [ eq10 ] ) could be checked again . ] to confirm the choice of this set of parameters we need other data at @xmath3 gev as well as at different energies . moreover , our formula i.e. , eq .
( [ eq10 ] ) can be also applied to the same kind data from heavy - ion collisions .
one of authors ( m. b. ) would like to thank the scandinavia - japan sasakawa foundation for financial support , and authors are also thankful for g. wilk s reading the manuscript .
they are indebted to useful conversations with h. boggild , t. csorgo , b. lorstad and l. muresan . 99 p. abreu _ et al . _ [ delphi collaboration ] , phys .
b * 355 * ( 1995 ) 415 .
m. biyajima , a. bartl , t. mizoguchi , o. terazawa and n. suzuki , prog .
* 84 * ( 1990 ) 931 [ addendum - ibid .
* 88 * ( 1992 ) 157 ] ; see also , m. biyajima , phys . lett .
b * 92 * ( 1980 ) 193 .
m. biyajima , t. mizoguchi , t. osada and g. wilk , phys . lett .
b * 353 * ( 1995 ) 340 ; ibid * 366 * ( 1996 ) 394 . t. osada , s. sano and m. biyajima , z. phys .
c * 72 * ( 1996 ) 285 .
g. alexander , rept .
* 66 * ( 2003 ) 481 .
e. o. alt , t. csorgo , b. lorstad and j. schmidt - sorensen , phys .
b * 458 * ( 1999 ) 407 .
t. mizoguchi and m. biyajima , phys .
b * 499 * ( 2001 ) 245 .
l. i. schiff , _ quantum mechanics , 2nd ed .
_ , ( mcgraw - hill , new york , 1955 ) . | in order to analyze data on charged pions correlation channels , @xmath0 and @xmath1 , we propose new interferometry approach using the coulomb wave function .
we show that to describe adequately data we have to introduce new parameter describing the contribution of @xmath2 process . using this new formula
we analyze data on @xmath0 and @xmath1 channels at @xmath3 gev by delphi collaboration , and estimate the magnitude of this new parameter as well as the degree of coherence .
bose - einstein correlation , coulomb wave function , @xmath4 annihilation |
a graph @xmath0 is said to be @xmath11-saturated if + it does not contain @xmath1 as a subgraph , but + the addition of any new edge ( from @xmath12 ) creates a copy of @xmath1 .
+ let @xmath13 denote the _ minimum _ size of an @xmath1-saturated graph on @xmath4 vertices .
given @xmath1 , it is difficult to determine @xmath13 because this function is not necessarily monotone in @xmath4 , or in @xmath1 .
recent surveys are by j. faudree , gould , and schmitt @xcite , and by pikhurko @xcite on the hypergraph case .
it is known @xcite that for every graph @xmath1 there exists a constant @xmath14 such that @xmath15 holds for all @xmath4 .
however , it is not known if the @xmath16 exists ; pikhurko @xcite has an example of a four graph set @xmath17 when @xmath18 oscillates , it does not tend to a limit .
since the classical theorem of erds , hajnal , and moon @xcite ( they determined @xmath19 for all @xmath4 and @xmath20 ) , and its generalization for hypergraphs by bollobs @xcite , there have been many interesting hypergraph results ( e.g. , kalai @xcite , frankl @xcite , alon @xcite , using lovsz algebraic method ) but here we only discuss the graph case .
remarkable asymptotics were given by alon , erds , holzman , and krivelevich @xcite ( saturation and degrees ) .
bohman , fonoberova , and pikhurko @xcite determined the sat - function asymptotically for a class of complete multipartite graphs .
more recently , for multiple copies of @xmath21 faudree , ferrara , gould , and jacobson @xcite determined @xmath22 for @xmath23 .
what is the saturation number for the cycle , @xmath8 ?
this has been considered by various authors , however , in most cases it has remained unsolved . here
relatively tight bounds are given .
[ th:1 ] for all @xmath24 and @xmath25 @xmath26 the construction giving the upper bound is presented at the end of this section , the proof of the lower bound ( which works for all @xmath27 ) is postponed to section [ s : sat_lower ] .
the case of @xmath28 is trivial ; the cases @xmath29 and @xmath30 were established by ollmann @xcite in 1972 and by ya - chen @xcite in 2009 , resp .
@xmath31 actually , ( [ fur : eqc5 ] ) was conjectured by fisher , fraughnaugh , langley @xcite .
later ya - chen @xcite determined @xmath32 for all @xmath4 , as well as all extremal graphs .
the best previously known general lower bound came from barefoot , clark , entringer , porter , szkely , and tuza @xcite , and the best upper bound ( a clever , complicated construction resembling a bicycle wheel ) came from gould , uczak , and schmitt @xcite @xmath33 where @xmath34 for @xmath9 even @xmath35 , @xmath36 for @xmath9 odd @xmath37 .
although there is still a gap , theorem [ th:1 ] supersedes all earlier results for @xmath38 except the construction giving @xmath39 for @xmath40 from @xcite .
our new construction for a @xmath9-cycle saturated graph for @xmath41 can be read from the picture below . to be precise , define the graph @xmath42 on @xmath4 vertices , for arbitrary @xmath43 as follows .
write @xmath4 in the form @xmath44 where @xmath45 is an integer and @xmath46 .
the vertex set @xmath47 consists of the pairwise disjoint sets @xmath48 , @xmath49 , @xmath50 , @xmath51 , and @xmath52 for @xmath53 , @xmath54 where @xmath55 , @xmath56 , @xmath57 , and @xmath58 and @xmath59 , @xmath60 , @xmath61 , @xmath62 , @xmath63 .
we also denote @xmath64 by @xmath65 and @xmath66 by @xmath67 .
the edge set of @xmath1 does not contain @xmath68 and it consists of an almost complete graph @xmath69 minus an edge on @xmath70 , a @xmath71 minus an edge on @xmath72 , @xmath73 pending edges connecting @xmath74 and @xmath75 , and @xmath76 paths @xmath77 of length @xmath78 with vertex sets @xmath79 with endpoints @xmath80 and @xmath81 .
the number of edges @xmath82 it is not difficult to check that , indeed , @xmath1 is @xmath8-saturated ( see details in section [ s : hk ] ) .
after which , a little calculation yields the upper bound in ( [ eq : th1 ] ) .
we strongly believe that this construction is essentially optimal .
[ conj:1 ] there exists a @xmath83 such that @xmath84 holds for each @xmath85 .
first we check that @xmath42 is @xmath8-free .
if a cycle with vertex set @xmath88 is entirely in @xmath65 , then it is contained in @xmath89 , so @xmath90 . if @xmath88 contains a vertex @xmath91 then @xmath92 , the @xmath78 edges of the path @xmath77 are part of the cycle .
however , it is impossible to join @xmath80 and @xmath81 by a path of length 3 , so @xmath93 . the key observation to know that @xmath1 is @xmath8-saturated
is that @xmath80 and @xmath81 are connected inside @xmath65 by a path @xmath94 of any other lengths @xmath95 except for @xmath96 @xmath97 for example , @xmath98 is @xmath99 , @xmath100 , @xmath101 , etc .
also the vertices @xmath102 ( @xmath103 ) and @xmath104 are connected by a path @xmath105 of length @xmath106 inside @xmath65 for @xmath107 .
@xmath108 note that this is true for any @xmath109 but we will apply ( [ eq : um ] ) only for @xmath110 . now add an edge @xmath2 to @xmath1 from its complement .
we distinguish four disjoint cases .
@xmath111case 1.if @xmath2 is contained in the induced cycle @xmath112 then we get a path connecting @xmath80 and @xmath81 in @xmath112 of length @xmath76 , where @xmath76 is at least two and at most @xmath113 .
this path with @xmath114 form a @xmath9-cycle .
@xmath111case 2.if the endpoints of @xmath2 are @xmath91 and @xmath115 with @xmath116 then we may suppose that @xmath117 .
the vertex @xmath91 splits the path @xmath77 into two parts , @xmath118 and @xmath119 , where @xmath118 starts at @xmath80 and has length @xmath120 , and @xmath119 ends at @xmath121 and has length @xmath122 .
consider the path @xmath123 , its length is @xmath124 .
this length is between 3 and @xmath125 so we can apply ( [ eq : tl ] ) to add an appropriate @xmath126 to complete @xmath127 to a @xmath9-cycle unless @xmath128 . in the latter , the edge @xmath99 together with @xmath129 , @xmath2 , and @xmath119 form a @xmath8 .
@xmath111case 3.if the endpoints of @xmath2 are @xmath91 and @xmath130 , then again by symmetry , we may suppose that @xmath131 , so the length of @xmath118 is at most @xmath132 .
then , by ( [ eq : um ] ) there is an @xmath133 so that @xmath118 , @xmath2 and @xmath133 form a @xmath9-cycle .
@xmath111case 4.finally , @xmath2 is contained in @xmath65 .
+ for @xmath134 we use @xmath135 to get the @xmath9-cycle @xmath136 , + for @xmath137 we have the @xmath9-cycle @xmath138 , + for @xmath139 we have to use @xmath135 , i.e. , here we need again that @xmath45 , + for @xmath140 we have the @xmath9-cycle @xmath141 , + for @xmath142 we have the @xmath9-cycle @xmath143 , finally + for @xmath144 we have the @xmath9-cycle @xmath145 .
a graph @xmath0 is @xmath1-*semisaturated * ( formerly called _ strongly _ saturated ) if @xmath3 contains more copies of @xmath1 than @xmath0 does for @xmath146 .
let @xmath147 be the minimum size of an @xmath1-semisaturated graph . obviously , @xmath148 it is known that @xmath149 ( it follows from frankl / alon / kalai generalizations of bollob ' as set pair theorem ) and @xmath150 ( tuza @xcite ) .
below we have a @xmath151-semisaturated graph on @xmath152 vertices and @xmath153 edges .
every vertex can be reached by a path of length 2 from @xmath154 .
joining one , two or three triangles to the central vertex @xmath154 one obtains @xmath151 semisaturated graphs with @xmath155 , @xmath156 , or @xmath157 vertices and @xmath158 , @xmath159 , or @xmath160 edges , resp .
leaving out a pendant edge , we can extend these constructions for even values of @xmath4 @xmath161 the picture on the right is the extremal @xmath151-saturated graph by ( [ fur : eqc5 ] ) .
maybe equality holds in ( [ eq : c5 ] ) for @xmath163 . since @xmath164 inequalities ( [ fur : eqc5 ] ) and ( [ eq : c5 ] )
imply that @xmath165 our next theorem shows that a similar statement holds for every cycle @xmath8 with @xmath166 ( and probably for @xmath167 , too ) .
[ th:2 ] for all @xmath168 @xmath169 the proof of the lower bound is postponed to section [ s : ssat_lower3 ] . the construction yielding the upper bound is presented in the next two sections where we describe a way to improve the @xmath170 term as well as give better constructions for @xmath171 .
we believe that our constructions are essentially optimal .
[ conj:2 ] there exists a @xmath83 such that @xmath172holds for each @xmath85 .
in this section we define an infinite class of @xmath8-semisaturated graphs , @xmath173 ( more precisely @xmath174 ) . call a graph @xmath0 @xmath9-_suitable _ with special vertices @xmath80 and @xmath81 if @xmath111(s1)@xmath0 is @xmath8-semisaturated , @xmath111(s2)@xmath175 a path @xmath94 in @xmath0 with endpoints @xmath80 and @xmath81 and of length @xmath95 for all @xmath176 , and @xmath111(s3 ) for every @xmath177 , and integers @xmath178 and @xmath179 with @xmath180 and @xmath181 @xmath111 @xmath175 an @xmath182 and a path @xmath183 of length @xmath184 and with endpoints @xmath102 and @xmath185 . for example , it is easy to see , that a * wheel * with @xmath73 spikes @xmath186 is such a graph , @xmath187 , @xmath188 .
it is defined by the @xmath189-element vertex set @xmath190 and by @xmath191 edges joining @xmath80 to all other @xmath102 s , forming a cycle @xmath192 of length @xmath193 , and joining each @xmath75 to @xmath102 .
define the graph @xmath174 as follows , when @xmath4 is in the form @xmath194 where @xmath195 is an integer .
the vertex set @xmath47 consists of the pairwise disjoint sets @xmath65 and @xmath52 for @xmath53 , @xmath196 where @xmath197 , @xmath198 and @xmath199 .
the edge set of @xmath1 consists of a copy of @xmath0 with vertex set @xmath65 , and @xmath76 paths with endpoints @xmath80 and @xmath81 and vertex sets @xmath112 .
the number of edges is @xmath200 it is not difficult to check that , indeed , @xmath1 is @xmath8-semisaturated , the details are similar ( but much simpler ) to those in section [ s : hk ] , so we do not repeat that proof .
finally , considering @xmath201 ( where now @xmath202 ) we obtain that for all @xmath203 @xmath204 [ cor : c5 ] @xmath205
in this section we define another infinite class of @xmath8-semisaturated graphs , @xmath206 ( more precisely @xmath207 ) yielding the upper bound ( [ eq : th2 ] ) in theorem [ th:2 ] .
call a graph @xmath0 @xmath208-_suitable _ with special vertices @xmath80 and @xmath81 if ( s1 ) and ( s2 ) hold but ( s3 ) is replaced by the following @xmath111(s3)@xmath209 for every @xmath177 , and integers @xmath210 either there exists a path @xmath211 @xmath111 ( of length @xmath178 and with endpoints @xmath80 and @xmath185 ) or a path @xmath212 in the following cases @xmath111@xmath180 and @xmath213 , @xmath111@xmath214 and @xmath215 .
it is easy to see , that the wheel @xmath186 with @xmath73 spikes is such a graph , @xmath216 , @xmath188 .
define the graph @xmath207 as follows , when @xmath4 is in the form @xmath217 where @xmath218 is an integer and @xmath219 .
the vertex set @xmath47 consists of the pairwise disjoint sets @xmath65 , @xmath52 and @xmath51 for @xmath53 , @xmath220 where @xmath197 , @xmath221 , @xmath222 and @xmath199 .
the edge set of @xmath1 consists of a copy of @xmath0 with vertex set @xmath65 , and @xmath76 paths with endpoints @xmath80 and @xmath81 and vertex sets @xmath112 and finally @xmath223 spikes , a matching with edges from @xmath224 to @xmath51 .
the number of edges is @xmath225 it is not difficult to check that @xmath1 is @xmath8-semisaturated , the details are similar ( but simpler ) to those in section [ s : hk ] . as an example we present one case .
add the edge @xmath226 to @xmath1 where @xmath177 and @xmath227 .
let us denote the ( unique ) neighbor of @xmath228 by @xmath229 , @xmath230 .
the distance of @xmath229 to @xmath80 is denoted by @xmath95 .
then the length of the @xmath231 path is @xmath232 and the length of the @xmath233 path is @xmath234 and one can find a @xmath8 through @xmath226 using property ( s3)@xmath209 . considering @xmath235 ( with @xmath218 ) we obtain from ( [ eq : n ] ) and ( [ eq : c3 ] ) that for all @xmath236 @xmath237 using @xmath238 , it is easy to see that ( [ eq:10 ] ) holds for all @xmath239 , leading to the upper bound in ( [ eq : th2 ] ) . one can slightly improve ( [ eq:6 ] ) and ( [ eq:10 ] ) if there are special graphs thinner than the wheel @xmath240 .
[ prob : special ] determine @xmath241 , the minimum size of a @xmath9-vertex @xmath9-special graph ( i.e. , one satisfying ( s1)(s3 ) ) .
determine @xmath242 , the minimum size of a @xmath9-vertex @xmath243-special graph ( i.e. , one satisfying ( s1 ) , ( s2 ) and ( s3)@xmath209 ) .
suppose that @xmath0 is a @xmath8-semisaturated graph where @xmath244 , @xmath245 . obviously , @xmath0 is connected .
let @xmath246 be the set of vertices of degree one , @xmath247 , its size is @xmath248 and its elements are denoted as @xmath249 . denote the neighbor of @xmath250 by @xmath251 , @xmath252 and let @xmath253 .
we also denote the neighborhood of any vertex @xmath254 by @xmath255 or briefly by @xmath256 . [
le : degree1 ] ( the neighbors of degree one vertices.)(i ) @xmath257 for @xmath258 , so @xmath259 .
( ii ) @xmath260 for every @xmath261 , ( iii ) if @xmath262 , then @xmath263 is also a @xmath8-semisaturated graph . _
proof._if @xmath264 , then the addition of @xmath265 to @xmath0 does not create a new @xmath9-cycle .
if @xmath266 and @xmath267 , the addition of @xmath268 to @xmath0 does not create a new @xmath9-cycle .
finally , ( iii ) is obvious .
split @xmath88 and @xmath269 according to the degrees of their vertices .
thus , divide @xmath270 into five parts @xmath271 , @xmath272 and @xmath273 , @xmath274 and @xmath275 .
[ le : structure ] ( the structure of @xmath8-saturated graphs .
see @xcite ) .
suppose that @xmath0 is a @xmath8-saturated graph ( and @xmath244 ) . then ( iv ) if @xmath276 is a path in @xmath0 ( with @xmath277 , @xmath278 ) , then @xmath279 .
so there are no edges from @xmath280 to @xmath88 ( or to @xmath246 ) .
( v ) if @xmath281 is an edge of @xmath0 ( with @xmath282 ) , then @xmath283 . so there are no edges in @xmath284 , no edges from @xmath284 to @xmath285 . in other words ,
every @xmath286 has one neighbor in @xmath246 and two in @xmath287 .
( vi ) the induced graph @xmath288 $ ] consists of paths of length at most @xmath125 .
[ cl:8 ] suppose that @xmath0 is a @xmath8-semisaturated graph on @xmath4 vertices with minimum degree at least @xmath289 , @xmath244
. then every vertex @xmath290 is contained in some cycle of length at most @xmath291 .
consider two arbitrary vertices @xmath292 in the neighborhood @xmath293 .
if @xmath294 , then @xmath290 is contained in a triangle .
if @xmath295 , then @xmath296 contains a new @xmath9-cycle ; there is a path @xmath297 of length @xmath298 in @xmath0 with endpoints @xmath299 and @xmath300 .
if @xmath297 avoids @xmath290 , then @xmath297 together with @xmath301 form a @xmath291 cycle .
if @xmath290 splits @xmath297 into two paths @xmath302 , where @xmath303 starts in @xmath304 and ends in @xmath290 , then either @xmath305 , or @xmath306 , or both form a proper cycle of length at most @xmath193 .
note that claim [ cl:8 ] itself ( and the connectedness of @xmath0 ) immediately imply @xmath307 we can do a bit better repeatedly using the semisaturatedness of @xmath0 .
[ le : lower ] suppose that @xmath0 is a @xmath8-semisaturated graph on @xmath4 vertices with minimum degree at least @xmath289 , @xmath244 .
then @xmath308 _ proof._we define an increasing sequence of subgraphs @xmath309 with vertex sets @xmath310 such that @xmath311 is a subgraph of @xmath312 and @xmath313 ( for @xmath314 ) .
this , together with @xmath315 imply the claim .
@xmath316 is the shortest cycle in the graph @xmath0 .
its length is at most @xmath291 so ( [ eq:14 ] ) obviously holds .
if @xmath311 is defined and one can find a path @xmath297 of length at most @xmath9 with endpoints in @xmath317 but @xmath318 , then we can take @xmath319 . from now on , we suppose that such a _ short returning _ path does not exist .
our procedure stops if @xmath320 . in the case of @xmath321
, the connectedness of @xmath0 implies that there exists an @xmath322 edge with @xmath323 and @xmath324 . since @xmath325
we have another edge @xmath326 , @xmath327 .
we have @xmath328 , otherwise one gets a path @xmath329 of length smaller than @xmath9 with endpoints in @xmath317 but going out of @xmath311 , contradicting our earlier assumption .
similarly , we obtain that @xmath330 contains no edge , otherwise we can define @xmath331 as either @xmath332 plus the three edges of a triangle @xmath322 , @xmath333 , @xmath334 or we add four edges @xmath322 , @xmath335 , @xmath336 , and @xmath337 but only three vertices ( namely @xmath154 , @xmath299 , and @xmath300 ) .
the obtained @xmath312 obviously satisfies ( [ eq:13 ] ) in both cases .
similarly , if there is a cycle @xmath50 of length at most @xmath193 contaning @xmath154 , then we can define @xmath331 as @xmath332 plus @xmath338 and @xmath322 . from now on , we suppose that such a _ short cycle through _ @xmath154 does not exist .
fix a neighbor @xmath339 of @xmath154 , @xmath327 .
since @xmath340 , @xmath0 contains a path @xmath297 of length @xmath193 with endvertices @xmath229 and @xmath339 .
since @xmath0 does not contain a short returning path nor a short cycle through @xmath154 , we obtain that @xmath297 avoids @xmath154 and @xmath341 .
if the cycle @xmath342 of length @xmath291 has any diagonal edge then @xmath312 is obtained by adding @xmath50 together with its diagonals . from now on
, we suppose that @xmath50 does not have any diagonals
. more generally , if there is any _ diagonal _ _ path _ @xmath297 of length @xmath343 with edges disjoint from @xmath344 but with endpoints in @xmath345 then we can define @xmath346 and have added @xmath347 vertices and @xmath348 edges , obviously satisfying ( [ eq:13 ] ) .
however , such a diagonal path exists .
let @xmath349 be the other neighbor of @xmath229 in @xmath50 .
since @xmath350 , there is a path @xmath351 of length @xmath193 with endpoints @xmath290 and @xmath339 .
this @xmath351 must have edges outside @xmath344 so it can be shortened to a diagonal path @xmath297 of length at most @xmath193 .
this completes the proof of the lemma .
in this section we finish the proof of theorem [ th:2 ] .
let @xmath0 be a @xmath8-semisaturated graph on @xmath4 vertices with minimum number of edges , @xmath244 .
let @xmath246 be the set of degree one vertices , @xmath352 .
by lemma [ le : degree1 ] @xmath353 , and for @xmath354 we have @xmath355 and @xmath356 is a @xmath8-semisaturated graph on @xmath357 vertices with minimum degree at least 2
. then lemma [ le : lower ] can be applied to @xmath358 .
we obtain @xmath359 since @xmath360 , this is already a better lower bound than the one in ( [ eq : old ] ) from @xcite .
in this section we finish the proof of theorem [ th:1 ] .
let @xmath0 be a @xmath8-saturated graph on @xmath4 vertices , @xmath244 .
let us consider the partition of @xmath361 defined in section [ s : ssat_lower1 ] , where @xmath246 is the set of degree one vertices , @xmath88 is their neighbors . by lemma [ le : degree1 ] @xmath362 . to simplify notations we use @xmath363 , @xmath364 , @xmath365 , and @xmath366 .
we have @xmath367 by definition of the parts we have the lower bound @xmath368 this yields @xmath369 now we estimate the number of edges by considering four disjoint subsets of @xmath370 .
the part @xmath246 is adjacent to @xmath371 edges , and according to lemma [ le : structure ] , @xmath280 is adjacent to at least @xmath372 edges , @xmath284 is adjacent to exactly @xmath373 edges from which @xmath374 has already been counted at @xmath246 , and finally @xmath375 is adjacent to at least another @xmath376 edges .
we obtain @xmath377 therefore we get @xmath378 d. c. fisher , k. fraughnaugh , l. langley , _ on @xmath151-saturated graphs with minimum size _ , proceedings of the twenty - sixth southeastern international conference on combinatorics , graph theory and computing ( boca raton , fl , 1995 ) , _ congr . numer_.
* 112 * ( 1995 ) , 4548 .
l. taylor ollmann , _
@xmath385 saturated graphs with a minimal number of edges _ , in proceedings of the third southeastern conference on combinatorics , graph theory , and computing ( florida atlantic univ . ,
boca raton , fla . , 1972 ) , pages 367392 , boca raton , fla . , 1972 .
florida atlantic univ . | a graph @xmath0 is called @xmath1-saturated if it does not contain any copy of @xmath1 , but for any edge @xmath2 in the complement of @xmath0 the graph @xmath3 contains some @xmath1 .
the minimum size of an @xmath4-vertex @xmath1-saturated graph is denoted by @xmath5 .
we prove @xmath6 holds for all @xmath7 , where @xmath8 is a cycle with length @xmath9 .
we have a similar result for semi - saturated graphs @xmath10 we conjecture that our three constructions are optimal . |
the intricate formation of the large scale structure of the present - day universe is formed by an interplay between random gaussian fluctuations and gravitational instability . when gravitational instabilities start dominating the dynamical evolution of the matter content of the universe , the structure formation evolves from a linear to a highly nonlinear regime . in this framework , voids are formed in minima and haloes are formed in maxima of the same primordial gaussian field , and later on these features present different type dynamical evolutions due to their initial conditions in the nonlinear regime .
this fact has been known since early studies that showed voids are integral features of the universe @xcite .
@xcite show that the distribution of voids can be affected by their environments . as a result
, the void size distribution may play a crucial role to understand the dynamical processes affecting the structure formation of the universe @xcite .
the early statistical models of void probability functions ( vpfs ) @xcite are based on the counts in randomly placed cells following the prescription of @xcite .
apart from vpfs , the number density of voids is another key statistic to obtain the void distribution .
recently @xcite show that void size distributions obtained from the cosmic void catalog ( cvc ) satisfy a @xmath0-parameter log - normal probability function .
this is particularly interesting , because observations and theoretical models based on numerical simulations of galaxy distributions @xcite show that the galaxy mass distribution satisfies a log - normal function rather than a gaussian .
taking into account that voids are integral features of the universe , it may be expected to obtain a similar distribution profile for voids . apart from this , @xcite discuss a possible quantitative relation between the shape parameters of the void size distribution and the environmental affects . following up on the study of @xcite , we here extend their analysis of void size distributions to all simulated and mock samples of cvc of @xcite .
the three main catalogs under study are dark matter , halo and galaxy catalogs .
therefore , we confirm that the system of @xmath0-parameter log - normal distribution obtained by @xcite provides a fairly satisfactory model of the size distribution of voids .
in addition to this , we obtain equations which satisfy linear relations between maximum tree depth and the shape parameters of the void size distribution as proposed by @xcite .
extending the study by @xcite , we here fully investigate the void size distribution function statistically in simulations and mocks catalogs of the public cvc of @xcite .
it is useful to note that all the data of cvc are used here generated from a @xmath2 cold dark matter ( @xmath2cdm ) n - body simulation by using an adaptive treecode 2hot @xcite .
in addition , in all data sets voids are identified with the modified version of the parameter - free void finder zobov @xcite .
the data sets of cvc , we used here , can be categorized into three main groups ; * dark matter ( dm ) simulations are dm full , dm dense and dm sparse .
although these dark matter simulations have the same cosmological parameters from the wilkinson microwave anisotropy probe ( wmap ) seven year data release ( dr@xmath3 ) @xcite as well as the same snapshot at z=0 , they have different tracer densities of @xmath4 , @xmath5 , and @xmath6 particles per @xmath7 which are respectively dm full , dm dense , and dm sparse .
also the minimum effective void radii @xmath8 and @xmath9 mpc / h are obtained from the simulations for dm full , dm dense and dm sparse respectively .
* halo catalog in which two halo populations are generated ; haloes dense and haloes sparse . in the halo catalog the halo positions are used as tracers to find voids .
the minimum resolvable halo mass of haloes dense is @xmath10 @xmath11 while the minimum resolvable halo mass of haloes sparse data set is @xmath12 @xmath11 . in addition , the minimum effective void radii of haloes dense and sparse are @xmath13 mpc / h respectively . the main reason to construct these halo populations with different minimum resolvable halo masses is to compare the voids in halos to voids in relatively dense galaxy environments , see @xcite for more details . * galaxy catalogues
; there are two galaxy mock catalogues which are produced from the above halo catalog by using the halo occupation distribution ( hod ) code of @xcite and the hod model by @xcite .
these galaxy mock catalogs are called hod dense and hod sparse @xcite .
the hod dense catalog has @xmath14 voids with effective minimum radii @xmath15 mpc / h and includes relatively high - resolution galaxy samples with density @xmath5 dark matter particles per cubic mpc / h matching the sloan digital sky survey ( sdss ) dr@xmath3 main sample @xcite using one set of parameters found by @xcite ( @xmath16 , @xmath17 , @xmath18 , @xmath19 ) .
the hod sparse mock catalog consists of @xmath20 voids with @xmath9 @xmath21 effective minimum radii ( @xmath22 @xmath21 ) and this void catalog represents a relatively low resolution galaxy sample with density @xmath23 particles per cubic @xmath21 matching the number density and clustering of the sdss dr@xmath24 galaxy sample @xcite using the parameters found by @xcite ( @xmath25 , @xmath26 @xmath27 , @xmath28 , @xmath29 , and @xmath30 chosen to fit the mean number density ) .
in addition to this another mock galaxy catalog is used here ; n - body mock catalog which is a single hod mock galaxy catalog in real space at @xmath31 , generated by a dark matter simulation of @xmath32 particles ( with a particle mass resolution @xmath33 @xmath34 in a @xmath35 gpc / h box and is tuned to sdss dr@xmath24 in full cubic volume by using the hod parameters found in @xcite and it consists of @xmath36 voids @xcite . although the n - body mock catalog is processed slightly differently than hod sparse and hod dense , it is a hod mock catalog and it uses planck first- year cosmological parameters @xcite . in the following section , we examine the above data sets from a statistical perspective such as histograms , parameters of location ( range , mean , median ) , mode or dispersion ( standard deviation ) and shape ( skewness , kurtosis ) by following the previous study of @xcite . from a statistical perspective , we also investigate the connection between the distribution and the environment of void populations .
as a first step , the raw data plots of void size distributions are obtained for dm full , dm dense , dm sparse , haloes dense and haloes sparse .
note that the void size distributions for hod dense , hod sparse and n - body mock data sets are discussed in great detail in @xcite . in the raw void
size distributions , an unexpected local peak is observed around the value @xmath37 mpc / h in the dm full sample and in the dm dense sample around @xmath38 mpc / h , see fig 1 upper left and lower left panels . a similar behavior is observed by @xcite in the n - body mock sample around the value @xmath39 mpc / h .
[ cols= " < , < " , ]
here , extending our previous study of @xcite to attempt to find a universal void size distribution , we investigate the statistical properties of the void size distribution such as the shape parameters and their relations to the void environment of cvc by using the moment method by following @xcite . as aforementioned ,
the moment method is easy to apply .
therefore , we here confirm our previous result on the size distributions of voids which states that the @xmath0-parameter log - normal distribution gives a satisfactory model of the size distribution of voids , which is obtained from simulation and mock catalogs of cvc ; n - body mock , dm full , dm dense , dm sparse , haloes dense , haloes sparse , hod sparse and hod dense ( see fig 6 , also fig 3 in @xcite ) . on the other hand , we should keep in mind that all the data sets of cvc are generated by a single n - body simulation that operates counting scales as nlogn .
therefore the nature of these data sets may enforce us to obtain such a unique void size distribution . at this point being critical
is essential before stating that there is a universal void size distribution satisfying the @xmath0-parameter log - normal . as a result , a thorough investigation of the void size distribution by using other voids catalogs to unveil the truth beyond the relation between the shape of the void size distribution and the void environment has a key importance .
especially , taking into account that @xcite point out some problems and inconsistencies in cvc such as the identification of some overdense regions as voids in the galaxy data of the sdss dr@xmath3 @xcite .
processing from the problems of the cvc , @xcite provide an alternative public catalogue of voids , obtained from using an improved version of the same watershed transform algorithm .
therefore , it is essential to extend our analysis of void size distributions to the catalog given by @xcite .
again , this is particularly important to confirm whether the @xmath0-parameter log - normal void size distribution is valid in a different void catalog .
if the 3-parameter log - normal distribution fits another simulated / mock void catalogues , then this may indicate that voids have universal ( redshift independent ) size distributions given by the log - normal probability function .
apart from this , @xcite show that void average density profile can be represented by an empirical function in @xmath2cdm n - body simulations by using zobov .
this function is universal across void size and redshift . following this , @xcite investigate the density profiles of voids which are identified by again using the zobov in mock luminous red galaxy catalogues from the jubilee simulation , and in void catalogues constructed from the sdss lrg and main galaxy samples . as a result ,
@xcite show that the scaled density profiles of real voids show a universal behavior over a wide range of galaxy luminosities , number densities and redshifts .
processing from these results , there is a possibility that the @xmath0-parameter log - normal void size distribution may be a universal distribution for voids in simulated as well as real data samples .
that is why , it is critical to extend our analysis to other simulated as well as real data sets
. we also observe that the number of nonzero and zero void central densities in the samples have important effects on the shape of the @xmath0-parameter log - normal void size distributions . as is seen in table 1 and fig 1 ,
if the percentage of number of nonzero central densities reaches @xmath40 in a simulated or mock sample in cvc , then a second population emerges in the void size distribution .
this second population presents itself as a second peak in the log - normal size distribution , at larger radius . also , we here obtain a linear relation between the maximum tree depth and the skewness of the samples , and this relation is given by equation [ eqn : mtd - skew ] ( see fig 3 ) .
this linear relation indicates that if there is a void in a simulated / mock sample with a high maximum tree depth , then we expect more skewed log - normal distribution .
therefore , there is a direct correlation between the void substructure and the skewness of the void size distribution .
the possibility of this relation is mentioned by @xcite .
therefore , we here confirm that the skewness of a void size distribution is a good indicator of void substructures in a simulated / mock sample .
aforementioned , the minimum radius cut of cvc samples defined by two density - based criterion as given by @xcite can affect the relation skewness - mtd .
the minimum radius cut of cvc is particularly important because it may not only affect the resulting skewness of the data sets but also other shape parameters of the void size distributions , which can violate the confirmation of the log - normal distribution of the samples .
that is why , it is important to study raw samples to understand the effect of the minimum radius - cut .
in addition to skewness - mtd linear relation , another linear correlation is obtained between the maximum tree depth and the variance of the sparse samples of cvc ( see equation [ eqn : mtd - variance - sparse ] ) . as is seen from fig 4 ,
sparse samples with high maximum tree depth tend to be more dispersed than a sparse sample with lower maximum tree depth . on the other hand , although we obtain a linear relation between the maximum tree depth and the variance of the dense samples ( see equation [ eqn : mtd - variance - dense ] ) , due to the lack of dense samples with merging tree depth , this relation does not provide enough information to define a relation between these parameters ( see fig 4 , red dotted line ) .
but it is obvious that the relation between maximum tree depth and variance shows two different behaviors for the sparse and the dense samples .
this is an expected result since variance is the indicator of dispersion by definition .
while sparse samples are highly dispersed with high variance values , the dense samples are expected to show lower variance / dispersion than the sparse samples .
these relations indicate that there is a direct correlation between the shape parameters of the void size distribution such as skewness , variance and the void substructures .
our next goal is to address the following questions : is it possible to relate the shape parameters of the void size distribution to the environment in real data samples ? do the shape parameters change in time , indicating the dynamical evolution of the void size distribution ?
is the @xmath0-parameter log - normal void size distribution universal ?
the authors would like to thank paul sutter and his team for constructing and sharing the cosmic void catalog .
all void samples used here can be found in folder void catalog : 2014.06.08 at http://www.cosmicvoids.net .
abazajian , k. n. 2009 , , 182 , 543 bernardeau , f. 1992 , , 392 , 1 bernardeau , f. 1994 , , 291 , 697 bouchet , f. r. , strauss , m. a. , & davis , m. et al .
1993 , , 417 , 36 chincarini g. , & rood h. j. 1975 , nature , 257 , 294 coles , p. , & jones , b. 1991 , , 248 , 1 conover , w. j. 1980 , practical nonparametric statistics , 2nd ed .
, new york , ny . , john wiley & sons conover , w. j. 1980 , practical nonparametric statistics , 3rd ed .
, new york , ny . , john wiley & sons croton , d. j. , farrar , g. r. , & norberg , p. et al .
2005 , , 356 , 1155 dawson , k. s. et al .
2013 , , 145 , 10 einasto , j. , joeveer , m. , & saar e. 1980 , nature , 283 , 47 elizalde , e. , & gaztanaga , e. 1992 , , 254 , 247 fry , j. n. 1986 , , 306 , 358 goldberg , d. m. , & vogeley , m. s. 2004 , , 605 , 1 gregory s. a. , & thompson l. a. 1978 , , 222 , 784 hamilton , a. j. s. 1985 , , 292 , l35 hamaus , n. , sutter , p. m. , & wandelt , b. d. 2014 , physical review d , 112 , 25 hoyle , f. , rojas , r. r. , vogeley , m. s. , & brinkmann , j. et al .
2005 , , 620 , 618 johnson , n.l , kotz , s , & balakrishnan , n. 1994 , continuous univariate distributions , vol . 1 ( 2nd ed ; wiley ) kayo , i. , taruya , a. , & suto , y. 2001 , , 561 , 22 kendall , m. , & stuart , a. 1977 , the advanced theory of statistics .
vol 1 : distribution theory , 2nd ed . , ( new york , ny . , macmillan ) kofman , l. , bertschinger , e. , gelb , j. m. , nusser , a. , & dekel , a. 1994 , , 420 , 44 komatsu , e. et al .
2011 , , 192 , 18 lavaux , g. , & wandelt , b. d. 2012 , , 754 , 109 manera , m. , et al .
2013 , , 428 , 1036 neyrinck , m. c. 2008 , , 386 , 2101 nadathur , s. , & hotchkiss , s. 2014 , , 440 , 1248 nadathur , s. 2015 , , 449 , 3997 planck collaboration , 2014 , , 571 , a19 russell , e. 2013 , , 436 , 3525 russell , e. 2014 , , 438 , 1630 pycke , j .-
, & russell , e. 2016 , , 821 , 110 sheth , r. k. , & van de weygaert , r. 2004 , , 350 , 517 sheskin , d. j. 2011 , handbook of parametric and nonparametric statistical procedures ( boca raton : fl , crc press ) strauss , m. a. 2002 , , 124 , 1810 sutter , p. m. , lavaux , g. , wandelt , b. d. , & weinberg , d. h. 2012 , , 761 , 44 sutter , p. m. , lavaux , g. , hamaus , n. , wandelt , b. d. , weinberg , d. h. , & warren , m. s. , 2014 , , 462 , 471 sutter , p. m. , lavaux , g. , wandelt , b. d. , weinberg , d. h. , warren , m. s. , & pisani , a. , 2014 , , 442 , 3127 taylor , a. n. , & watts , p.
i. r. 2000 , , 314 , 92 tinker , j. l. , weinberg , d. h. , & zheng , z. , 2006 , , 368 , 85 van de weygaert r. , & platen e. 2011 , international journal of modern physics conference series , 1 , 41 , s. d. m. 1979 , , 186 , 145 zehavi , i. et al .
2011 , , 736 , 59 zheng , z. , coil , a. l. , & zehavi , i. 2007 , , 667 , 760 | following up on previous studies , we here complete a full analysis of the void size distributions of the cosmic void catalog ( cvc ) based on three different simulation and mock catalogs ; dark matter , haloes and galaxies . based on this analysis
, we attempt to answer two questions : is a @xmath0-parameter log - normal distribution a good candidate to satisfy the void size distributions obtained from different types of environments ? is there a direct relation between the shape parameters of the void size distribution and the environmental effects ? in an attempt to answer these questions
, we here find that all void size distributions of these data samples satisfy the @xmath0-parameter log - normal distribution whether the environment is dominated by dark matter , haloes or galaxies .
in addition , the shape parameters of the @xmath0-parameter log - normal void size distribution seem highly affected by environment , particularly existing substructures .
therefore , we show two quantitative relations given by linear equations between the skewness and the maximum tree depth , and variance of the void size distribution and the maximum tree depth directly from the simulated data .
in addition to this , we find that the percentage of the voids with nonzero central density in the data sets has a critical importance . if the number of voids with nonzero central densities reaches @xmath1 in a simulation / mock sample , then a second population is observed in the void size distributions .
this second population emerges as a second peak in the log - normal void size distribution at larger radius . |
let @xmath3 be a sample of the variable set @xmath0 where @xmath1 is an indicator variable and @xmath2 is an explanatory variable .
conditionally on @xmath2 , @xmath1 follows a bernoulli distribution with parameter @xmath4 .
usual examples are response variables @xmath1 to a dose @xmath2 or to an expository time @xmath2 , economic indicators .
the variable @xmath2 may be observed at fixed values @xmath5 ,
@xmath6 on a regular grid @xmath7 or at irregular fixed or random times @xmath8 , for a continuous process @xmath9 . exponential linear models with known link functions
are often used , especially the logistic regression model defined by @xmath10 with a parametric function @xmath11 .
the inverse function of @xmath12 is easily estimated using maximum likelihood estimators of the parameters and many authors have studied confidence sets for the parameters and the quantiles of the model . in a nonparametric
setting and for discrete sampling design with several independent observations for each value @xmath13 of @xmath2 , the likelihood is written @xmath14^ { 1_{\{x_i = x_j\}}}.\ ] ] the maximum likelihood estimator of @xmath15 is the proportion of individuals with @xmath16 as @xmath17 , @xmath18 regular versions of this estimator are obtained by kernel smoothing or by projections on a regular basis of functions , especially if the variable @xmath2 is continuous .
let @xmath19 denote a symmetric positive kernel with integral 1 , @xmath20 a bandwidth and @xmath21 , with @xmath22 as @xmath23 .
a local maximum likelihood estimator of @xmath12 is defined as @xmath24 or by higher order polynomial approximations @xcite . under regularity conditions of @xmath12 and @xmath19 and ergodicity of the process @xmath25
, the estimator @xmath26 is @xmath27-uniformly consistent and asymptotically gaussian .
when @xmath12 is monotone , the estimators are asymptotically monotone in probability . for large @xmath28 , the inverse function @xmath29
is then estimated by @xmath30 if @xmath12 is decreasing or by @xmath31 if @xmath12 is increasing .
the estimator @xmath32 is also @xmath27-uniformly consistent and asymptotically gaussian @xcite . for small samples ,
a monotone version of @xmath33 using the greatest convex minorant or the smallest concave majorant algorithm may be used before defining a direct inverse .
other nonparametric inverse functions have been defined @xcite . under bias sampling ,
censoring or truncation , the distribution function of @xmath1 conditionally on @xmath2 is not always identifiable .
the paper studies several cases and defines new estimators of conditional and marginal distributions , for a continuous bivariate set @xmath0 and for a conditional bernoulli variable @xmath1 .
in case - control studies , individuals are not uniformly sampled in the population : for rare events , they are sampled so that the cases of interest ( individuals with @xmath16 ) are sufficiently represented in the sample but the proportion of cases in the sample differs from its proportion in the general population @xcite .
let @xmath34 be the sampling indicator of individual @xmath35 in the global population and @xmath36 the distribution function of @xmath37 conditionally on @xmath38 is given by @xmath39 let @xmath40 for individual @xmath35 , @xmath41 is observed conditionally on @xmath42 and the conditional distribution function of @xmath43 is defined by @xmath44 the probability @xmath45 is deduced from @xmath46 and @xmath47 by the relation @xmath48 and the bias sampling is @xmath49 the model defined by @xmath50 is over - parameterized and only the function @xmath51 is identifiable .
the proportion @xmath46 must therefore be known or estimated from a preliminary study before an estimation of the probability function @xmath12 . in the logistic regression model , @xmath52
$ ] is replaced by @xmath53 = \psi(x)-\log \theta$ ] .
obviously , the bias sampling modifies the parameters of the model but not this model and the only stable parametric model is the logistic regression .
let @xmath54 be the inverse of the proportion of cases in the population , @xmath55 under the bias sampling , @xmath56 @xmath54 is modified by the scale parameter @xmath57 : it becomes @xmath58 .
the product @xmath59 may be directly estimated by maximization of the likelihood and @xmath60 in a discrete sampling design with several independent observations for fixed values @xmath13 of the variable @xmath2 , the likelihood is @xmath61^ { 1_{\{x_i = x_j \}}}\ ] ] and @xmath62 is estimated by @xmath63 for random observations of the variable @xmath2 , or for fixed observations without replications , @xmath64 is estimated by @xmath65 if @xmath46 is known , nonparametric estimators of @xmath12 are deduced as @xmath66
consider that @xmath1 is observed under a fixed truncation of @xmath2 : we assume that @xmath67 is observed only if @xmath68 $ ] , a sub - interval of the support @xmath69 of the variable @xmath2 , and @xmath70}(x)$ ] .
then @xmath71 and the conditional probabilities of sampling , given the status value , are @xmath72 if the ratio @xmath73 is known or otherwise estimated , the previous estimators may be used for the estimation of @xmath45 from the truncated sample with @xmath74 .
for a random truncation interval @xmath75 $ ] , the sampling indicator is @xmath76}(x)$ ] and the integrals of @xmath12 are replaced by their expectation with respect to the distribution function of @xmath77 and @xmath78 and the estimation is similar .
consider then @xmath0 a two - dimensional variable in a left - truncated transformation model : let @xmath1 denote a response to a continuous expository variable @xmath2 , up to a variable of individual variations @xmath79 independent of @xmath2 , @xmath80 @xmath81 with distribution function @xmath82 .
the distribution function of @xmath1 conditionally on @xmath2 is defined by @xmath83 and the function @xmath84 is continuous .
the joint and marginal distribution functions of @xmath2 and @xmath1 are denoted @xmath85 , with support @xmath86 , @xmath87 , with bounded support @xmath69 , and @xmath88 , such that @xmath89 and @xmath90 the observation of @xmath1 is supposed left - truncated by a variable @xmath91 independent of @xmath67 , with distribution function @xmath92 : @xmath1 and @xmath91 are observed conditionally on @xmath93 and none of the variables is observed if @xmath94 .
denote @xmath95 for any distribution function @xmath96 and , under left - truncation , @xmath97 obviously , the mean of @xmath1 is biased under the truncation and a direct estimation of the conditional distribution function @xmath98 is of interest for the estimation of @xmath99 instead of the apparent mean @xmath100 .
the function @xmath101 is also written @xmath102 with @xmath103 and the expressions ( [ ayx])-([byx ] ) of @xmath77 and @xmath78 imply that @xmath104 and @xmath105 .
an estimator of @xmath106 is obtained as the product - limit estimator @xmath107 of @xmath108 based on estimators of @xmath77 and @xmath78 : for a sample @xmath109 , let @xmath110 in @xmath111 $ ] and @xmath112 with @xmath113 .
that is a nonparametric maximum likelihood estimator of @xmath98 , as is the kaplan - meier estimator for the distribution function of a right - censored variable .
then an estimator of @xmath114 may be defined as an estimator of @xmath115 , @xmath116 by the same arguments , from means in ( [ ayx])-([byx ] ) , @xmath117 is estimated by @xmath118 the distribution function @xmath92 is simply estimated by the product - limit estimator for right - truncated variables @xcite @xmath119 and an estimator of @xmath120 is deduced from those of @xmath98 , @xmath87 and @xmath84 as @xmath121 the means of @xmath91 and @xmath122 are estimated by @xmath123 the estimators @xmath124 and @xmath125 are known to be @xmath27-uniformly consistent and asymptotically gaussian . for the further convergence restricted to the interval @xmath126 , assume c1 . @xmath127 and @xmath128 as @xmath23 , @xmath129 , @xmath130 and @xmath131 .
the conditional probability @xmath51 is strictly positive in the interior of @xmath69 , + c3 .
the distribution function @xmath132 is twice continuously differentiable with respect to @xmath110 and differentiable with respect to @xmath133 .
@xmath134 for a @xmath135 in @xmath136 $ ] .
let us denote @xmath137 , @xmath138 , and @xmath139 .
[ cvanbn ] @xmath140 and @xmath141 , @xmath142 if @xmath143 , @xmath144 and @xmath145 converge in distribution to gaussian processes with mean zero , variances @xmath146 and @xmath147 respectively , and the covariances of the limiting processes are zero .
the proof relies on an expansion of the form @xmath148 with @xmath149 and @xmath150 , where @xmath151 @xmath152 a similar approximation holds for @xmath153 .
the biases and variances are deduced from those of each term and the weak convergences are established as in @xcite . from proposition [ cvanbn ] and applying the results of the nonparametric regression , [ cvfnmn ] the estimators @xmath154 , @xmath155 ,
@xmath156 converge @xmath27-uniformly to @xmath98 , @xmath84 , @xmath120 , @xmath157 and @xmath158 converge @xmath27-uniformly to @xmath159 and @xmath160 respectively .
the weak convergence of the estimated distribution function of truncated survival data was proved in several papers @xcite . as in @xcite and by proposition
[ cvanbn ] , their proof extends to their weak convergence on @xmath161 under the conditions @xmath162 and @xmath163 on @xmath164 , which are simply satisfied if for every @xmath110 in @xmath164 , @xmath165 and @xmath166 .
[ cvfyx ] @xmath167 converges weakly to a centered gaussian process @xmath168 on @xmath86 .
the variables @xmath169 , for every @xmath110 in @xmath164 , and @xmath170 converge weakly to @xmath171 and @xmath172 .
if @xmath84 is supposed monotone with inverse function @xmath173 , @xmath2 is written @xmath174 and the quantiles of @xmath2 are defined by the inverse functions @xmath175 and @xmath176 of @xmath98 at fixed @xmath133 and @xmath110 , respectively , are defined by the equivalence between @xmath177 where @xmath178 is the inverse of @xmath120 at @xmath179 . finally ,
if @xmath84 is increasing , @xmath180 is decreasing in @xmath110 and increasing in @xmath133 , and it is the same for its estimator @xmath181 , up to a random set of small probability
. the thresholds @xmath175 and @xmath176 are estimated by @xmath182 as a consequence of theorem [ cvfyx ] and generalizing known results on quantiles for @xmath183 , @xmath184 converges @xmath27-uniformly to @xmath185 on @xmath186 .
for every @xmath133 and ( respect . ) @xmath110 , @xmath187 and @xmath188 converge weakly to the centered gaussian process @xmath189^{-1}$ ] and , respect .
, @xmath190^{-1}$ ] .
the variable @xmath1 is supposed left - truncated by @xmath91 and right - censored by a variable @xmath122 independent of @xmath191 .
the notations @xmath51 and those of the joint and marginal distribution function of @xmath2 , @xmath1 and @xmath91 are in section [ ytruncation ] and @xmath192 is the distribution function of @xmath122 .
the observations are @xmath193 , and @xmath194 , conditionally on @xmath195 .
let @xmath196 the estimators are now written @xmath197 if @xmath1 is only right - truncated by @xmath122 independent of @xmath67 , with observations @xmath67 and @xmath122 conditionally on @xmath198 , the expressions @xmath51 , @xmath77 and @xmath78 are now written @xmath199 the distribution function @xmath192 and @xmath98 are both identifiable and their expression differs from the previous ones , @xmath200 the estimators are now @xmath201 if @xmath1 is left and right - truncated by variables @xmath91 and @xmath122 independent and independent of @xmath67 , the observations are @xmath0 , @xmath122 and @xmath91 , conditionally on @xmath202 , @xmath203 the distribution functions @xmath192 , @xmath92 and @xmath98 are identifiable , with @xmath98 defined by @xmath204 and @xmath205.\end{aligned}\ ] ] their estimators are @xmath206 the other nonparametric estimators of the introduction and the results of section [ ytruncation ] generalize to all the estimators of this section .
right and left - truncated distribution functions @xmath98 and the truncation distributions are estimated in a closed form by the solutions a self - consistency equation @xcite .
the estimators still have asymptotically gaussian limits even with dependent truncation distributions , when the martingale theory for point processes does not apply .
consider model ( [ model ] ) with an independent censoring variable @xmath122 for @xmath1 . for observations by intervals , only @xmath122 and
the indicators that @xmath1 belongs to the interval @xmath207-\infty , c]$ ] or @xmath207c , \infty[$ ] are observed .
the function @xmath98 is not directly identifiable and efficient estimators for @xmath84 and @xmath98 are maximum likelihood estimators .
let @xmath208 and assume that @xmath120 is @xmath209 .
conditionally on @xmath122 and @xmath210 , the log - likelihood of @xmath211 is @xmath212 and its derivatives with respect to @xmath114 and @xmath120 are @xmath213 for every @xmath214 s.t . @xmath215 and @xmath216 . with @xmath217
, @xmath218 then @xmath219 belongs to the tangent space for @xmath120 and the estimator of @xmath220 must be determined from the estimator of @xmath120 through the conditional probability function of the observations @xmath221 pinon , c. and pons , o. ( 2006 ) .
nonparametric estimator of a quantile function for the probability of event with repeated data , in _ dependence in probability and statistics _ , _ lecture notes in statistics _ , * 187 * , pp . 475489 .
springer , new york .
pons , o.(2007 ) .
estimation for the distribution function of one- and two - dimensional censored variables or sojourn times of markov renewal processes . _ communications in statistics - theory and methods _ ,
* 36 * issue 14 , 25272542 . | efficient estimation under bias sampling , censoring or truncation is a difficult question which has been partially answered and the usual estimators are not always consistent .
several biased designs are considered for models with variables @xmath0 where @xmath1 is an indicator and @xmath2 an explanatory variable , or for continuous variables @xmath0 .
the identifiability of the models are discussed .
new nonparametric estimators of the regression functions and conditional quantiles are proposed . |
one of the most popular explanations of accelerated expansion of the universe is that it is caused by some scalar field , for which the strong condition of energy domination is nt satisfied : @xmath0 .
different models of scalar field are considered : quintessence , phantom , quintom , tachyon and other more exotic models of scalar field ( see for example , @xcite and citations there in ) . to describe its impact on expansion of the universe or other components the hydrodynamics approximation is used -
the field represented by some ideal fluid with equation of state @xmath1 , where @xmath2 is called the parameter of equation of state . in the simplest case
the quantity @xmath3 is a constant , but there are more general models , where parameter of equation of state changes with time . beside this , there are models , in which scalar field non - gravitationally interacts with other fields @xcite . in the lagrangian of scalar field and fields with which it interacts the additional term appears . in the general case
this lagrangian can be written in the form : @xmath4 , where @xmath5 - kinetic term of field , @xmath6 - the variable of scalar field , which is the dark energy , @xmath7 - the fields of other types of matter .
in general , the form of @xmath8 is unknown .
there are no any physical principle or any experimental data , from which it is possible to derive the lagrangian of interaction of scalar field of dark energy with other fields yet . nowadays ,
the attempts to build unified theory of field are being made , and the scalar field some how must be coupled with other fields .
one can read about possibility of such interaction and observational data which points on it in works @xcite .
below we analyse the impact of non - gravitational interaction between dark energy and dark matter on the dynamics of the expansion of the universe and on the evolution of the densities of these components . as we do nt know the form of the lagrangian of scalar field we describe its behaviour phenomenologically .
evolution of homogeneous and isotropic universe and its components are described by system of einstein s equations and by equations which express the conservation laws of energy and momentum considering additional interaction between components .
we assume that dark energy interacts gravitationally and non - gravitationally with dark matter only and interacts gravitationally only with baryonic matter or relativistic matter .
all components are described by ideal fluid approximation . as the basis we took the model of dark energy @xcite which had been studied as minimally coupled , which gives the possibility to distinguish the impact of interaction between dark components on dynamics of the expansion of the universe and evolution of their densities .
the following features have been generally accepted as properties of the observational universe : homogeneity , isotropy , flatness of 3-space and the accelerated expansion at large scales .
the metric of 4-space of such world is the friedmann - lemaitre - robertson - walker metric ( flrw ) , which in comoving spherical coordinates in conformal representation have such form : @xmath9,\ ] ] where @xmath10 is the diagonal metric tensor with only one unknown function @xmath11 , the scale factor , which describes the expansion of the universe .
hereafter indices @xmath12 take the values 0 , 1 , 2 , 3 , where @xmath13-component will be always a time variable . in the model of universe with zero curvature of 3-space
it is convenient to normalize it by 1 at the present epoch : @xmath14 .
the first and second time derivatives of this factor describe the rate and the acceleration of the expansion of the universe , i.e. the dynamics of such expansion .
the variable @xmath15 is conformal time , which is related with physical cosmological time @xmath16 by the simple differential relation : @xmath17 , @xmath18 - the speed of light .
the large set of independent observational data ( see the chapter 1 in the book @xcite ) indicates that @xmath19 and @xmath20 , or in conformal time @xmath21 and @xmath22 , hereafter the dot denotes the derivative @xmath23 . traditionally in cosmology the expansion rate of the universe
is described by relative quantity @xmath24 , which is called the hubble parameter , and the acceleration by the dimensionless quantity @xmath25 , which is called the deceleration parameter : @xmath26 the equations for them are given by the einstein s equations ( 1915 ) , which in general case have the form @xmath27 where @xmath28 is the ricci tensor of covariant curvature , @xmath29 is its convolution , the scalar of 4-space curvature and @xmath30 is the total covariant energy - momentum tensor of the components of the universe .
we specify the energy - momentum tensor in the following way : consider that the universe is uniformly filled by non - relativistic matter - baryonic matter ( @xmath31 ) and dark matter ( @xmath32 ) , relativistic one ( @xmath33 ) - cosmic microwave background and relict neutrino , and also dark energy ( @xmath34 ) . each of it will be described by energy - momentum tensor of ideal fluid , @xmath35 where @xmath36 is the ( energy ) density of @xmath37-component , @xmath38 - its pressure , and @xmath39 - 4-vector of velocity .
then the einstein s equations give the equations for @xmath24 and @xmath25 @xmath40 called the friedmann equations . hereafter the variables , in which @xmath41 , was used .
to solve those equations one would need to specify the equations of state of each component and non - gravitational interaction between them , if such is present .
lets take the equation of state in the form @xmath42 , with the given known parameters of equation of state for non - relativistic and relativistic components ( @xmath43 , @xmath44 ) and unknown for dark energy @xmath45
. we shall find it for scalar - field dark energy from the condition that the ratio of variations of pressure to density is constant and from specific form of non - gravitational interaction with dark matter .
to find the dependencies of densities on time @xmath15 ( or on scale factor @xmath46 ) the equations ( [ h2])-([qh2 ] ) must be complemented by the conservation equations for each component . in general case ,
if there is any other interaction between components besides gravitational one , the equations of conservation of energy and momentum will be written in such form : @xmath47 where @xmath48 is the 4-vector of the change of density flow of energy - momentum to / from component @xmath37 as the result of interaction with other components , and `` ; '' - denotes the covariant derivative with respect to @xmath49 .
for minimally coupled components @xmath50 .
the second equation in ( [ jin ] ) is the result of bianchi identities .
hereafter we will suppose that non - gravitational interaction is present only between dark energy and dark matter , so the non - zeros are only @xmath51 - for dark energy and @xmath52 - for dark matter .
from the second equation in ( [ jin ] ) it follows that @xmath53 then the conservation laws for dark energy and dark matter will be written in following form : @xmath54 for baryonic matter and relativistic matter the conservation laws will be with zero flow of energy - momentum , caused by interaction , because we shall analyse the dynamics of expansion of universe in the epoch after recombination , when their motion is free and they interact only gravitationally : @xmath55 the conservation equations ( [ tei_dedm ] ) and ( [ tei_br ] ) for homogeneous isotropic universe contain only the equations of discontinuity , which in the metrics ( [ ds ] ) have the form @xmath56 since the universe is homogeneous and isotropic on large scales and the perturbations are absent , the 4-vector of energy - momentum flow @xmath57 has only one non - zero component @xmath58 - the change of energy density per unit time .
similarly to ( [ eq de ] ) and ( [ eq dm ] ) the equations for baryonic and relativistic components with zero right hand side give the well known expressions : @xmath59 , @xmath60 , hereafter up or down index ( 0 ) will mean the quantity at the present epoch .
now let us consider the dependence of parameter of equation of state @xmath61 on scale factor @xmath46 .
for this we shall introduce the new parameter - the adiabatic speed of sound @xmath62 .
then , using equation ( [ eq de ] ) , we shall obtain such differential equation for @xmath63 : @xmath64 in general case the quantity @xmath65 can depend on time , which must be given or obtained from known or given other physical properties of dark energy . in this work we shall consider it as constant : @xmath66 @xcite
. this equation must be solved together with equations ( [ eq de ] ) , ( [ eq dm ] ) , but for this the @xmath58 must be given or to derived from some considerations ( see the reviews @xcite ) .
as we do nt know anything about such interaction , it is naturally to suggest that it is a function of energies of these two components : @xmath67 for small values of densities it can be represented as @xmath68 where @xmath69 - the constants , which give the strength and sign of interaction .
we suppose , that interaction between dark energy and dark matter must nt depend on the expansion rate of the universe and conformal time @xmath15 explicitly .
therefore the expression for interaction @xmath58 we took in the form ( [ jahf ] ) and ( [ jah ] ) with factor @xmath70 , which removes the explicit dependence of interaction between dark components on the hubble parameter @xmath24 and the conformal time @xmath15 , as such factor is present in the right hand side of equations ( [ eq de ] ) , ( [ eq dm ] ) .
this is obviously , if one switches from the differentiation with respect to @xmath15 to the differentiation with respect to @xmath46 in these equations : @xmath71 the change of dark components energy densities in this case is described by integral - differential equation , which can be solved numerically . in this article
we consider only some partial cases of such interaction , for which the analytical solutions exist : @xmath72 where @xmath73 is the critical density at present epoch . the component s densities at present epoch are convenient to represent in the units of critical one with the dimensionless parameter of density @xmath74 : @xmath75 . in literature also
the other kinds of interactions are considered , in particular , @xmath76 , where @xmath6 is the dark energy scalar field .
it can be rewritten in the form @xmath77 , where @xmath78 now depends on @xmath46 .
the models with such interaction and interaction ( [ jdm ] ) are considered in @xcite .
further we shall consider the three cases of interaction ( [ j0 ] ) , ( [ jde ] ) and ( [ jdm ] ) between dark energy and dark matter ( the de - dm interaction ) , which are reduced to analytical solutions of conservation equations ( [ eq de ] ) and ( [ eq dm ] ) for energy densities de and dm and the equation ( [ dw ] ) for equation of state parameter of dark energy .
henceforth we shall call the dark energy as quintessence , if its density decreases in the process of expansion of universe , and the dark energy , the density of which increases in the process of expansion of universe - the phantom .
consider the interaction ( [ j0 ] ) , which does not depend on the densities of dark components . in this case from equations ( [ eq de ] ) and ( [ dw ] ) for arbitrary @xmath79 we obtain @xmath80 and as the result we have obtained such equation for @xmath61 : @xmath81 this is the riccati equation , which has the partial solution : @xmath82 .
with its help we find the general solution : @xmath83}{(1+w_0)\omega_{de}-(w_0-c_a^2)\omega_{de}a^{3(1+c_{a}^{2 } ) } + \alpha(1-a^{3(1+c_{a}^{2})})}-1.\ ] ] one can see , that the character of equation of state parameter s evolution depends on the values of all parameters of dark energy and parameter of interaction @xmath84 .
moreover , if @xmath85 , then @xmath61 goes to @xmath86 , the equation of state parameter of minimally coupled dark energy @xcite : @xmath87 thus , the introduction of the even simplest interaction have changed the character of equation of state parameter s evolution significantly , which can be found from the comparison of fig .
[ fig : rcr1 ] with fig . 1 in @xcite and fig . 1 in @xcite .
in particular , the quintessence dark energy can gain in the future such properties , in which @xmath88 , while remaining the quintessence , namely such that density of which decreases in the process of expansion of the universe . and vice versa for phantom . considering ( [ rw ] ) and ( [ wc ] ) the conservation equations ( [ eq de ] ) and ( [ eq dm ] ) have exact analytical solutions : @xmath89 where @xmath90 is the known solution for minimally coupled dark energy ( @xmath91 ) @xmath92 which is smooth function of @xmath46 for the arbitrary parameters @xmath93 and @xmath94 @xcite .
the solutions ( [ r de0 ] ) and ( [ r dm0 ] ) are also the smooth functions for @xmath95 and arbitrary parameters of dark energy .
they allow the negative values of density of dark components for certain values of parameters of dark energy and the parameter of interaction .
the condition of positivity of density of dark matter for all @xmath46 is @xmath96 .
the conditions of positivity of density of quintessence dark energy for all @xmath46 are @xmath97 , and of phantom - @xmath98 .
thereby , the interaction ( [ j0 ] ) provides the positive values of density of dark components only in the range of values of parameter of interaction @xmath99 in the models with quintessence and phantom dark energy accordingly .
it is interesting , that in the case of quintessence dark energy the expansion of universe approaches exponential one and asymptotically constant energy densities of both dark components , and in the case of phantom - the singularity of big rip at constant values of density of dark matter . , the density of dark energy @xmath100 and the density of dark matter @xmath101 at the different values of interaction parameter @xmath84 ( the dotted line - @xmath102 , the solid line - @xmath103 , the dashed line - @xmath104 , the dash - dotted line - @xmath105 ) : a - the quintessence dark energy ( @xmath106 , @xmath107 ) , b - the phantom ( @xmath108 , @xmath109 ) . here
@xmath110 is the critical density at present epoch ( @xmath111 ) , @xmath112 , @xmath113,title="fig:",scaledwidth=43.0% ] , the density of dark energy @xmath100 and the density of dark matter @xmath101 at the different values of interaction parameter @xmath84 ( the dotted line - @xmath102 , the solid line - @xmath103 , the dashed line - @xmath104 , the dash - dotted line - @xmath105 ) : a - the quintessence dark energy ( @xmath106 , @xmath107 ) , b - the phantom ( @xmath108 , @xmath109 ) . here
@xmath110 is the critical density at present epoch ( @xmath111 ) , @xmath112 , @xmath113,title="fig:",scaledwidth=43.0% ] let us analyse the behaviour of the quantities @xmath61 and @xmath100 with the expansion of universe considering the conditions ( [ r demc1 ] ) and ( [ r demc2 ] ) . for the quintessence dark energy ( @xmath114 ) , when @xmath115 we have @xmath116 , @xmath117 , and when @xmath118 we have @xmath119 @xmath120 for the phantom dark energy ( @xmath121 ) , when @xmath115 we have @xmath119 @xmath122 and when @xmath118 we have @xmath116 , @xmath117 . in the fig .
[ fig : rcr1 ] the dependencies of quantities @xmath61 , @xmath100 and @xmath101 on @xmath46 in the range of its values @xmath123 $ ] are given .
we see , that with values of parameter @xmath84 at the upper bound for the model with quintessence dark energy ( @xmath124 ) when @xmath118 we have @xmath125 , @xmath126 . for the dark energy , with the parameters of phantom , for the upper bound ( @xmath127 )
when @xmath118 we have @xmath128 , @xmath129 , namely the dark energy in this case is similar to the cosmological constant .
we see also that at negative @xmath84 the density of dark matter @xmath101 in the process of expansion of universe at first is the positive , and then becomes negative .
let us consider now the impact of interaction ( [ j0 ] ) on the dynamics of expansion of universe .
for this we substitute in the friedmann equations ( [ h2 ] ) and ( [ qh2 ] ) the expressions for energy densities of dark energy ( [ r de0 ] ) and dark matter ( [ r dm0 ] ) founded above , and also the expressions for the density of baryonic matter @xmath130 and relativistic matter @xmath131 .
we calculate the values @xmath132 and @xmath25 for the standard cosmological model with @xmath112 , @xmath113 , @xmath133 , @xmath134 and @xmath135 .
the results are shown in the fig .
[ fig : rcr2 ] .
the dependencies of @xmath132 and @xmath25 on @xmath46 are given for the range of its values @xmath136 $ ] . here at @xmath137 for the quintessence dark energy , when @xmath115 , then @xmath138 , @xmath139 , and when @xmath118 , then @xmath140 for the phantom dark energy , when @xmath115 , then @xmath138 , @xmath139 , and when @xmath118 , then @xmath138 , @xmath141 .
for the upper bounds of interaction parameter @xmath142 for the quintessence dark energy we have @xmath138 , @xmath139 when @xmath115 and @xmath143 , @xmath144 when @xmath118 .
for the dark energy with the upper bound @xmath142 with the parameters of phantom we have the dark energy with constant @xmath61 and @xmath100 . in this case
@xmath138 , @xmath139 when @xmath115 and @xmath145 , @xmath144 when @xmath118 . and deceleration parameter @xmath25 at the different values of interaction parameter @xmath84 ( the dotted line - @xmath102 , the solid line - @xmath103 , the dashed line - @xmath104 , the dash - dotted line - @xmath105 ) : a - the model with quintessence dark energy , b - with the phantom .
the parameters of models are the same as in the fig .
[ fig : rcr1],title="fig:",scaledwidth=43.0% ] and deceleration parameter @xmath25 at the different values of interaction parameter @xmath84 ( the dotted line - @xmath102 , the solid line - @xmath103 , the dashed line - @xmath104 , the dash - dotted line - @xmath105 ) : a - the model with quintessence dark energy , b - with the phantom .
the parameters of models are the same as in the fig .
[ fig : rcr1],title="fig:",scaledwidth=43.0% ]
let the expression for density of energy flow @xmath58 is such as in ( [ jde ] ) . then substituting it in the equation ( [ dw ] ) for @xmath61
, we shall obtain @xmath146 this equation has the exact analytical solution @xmath147 where @xmath148 is the initial condition .
now , substituting the expression ( [ w1 ] ) for @xmath61 in expression ( [ rw ] ) for density @xmath100 , we obtain the exact analytical solution of equation ( [ eq de ] ) for @xmath149 : @xmath150 as one can see , the dependencies @xmath63 and @xmath149 have three parameters @xmath151 , which define the general properties and type of dark energy .
let us investigate the behaviour of @xmath61 and @xmath100 , when @xmath115 and @xmath118 .
if @xmath152 , then when @xmath115 we have @xmath116 , @xmath117 , and when @xmath118 we have @xmath153 , @xmath154 . if @xmath155 , then when @xmath115 we have @xmath153 , @xmath154 , and when @xmath118 we have @xmath116 , @xmath117 . at the values of parameters @xmath156 and @xmath157 , for which the inequalities @xmath158 and @xmath159 , or @xmath160 and @xmath161 are valid , the dependence of the equation of state parameter of dark energy @xmath61 on @xmath46 has the singularity of the second kind . from the definition of @xmath61 it is clear , that in that case the always smooth dependence @xmath162 changes the sign in the process of expansion of universe , namely @xmath163 at the point @xmath164}.\ ] ] namely , the solution ( [ r de1 ] ) allows also @xmath165 . we suppose , that the energy densities of the dark components of universe can be only positive as during all its past and at the present epoch , and always in the future . for the equation of state parameter and the density of dark energy
we see here the new behaviour , different from the case @xmath166 , studied in detail in the works @xcite.the phantom divide line is shifted by the quantity of interaction parameter @xmath157 and becomes equal @xmath167 , if the quintessence dark energy is define as such , the density of which decreases ( @xmath152 ) , and the phantom dark energy is define as such , the density of which increases ( @xmath155 ) in the process of expansion of the universe .
the condition that @xmath168 at arbitrary @xmath169 , for quintessence dark energy is the following : @xmath170 and for phantom one : @xmath171 now we shall find the dependence of density of dark matter @xmath101 on @xmath46 . substituting the expression ( [ r de1 ] ) in the equation ( [ eq dm ] ) , we obtain @xmath172 where @xmath173 note , that the right hand side of equation ( [ eq1 dm ] ) there is a regular function for an arbitrary finite value @xmath95 and arbitrary values of parameters @xmath93 , @xmath94 and @xmath157 . in the case of @xmath174 , when @xmath175 and @xmath176 , the solution of ( [ eq1 dm ] ) is @xmath177 the condition that @xmath178 for any @xmath95 , is the restriction of the range of values of the interaction parameter : @xmath179 so , in this particular case @xmath129 the interaction between the dark matter and dark energy of the form ( [ jde ] ) is such , that the energy flows from the dark energy to dark matter .
the flow rate decreases in the process of expansion of the universe and goes to an asymptotic regime @xmath180 such that @xmath181 when @xmath118 . , the density of dark energy @xmath100 and the density of dark matter @xmath101 at the different values of interaction parameter @xmath157 ( the dotted line - @xmath182 , the solid line - @xmath183 , the dashed line - @xmath184 , the dash - dotted line - @xmath185 ) : a - the quintessence dark energy ( @xmath106 , @xmath107 ) , b - the phantom one ( @xmath108 , @xmath109 ) . here
@xmath110 is the critical density at present epoch ( @xmath111 ) , @xmath112 , @xmath113,title="fig:",scaledwidth=43.0% ] , the density of dark energy @xmath100 and the density of dark matter @xmath101 at the different values of interaction parameter @xmath157 ( the dotted line - @xmath182 , the solid line - @xmath183 , the dashed line - @xmath184 , the dash - dotted line - @xmath185 ) : a - the quintessence dark energy ( @xmath106 , @xmath107 ) , b - the phantom one ( @xmath108 , @xmath109 ) . here
@xmath110 is the critical density at present epoch ( @xmath111 ) , @xmath112 , @xmath113,title="fig:",scaledwidth=43.0% ] the general solution of equation ( [ eq1 dm ] ) is @xmath186 .
\label{r dm1}\ ] ] one can see , that the solution is regular at the all interval @xmath95 for an arbitrary values of parameters @xmath93 , @xmath94 and @xmath157 .
if @xmath187 , then @xmath128 , and the expressions for energy densities of dark components ( [ r de1 ] ) and ( [ r dm1 ] ) coincide with the corresponding expressions in the work @xcite . in the particular case @xmath188 , which is the case for quintessence dark energy , @xmath189 and the expression ( [ r dm1 ] )
is simplified : @xmath190 it is easy to see , that in this case the condition @xmath178 for an arbitrary @xmath95 together with the condition @xmath168 ( [ um de1 ] ) is satisfied at @xmath191 and at the realistic values of equation of state parameter @xmath192 @xcite .
let us find the range of values @xmath157 , at which @xmath178 for an arbitrary @xmath95 and an arbitrary parameters of dark energy
. in the case of quintessence dark energy with @xmath168 the value of density of dark matter @xmath101 always will be positive with the condition @xmath193 and in the case of phantom - with condition @xmath194 here the conditions of positivity of density of dark energy ( [ um de1 ] ) and ( [ um de2 ] ) are take into account .
the upper bound of values of @xmath157 from the inequalities ( [ um dm0 ] ) and ( [ um dm1 ] ) we denote @xmath195 . on fig .
[ fig : de1 ] we represent the dependencies @xmath63 , @xmath149 and @xmath196 for the model of universe with quintessence ( left ) and phantom ( right ) dark energy and the three values of interaction parameter @xmath197 , and its upper limit value @xmath198 .
the parameters are chosen the same as in the fig .
[ fig : rcr1 ] .
the impact of value of interaction parameter @xmath157 on the evolution of equation of state parameter is equivalent to the shift of phantom divide line ( @xmath199 ) on the quantity @xmath157 . in the case of @xmath200 ,
when the energy flows from dark energy to dark matter , the density of quintessence dark energy decreases faster , and the density of phantom one increases slower , than in the case without of interaction ( @xmath166 ) . in the case of quintessence dark energy the density of dark matter goes to the constant value , and in the case of phantom - slowly goes to infinity .
also we see , that for the upper limit value of interaction parameter @xmath201 for the dark energy with parameters of phantom we have the case , which we have considered earlier , in which the quantities @xmath61 and @xmath100 are constant . in the case of @xmath202 ,
when the energy flows from dark matter to dark energy , the density of quintessence dark energy decreases slower , and of phantom increases faster , than in the case without the interaction .
the density of dark matter in this case decreases fast , and goes to zero and to the negative values in future , which we consider as non - physical behaviour .
this means , that the interaction of type ( [ jde ] ) can take place only when @xmath200 . and deceleration parameter @xmath25 at the different values of interaction parameter @xmath157 ( the dotted line - @xmath182 , the solid line - @xmath183 , the dashed line - @xmath184 , the dash - dotted line - @xmath185 ) : a - the quintessence dark energy , b - the phantom one .
the parameters of models are the same as in the fig .
[ fig : rcr1],title="fig:",scaledwidth=43.0% ] and deceleration parameter @xmath25 at the different values of interaction parameter @xmath157 ( the dotted line - @xmath182 , the solid line - @xmath183 , the dashed line - @xmath184 , the dash - dotted line - @xmath185 ) : a - the quintessence dark energy , b - the phantom one .
the parameters of models are the same as in the fig .
[ fig : rcr1],title="fig:",scaledwidth=43.0% ] consider now the impact of interaction ( [ jde ] ) on the dynamics of expansion of universe .
substituting the expressions for density of all components in the friedmann equations ( [ h2 ] ) and ( [ qh2 ] ) , we obtain the dependence of quantities @xmath203 and @xmath25 on @xmath46 .
the results are represented in the fig .
[ fig : de2 ] for the same parameters as in the fig .
[ fig : rcr2 ] : left - the model with quintessence dark energy , right - with phantom one . on the all panels of both figures there are also the curves , which correspond to the upper bounds of values @xmath157 from inequalities ( [ um dm0 ] ) and ( [ um dm1 ] ) . here
we see , that for the quintessence dark energy ( @xmath152 ) @xmath138 , @xmath139 when @xmath115 and @xmath204 , @xmath144 when @xmath118 . for the phantom ( @xmath155 )
@xmath138 , @xmath139 when @xmath115 and @xmath138 , @xmath205 when @xmath118 . for the upper bound of interaction parameter ( @xmath201 ) , for the dark energy with constant @xmath61 and @xmath100 with parameters of phantom @xmath138 , @xmath139
when @xmath115 and @xmath145 , @xmath144 when @xmath118 .
as we see , such interaction with the considered values of parameter @xmath157 has impact on the dynamics of expansion of universe in the past and present epochs ( on figure @xmath206 ) , but this impact is not large .
this means , that the high precision data is needed for finding the constraints of quantity @xmath157 .
let now the expression for energy density flow @xmath58 is such as in ( [ jdm ] ) .
substituting it in the equation ( [ eq dm ] ) for @xmath101 we obtain @xmath207 the general solution of this equation is the expression @xmath208 so , the density of dark matter is always smooth function , which takes only the positive values for an arbitrary @xmath95 and @xmath78 . in the case of @xmath209 ( the flow of energy from dark energy to dark matter ) the density of dark matter decreases slower for @xmath210 , than in the case of non - interacting components . the observational data on the large - scale structure of the universe and anisotropy of cosmic microwave background
constrain it value , @xmath211 . using the formula ( [ rw ] )
, we obtain the following ordinary differential equation for @xmath61 : @xmath212 as in the previous cases , we have the riccati equation with partial solution @xmath82 , using which the general solution can be found easily : @xmath213(1+c_{a}^{2 } ) } { 1+w_0+\gamma(1+c_{a}^{2})\frac{\omega_{dm}}{\omega_{de}}\frac{1-a^{3(c_a^2+\gamma)}}{c_a^2+\gamma}-(w_0-c_a^2)a^{3(1+c_a^2)}}-1\ ] ] we substitute it in the expression for @xmath100 ( [ rw ] ) and obtain @xmath214.\ ] ] first of all let us notice , that the obtained solutions for the density of dark matter @xmath196 and dark energy @xmath149 are the regular functions for all range of values of scale factor @xmath95 for an arbitrary values of parameters of dark energy @xmath215 , @xmath93 , @xmath94 and interaction parameter @xmath78 .
indeed , the expression ( [ r de2 ] ) is finite when @xmath216 , and when @xmath217 . in the particular case of constant @xmath61 ,
when @xmath187 , the expression ( [ r de2 ] ) coincide with the expression for the density of dark energy , given in the work @xcite .
let us analyse the behaviour of @xmath63 and @xmath149 when @xmath115 and @xmath118 considering the condition @xmath211 .
if @xmath114 , then for @xmath115 we have @xmath116 , @xmath117 , and for @xmath118 we have @xmath218 , @xmath219 and if @xmath220 this asymptotic value of the density of dark energy is positive .
while if @xmath121 , than when @xmath115 we have @xmath116 , @xmath117 , and when @xmath118 we have @xmath116 , @xmath117 .
so , as before , when @xmath114 we have the quintessence dark energy , and when @xmath121 - the phantom one . , the density of dark energy @xmath100 at the different values of interaction parameter @xmath78 ( the dotted line - @xmath221 , the solid line - @xmath222 , the dashed line - @xmath223 , the dash - dotted line - @xmath224 ) : a - the quintessence dark energy , b - the phantom .
the parameters are the same as in the fig .
[ fig : rcr1],title="fig:",scaledwidth=43.0% ] , the density of dark energy @xmath100 at the different values of interaction parameter @xmath78 ( the dotted line - @xmath221 , the solid line - @xmath222 , the dashed line - @xmath223 , the dash - dotted line - @xmath224 ) : a - the quintessence dark energy , b - the phantom .
the parameters are the same as in the fig .
[ fig : rcr1],title="fig:",scaledwidth=43.0% ] let us write now the conditions of positivity for the density of dark energy @xmath100 considering the condition @xmath225 for the realistic values of parameters @xmath226 , @xmath65 , @xmath215 , @xmath227 .
the density @xmath100 is always positive , if and only if @xmath228 and @xmath229 . for
the quintessence dark energy with @xmath230 the conditions are following : @xmath231 for the phantom dark energy when @xmath232 the conditions are following : @xmath233 from the conditions ( [ um2 de2 ] ) , ( [ um2 de3 ] ) one can see , that when @xmath234 the phantom dark energy can be the always positive only if @xmath235 . from the formula ( [ w2 ] )
one can see also , that there are such models , for which @xmath61 may cross the line @xmath236 , when the quantity @xmath100 is always positive .
we do not give here the conditions of positivity of density of dark energy for them because they are bulky and the parameter values region corresponding to them is narrow . and deceleration parameter @xmath25 at the different values of interaction parameter @xmath78 ( the dotted line - @xmath221 , the solid line - @xmath222 , the dashed line - @xmath223 , the dash - dotted line - @xmath224 ) : a - the quintessence dark energy , b - the phantom one .
the parameters of models are the same as in the fig .
[ fig : rcr1],title="fig:",scaledwidth=43.0% ] and deceleration parameter @xmath25 at the different values of interaction parameter @xmath78 ( the dotted line - @xmath221 , the solid line - @xmath222 , the dashed line - @xmath223 , the dash - dotted line - @xmath224 ) : a - the quintessence dark energy , b - the phantom one .
the parameters of models are the same as in the fig .
[ fig : rcr1],title="fig:",scaledwidth=43.0% ] in the fig .
[ fig : dm1 ] the behaviour of @xmath61 and @xmath100 for quintessence dark energy and phantom is shown for the same values of parameters as in the fig .
[ fig : rcr1 ] and for the values of interaction parameter @xmath237 . for the positive values of @xmath78 the quantity @xmath100 is always positive . for the quintessence dark energy
it is a monotonically decreasing function which asymptotically approaches to the constant value when @xmath118 , and for the phantom one it decreases at first to some minimal value , and then again increases to @xmath238 when @xmath118 . for the negative value @xmath239 for
both quintessence and phantom dark energies @xmath100 increases from @xmath240 to 0 and higher , it is the negative at first , and in the process of expansion becomes positive .
when changing the sign of @xmath100 on the opposite , as we see from the figures , @xmath61 has the singularity of second kind .
now , having the expressions for energy densities of components , the friedmann equations can be used for the analysis of impact of such model of interaction between the dark components on the dynamics of expansion of the universe .
we take following the expressions for energy densities of baryonic and relativistic matters : @xmath130 , @xmath131 . in the fig .
[ fig : dm2 ] the evolution of @xmath203 and @xmath25 is shown for the same values of parameters , as in the fig .
[ fig : rcr1 ] . for the interaction parameters @xmath237
one can see that for the quintessence dark energy @xmath138 , @xmath139 when @xmath115 , and @xmath241 when @xmath118 .
for the phantom dark energy @xmath138 , @xmath139 when @xmath115 and @xmath138 , @xmath141 when @xmath118 . under above conditions of positivity of @xmath100 ,
the quantity @xmath242 is positive always .
one can see , that this is true for the values of interaction parameter @xmath243 . for @xmath239 , despite of that @xmath100 becomes the negative , the total density of components stay always the positive , therefore the quantity @xmath242 is always positive , and @xmath25 smoothly goes from + 1 in the early universe to -1 in the future in the case of quintessence dark energy , or @xmath244 in the case of phantom .
the dynamics of expansion of the universe in the cosmological model with dynamical dark energy , which interacts with dark matter gravitationally and non - gravitationally was analysed .
the three types of interaction , which cause the energy - momentum exchange between them , - independent on densities of components , proportional to the density of dark energy and proportional to the density of dark matter were considered .
for all cases we have obtained the analytical dependencies @xmath63 , @xmath149 and @xmath196 , which are the exact solutions of the conservation equations of energy for the dark components .
it was shown , that in all cases the dependencies of energy densities on @xmath46 are the smooth function , which at the certain values of parameters of dark energy and the interaction parameter can take the negative values .
for each case the range of values of parameters , at which the density of dark components are positive for an arbitrary @xmath46 was found .
common for all cases is the condition of positivity of interaction parameter . in the representation ( [ j0])-([jdm ] ) this means , that only in the case of interaction in which the density of dark matter is always positive is the flow of energy from the dark energy to the dark matter , and the density of dark energy is positive , if and only if the value of interaction parameter not exceed some quantity , found for the each model .
the other common feature of such models are the non - zero asymptotic values of densities of dark components when @xmath118 , if the dark energy is the quintessence . if the dark energy is phantom one , then the asymptotic value of its density when @xmath115 is constant for all acceptable values of interaction parameter in the models ( [ j0 ] ) and ( [ jde ] ) . in the model ( [ jdm ] ) , when the energy flow is proportional to the density of dark matter , there are possible the variants of special behaviour of the phantom dark energy : @xmath117 when @xmath115 and @xmath118 , or @xmath245 when @xmath246 in the models ( [ j0 ] ) and ( [ jde ] ) .
the interaction , which causes the flow of energy from the dark matter to the dark energy , always leads to fast decrease of density of dark matter to zero and transition to the negative values , which we consider as the non - physical solution .
[ fig : rcr1 ] , [ fig : de1 ] , [ fig : dm1 ] confirm these conclusions . the type of interaction and the value of interaction parameter between the dark components , as seen from the fig .
[ fig : rcr2 ] , [ fig : de2 ] and [ fig : dm2 ] , have influence on the dynamics of expansion of universe - the hubble parameter and the deceleration parameter , which can be used for identifying the type and the strength of interaction , or at least the upper limits of values of interaction parameter .
the work was supported by the project of ministry of education and science of ukraine `` the dark components and evolutionary stages of formation of high - scale structure of universe , galaxies , stars and supernova remnants '' ( state registration number 0113u003059 ) .
novosyadlyj b. , sergijenko o. scalar field models of dark energy with barotropic equation of state : properties and observational constraints from different datasets
// proceedings of the 10th g. gamow s odessa astronomical conference - summer school astronomy and beyond : cosmomicrophysics , cosmology and gravitation , astrophysics , radio astronomy and astrobiology .
, , , 2010.-p.12 - 21 . | we consider the dark energy model with barotropic equation of state , which interacts with dark matter through gravitation and another force , causing the energy - momentum exchange between them .
both components are described in approximation of ideal fluids , which are parametrized by density and equation of state parameters .
three types of interactions between dark components are considered : the interaction independent from their densities , the one proportional to density of dark energy and the one proportional to density of dark matter .
the equations which describe the expansion dynamics of homogeneous and isotropic universe and evolution of densities of both components for different values of interaction parameter are obtained on the bases of the general covariant conservation equations and einstein s ones . for three kinds of interactions
we show the existence of the range of values of parameters of dark energy for which the densities of dark components and their sum are negative .
we find the conditions of positivity of density of dark energy and dark matter .
the constraints on the value of parameter of interaction are derived .
the dynamics of expansion of the universe with these interactions of dark energy and dark matter is analysed .
* key words : * non - minimally coupled dark energy , dark matter , expansion dynamics of the universe |
optical@xcite and angular resolved photo emission ( arpes ) data@xcite have given a wealth of information on quasiparticle dynamics in the cuprates both in their normal and superconducting state .
the methods are complementary but the exact quantitative relationship between the two is complex .
one important difference is that arpes gives direct information on angular variations around the fermi surface while optics involves an average over all the quasiparticles participating in the absorption . even if these anisotropies are not accounted for ( isotropic system ) there remain additional differences which have their fundamental origin in the fact that arpes measures directly the quasiparticle spectral density @xmath0 at fixed quasiparticle momentum * k * as a function of energy @xmath1 while optics involves the current - current correlation function which depends on the product of two spectral densities@xcite at the same momentum * k * but with frequencies displaced by the photon energy . in this paper
we focus on similarities as well as on essential differences between the information that is derived from these two different probes and particularly on how they are to be compared
. we will be interested in both , the normal and the superconducting state at zero and at finite temperature . in the literature on optics
it is the temperature and frequency dependence of the optical scattering rate @xmath2 which has been particularly emphasized.@xcite more recently the imaginary part@xcite of the memory function related to @xmath2 by kramers - kronig ( kk ) transform has also been used to compare directly with the energy dependence of the real part of the quasiparticle self energy determined by arpes.@xcite of particular interest in such a comparison is the relationship between the position in frequency of peak structures seen in these quantities and how they reflect corresponding peaks in the electron - boson spectral density @xmath3.@xcite for phonons @xmath3 is the well known electron - phonon spectral density while in the cuprates exchange of spin fluctuations@xcite is more appropriate although as yet there exists no consensus@xcite on this issue . here
we will consider several models for @xmath3 within an extended eliashberg formalism . in the superconducting state provision
is made for @xmath4-wave symmetry of the gap . to keep things simple the same form of @xmath3
is used in both gap and renormalization channels but with different magnitudes with the ratio between the two equal to @xmath5 . in sec .
[ sec:2 ] we consider only the normal state and coupling of quasiparticles to a single @xmath6 einstein mode .
we also employ a simplified approximate scattering time formulation of the optical conductivity formula in the normal state .
this allows some analytic results to be established and provides the motivation for introducing a new model scattering rate and its kk - transform determined solely from optical data but which is very close to the self energy itself . while the real part of the self energy has a logarithmic singularity at @xmath7 the imaginary part of the memory function does not .
instead , it shows a peak at @xmath8 . on the other hand ,
our model quantities reproduce well the self energy at @xmath9 and only very small differences arise at finite @xmath10 . in sec .
[ sec:3 ] we provide the more general formalism needed to treat the conductivity accurately as well as to take care of extended @xmath3 spectra and the superconducting state .
results in these cases are presented in sec .
[ sec:4 ] . for the superconducting state
additional complications arise because of the essential energy dependence of the self consistent quasiparticle density of states ( dos ) . in sec .
[ sec:5 ] we present our numerical results for our new model scattering rate based on the complete formula for the optical conductivity and solutions of the eliashberg equations and compare these with the imaginary part of the quasiparticle self energy .
we confirm the previous expectation , based on simplifying assumptions , that in the normal state these two quantities are close in magnitude as well as frequency dependence . in the superconducting state
they are not .
finally , conclusions are found in sec .
[ sec:6 ] .
a generalized drude form with frequency dependent optical scattering rate @xmath11 and optical effective mass ratio @xmath12_{op}$ ] has been used extensively to analyze optical data in a one component approach . the optical conductivity @xmath13 is written in terms of the memory function@xcite @xmath14 with @xmath15_{op}-1\}$ ] the optical mass renormalization parameter .
we have@xcite @xmath16}\ ] ] and @xmath11 and @xmath17 are related by kramers kroning , so that when the optical scattering rate is known , the optical mass renormalization can be obtained from the kk - transform .
the conductivity is a two particle property given by the current - current correlation function .
it is related in a complicated way to the quasiparticle self energy ( a one particle property ) @xmath18 . in the normal state@xcite at zero temperature @xmath19 @xmath20 where @xmath21 is the impurity scattering rate and @xmath22 accounts for the inelastic scattering . in terms of the electron - phonon or electron - spin fluctuation spectral density @xmath23 or @xmath3
respectively [ call it @xmath24 @xmath25 for coupling to a single einstein mode @xmath26 we obtain the simple , well known formula @xmath27 with @xmath28 the heavyside @xmath29-function .
it is clear from that the imaginary part has a finite jump at @xmath7 as well as a logarithmic singularity in its real part which can be used to identify boson structure in both cases .
the self energy can in principle be determined from arpes data which measures the quasiparticle spectral function @xmath30 with the greens function @xmath31^{-1}$ ] , where @xmath32 is the quasiparticle dispersion .
a usefull approximate expression for @xmath11 was obtained directly in second order perturbation theory by p.b .
allen.@xcite it is ( for @xmath9 ) @xmath33 details in the validity of this form , based on numerical evaluation of eqs . and are found in refs . and .
for @xmath34 @xmath35 where the subscript @xmath36 means transport ,
although for simplicity , here we will not make a distinction between quasiparticle and transport spectral density ; in general they are different .
( we have set the impurity term equal to zero for simplicity . ) note that this formula is less favorable for the identification of boson structures than is the imaginary part of the self energy .
while it is zero for @xmath37 and finite for @xmath38 it has no jump at @xmath7 .
instead , it smoothly increases from its zero value which makes it harder to identify the exact position of @xmath7 in @xmath11 . as a function of @xmath1 with the corresponding optical quantity @xmath39_{op}-1\}\equiv\omega% \lambda_{op}(\omega)$ ] .
coupling is to a single einstein oscillator with @xmath40mev and @xmath41mev in eqs . and .
there is a logarithmic singularity in @xmath42 at @xmath43 while @xmath44 only has a small peak at @xmath8 and a logarithmic singularity in the slope at @xmath43 .
, width=340 ] application of the kk - transform to eq .
gives immediately @xmath45 as the optical mass enhancement .
this is to be compared with the real part of the quasiparticle self energy , eq . .
in terms of the usual mass enhancement parameter @xmath46 [ defined as two times the first inverse moment of @xmath24 , @xmath47 , and the limit of @xmath42 as @xmath48 equals @xmath49 .
thus the slope of the self energy gives the value of @xmath46 as does also @xmath50 .
this is seen in fig .
[ fig:1 ] where we compare @xmath42 ( solid line ) and @xmath51 vs @xmath1 ( dashed line ) for parameters @xmath52mev and @xmath53mev .
while these functions agree in the @xmath48 limit they deviate from each other at finite frequencies .
in particular , at @xmath54 the quasiparticle self energy exhibits a logarithmic singularity while the optical mass renormalization @xmath51 has only a logarithmic singularity in its slope .
it can , furthermore , be shown that @xmath51 has its maximum at @xmath55 .
thus , while the boson structure in @xmath18 appears at the boson energy and is singular , the maximum in @xmath56 is instead displaced to @xmath57 and is associated with a rather broad peak in comparison .
we conclude from these considerations that the optical mass enhancement parameter @xmath56 can be used to identify boson structure in the imaginary part of the memory function as done recently by hwang _
et al._@xcite but its signature is much weaker than is the case in the quasiparticle self energy and @xmath7 is displaced by a factor of @xmath58 , a fact that does not appear to have been appreciated before , but is important to realize . in preparation for what will come later when we consider the superconducting state , we note that the self energy and , consequently , the corresponding optical quantity would also apply@xcite in an impurity model with constant quasiparticle scattering rate @xmath59 , but in which the normal state quasiparticle density of states , @xmath60 , has a gap of size @xmath7 at the fermi surface , i.e. : @xmath61 for @xmath62 and @xmath63 but is zero for @xmath64 .
this shows immediately that structure in @xmath60 can have an effect on the self energy which is , in some cases , indistinguishable from boson structure .
we return to this important point later .
how is this picture changed when we consider finite temperatures ? in this case
the formulas determining the conductivity as well as the quasiparticle self energy are more complex .
nevertheless , a simple picture still emerges . the normal state optical conductivity is now given by@xcite @xmath65\right.\nonumber\\ & & \times s^{-1}(t,\omega,\nu ) \biggr\ } , \label{eq:9}\end{aligned}\ ] ] where @xmath66 and the self energy is@xcite @xmath67 , \label{eq:11}\end{aligned}\ ] ] where @xmath68 is the digamma function and @xmath69 is the complex conjugate of @xmath70 .
for a general @xmath71 @xmath72.\nonumber\\ \label{eq:12}\end{aligned}\ ] ] shulga _ et al._@xcite were able to show that to a good approximation @xmath73 .
\label{eq:13}\end{aligned}\ ] ] at zero temperature eqs . and
reduce to the imaginary part of eq . and to eq .
, respectively .
kk - transforms can be used to get from eqs . and the real part of @xmath70 at finite @xmath10 as well as @xmath74 .
results are shown in fig .
[ fig:2 ] ( dotted line ) with its optical ( dashed line ) and model ( solid line ) counterparts as a function of @xmath1 .
the right hand frame is for temperature @xmath75 and the left hand frame is for @xmath76 .
the bottom frames gives corresponding quasiparticle scattering rates on which the results in the two top frames are based .
the parameters are @xmath77mev and @xmath40mev .
, width=340 ] for coupling to a single einstein mode as in fig .
[ fig:1 ] .
three curves are show in each frame .
the dotted line applies to the quasiparticle self energy , the dashed line to the corresponding optical quantity and the solid line to a model yet to be defined .
the top frames give real parts at @xmath76 ( left ) and at @xmath75 ( right ) .
the corresponding scattering rates on which these are based , are shown in the two bottom frames .
first we consider the top frames . in all cases
we see that temperature smooths out the corresponding boson structures , but even for @xmath75 they remain easily identifiable , although , the difference between quasiparticle and optical quantities is no longer as pronounced .
the solid curve in fig . [ fig:2 ] is based on the following observation . at @xmath9 , @xmath78/d\omega$ ]
given by eq .
equals exactly the quasiparticle scattering rate given as twice the imaginary part of the quasiparticle self energy in eq . .
thus , in this particular limit one can get the quasiparticle scattering rate directly from optics by taking the first derivative of @xmath79 .
of course , we have assumed that anisotropies in momentum space can be neglected . in general
, arpes will give information on the * k*-dependence of the quasiparticle self energy while optics is always an average .
neglecting such complications , it is interesting to introduce a model scattering rate @xmath80% /d\omega$ ] defined from optics alone and consider its relationship to the quasiparticle self energy at finite temperatures .
the formula for @xmath81 equivalent to eqs . and is @xmath82 .
\label{eq:14}\end{aligned}\ ] ] in all cases considered in fig .
[ fig:2 ] the model quantity ( solid lines ) deviates from its quasiparticle counter part ( dotted lines ) only very slightly while it deviates much more from the optical case ( dashed lines ) .
it is clear , therefore , that if one wishes to compare optical and quasiparticle quantities directly , it is better to use the kk - transform of @xmath83 than that for @xmath11 itself .
the peaks in the corresponding model mass enhancement will now correspond to the einstein frequency @xmath7 and it will be sharper than in the optical quantity itself which , as we have seen , is also shifted by @xmath58 .
note also , that on comparing top and bottom frame it is the function @xmath11 which is least useful in localizing boson structures as discussed in relation to eq . .
all the other functions have a sharper signature of the einstein frequency @xmath7 .
we next turn to a discussion of how the use of an extended spectrum for @xmath84 modifies our simple results and how the superconducting condensation further modifies boson structures in quasiparticle and optical quantities .
in general , coupling of the electrons to a boson spectrum such as phonons or spin fluctuations is not restricted to a single mode . the corresponding @xmath23 ( phonons ) or @xmath3 ( spin fluctuations )
can extend to @xmath85mev or so and even up to @xmath86mev ( of the order of @xmath87 in the @xmath88 model@xcite ) , respectively . here
we wish to consider such extended spectra and also consider the superconducting state assuming @xmath4-wave gap symmetry since we have the cuprates in mind . the equation for @xmath13@xcite [ eq:16 ] @xmath89\right\rangle_\theta , \label{eq:16a}\ ] ] where @xmath90 denotes the averaging over angle @xmath29 and the function @xmath91
is given by @xmath92\nonumber\\ & & + \frac{1}{e^\ast(\omega;\theta)- e(\omega+\nu;\theta)}\nonumber\\ & & \times\left[1+n^\ast(\omega;\theta)n(\omega+\nu;\theta)\right . \nonumber\\ & & \left .
+ p^\ast(\omega;\theta)p(\omega+\nu;\theta)\right ] , \label{eq:16b}\end{aligned}\ ] ] with @xmath93 and @xmath94 here @xmath95 , @xmath96 , and @xmath97 are the complex conjugates of @xmath98 , @xmath99 , and @xmath100 , respectively . the two fundamental functions @xmath101 and @xmath102 are related the renormalization and gap functions , respectively.@xcite they are for the gap channel [ eq:17 ]
@xmath103\\ \times\left\langle \frac{\tilde{\delta}(i\omega_m;\theta')\cos(2\theta ' ) } { \sqrt{\tilde{\omega}^2(i\omega_m)+\tilde{\delta}^2(i\omega_m ; \theta')}}\right\rangle_{\theta'}\\ + i\pi g\int\limits^\infty_{-\infty}\!dz\,\cos(2\theta ) i^2\chi(z)\left[n(z)+f(z-\nu)\right]\times\\ \times\left\langle \frac{\tilde{\delta}(\nu - z+i0^+;\theta')\cos(2\theta ' ) } { \sqrt{\tilde{\omega}^2(\nu - z+i0^+)-\tilde{\delta}^2(\nu - z+i0^+ ; \theta')}}\right\rangle_{\theta ' } , \label{eq:17a } % \end{eqnarray}\end{gathered}\ ] ] and in the renormalization channel @xmath104\\ \times\left\langle \frac{\tilde{\omega}(i\omega_m ) } { \sqrt{\tilde{\omega}^2(i\omega_m)+\tilde{\delta}^2(i\omega_m ; \theta')}}\right\rangle_{\theta'}\\ + i\pi\int\limits^\infty_{-\infty}\!dz\ , i^2\chi(z)\left[n(z)+f(z-\nu)\right]\\ \times\left\langle \frac{\tilde{\omega}(\nu - z+i0^+ ) } { \sqrt{\tilde{\omega}^2(\nu - z+i0^+)-\tilde{\delta}^2(\nu - z+i0^+ ; \theta')}}\right\rangle_{\theta ' } \label{eq:17b}. % \end{eqnarray}\end{gathered}\ ] ] here @xmath105 the gap is given by @xmath106 or , if the renormalization function @xmath107 is introduced in the usual way as @xmath108 then @xmath109 these equations are a minimum set and go beyond a bcs approach .
they include inelastic scattering known to be strong in the cuprate superconductors . in eq .
@xmath5 modifies the electron - boson spectral density @xmath3 from its value in the renormalized channel . in general the shape of @xmath3
could also be different but we have not included this possible complication here . in what follows , we will use for @xmath3 a form suggested by millis _
et al._@xcite ( mmp ) given by @xmath110 with @xmath111 a spin fluctuation frequency which can be used in fig . . the mmp model with @xmath112mev , @xmath113 augmented by a @xmath114-function at @xmath115mev with @xmath116 for a total of 1.5 ( dashed curve ) .
the model spectrum with @xmath117 obtained from a fit to the optical conductivity as described in ref .
( dotted curve ) .
coupling to an optical resonance at @xmath118mev is included as a peak . a pure mmp spectrum with @xmath112mev and @xmath119 ( solid curve ) .
bottom frame : the superconducting density of states @xmath120 at @xmath121k for the three spectra shown in the top frame .
, width=340 ] fitted to optical data and to emphasize structure we also will add in one case a @xmath114-function at some specific frequency .
such an mmp form for @xmath112mev is shown in the top frame of fig .
[ fig:3 ] ( solid line ) .
it has a @xmath119 .
also shown is the same mmp form now with an added @xmath114-function peak at @xmath115mev ( dashed line ) .
@xmath46 is again equal to 1.5 but now with only 0.554 in the mmp form and the rest in the @xmath114-function . in the superconducting state
the mmp form is modified because of the growth of an optical resonance@xcite at @xmath118mev which is the same frequency at which a spin one resonance is also observed in spin polarized inelastic neutron scattering@xcite for optimally doped ybco .
( in other cuprates the position of the observed resonance varies somewhat.@xcite ) this optical resonance grows as the temperature is lowered with the area under the resonance scaling as the superfluid density@xcite to a reasonable approximation .
such a spectrum has been derived from a fit to @xmath121k optical data on a yba@xmath122cu@xmath123o@xmath124 ( ybco@xmath124 ) twinned single crystal.@xcite this spectrum was also used later on to fit optical data for untwinned single crystals reported by c.c .
homes _ et al._@xcite and also to calculate the microwave conductivity of the ybco@xmath125 single crystals.@xcite this spectrum is shown as the dotted curve in the top frame of fig .
[ fig:3 ] .
it has a @xmath126 .
we will take this @xmath3 spectrum as representative of the oxides .
note that coupling to a collective mode at @xmath127 is also seen in the arpes data of campuzano _
et al._@xcite also note that in the above formulation we have neglected possible momentum space anisotropies .
recent arpes data by kaminski _
et al._@xcite justify this assumption . the bottom frame of fig .
[ fig:3 ] shows the quasiparticle density of states @xmath120 in the superconducting state for @xmath121k for the three spectra presented in the top frame
. this would be the classic way to see boson structure in conventional superconductors through tunneling which , so far , has been less successful in the high @xmath128 oxides .
this is the reason why optics and arpes became such important tools in studying the quasiparticle vs @xmath1 and optical effective mass renormalization @xmath51 for the three model electron - boson spectral densities @xmath3 shown in the top frame of fig . .
the top frame is for an mmp form with @xmath112mev and @xmath129 augmented by a @xmath114-function at @xmath115mev with @xmath116 , the middle is obtained with a model @xmath3 obtained from consideration of optical properties of ybco@xmath124 with @xmath117 ( dotted curve in the top frame of fig . ) , and the bottom frame is for the pure mmp with @xmath112mev and @xmath119 . in all cases the curves come in pairs @xmath42 in normal ( black dotted curve ) and superconducting state ( black solid curve ) and @xmath51 in normal ( black dashed curve ) and superconducting state ( black dash - dotted curve ) . in the middle frame
we also include as ( gray solid line ) the difference between superconducting and normal state of @xmath51 and as gray dotted curve the difference between superconducting and normal state @xmath42 . in the bottom frame
similar difference curves are shown but with two different values of @xmath10 in the normal state as labeled .
, width=340 ] properties .
note that the solid curve for @xmath120 which is based on an mmp form with @xmath112mev shows no distinct sharp structure .
thus , a relatively smooth spectrum extending over a large energy scale produces in the quasiparticle density of states only very small , gradual modulations which would be hard but not impossible to detect .
in fig . [ fig:4 ] we show our numerical results for the real part of the quasiparticle self energy @xmath42 ( only for @xmath34 and made positive ) as a function of frequency @xmath1 based on numerical evaluation of eqs . for the renormalized matsubara frequencies .
the black dotted curves apply to the normal and the black solid lines to the superconducting state .
both are at temperature @xmath121k . the spectrum @xmath3 used in the top frame of fig .
[ fig:4 ] is the mmp form of eq . with @xmath112mev
to which we have added a delta function contribution also at @xmath115mev .
this is shown as the dashed line in the top frame of fig .
[ fig:3 ] .
the resulting boson structure in the normal state self energy is sharp and falls exactly at @xmath115mev .
it is the presence of the @xmath114-function in @xmath3 which makes the peak so prominent .
the corresponding peak in the superconducting state ( solid black curve ) has shifted to higher frequency and falls slightly below @xmath130mev where @xmath131 is the gap amplitude . for an @xmath132-wave superconductor
the shift in the boson structure would fall exactly at @xmath133 but for @xmath4-wave it falls below this value because the gap is distributed in value and we are seeing the result of a distribution of shifts from 0 to @xmath131 .
the peak has also broadened and the weight under it appears to have increased . in this sense , the opening up of the superconducting gap in the quasiparticle density of states has not only shifted the boson structure but has also , in a sense , enhanced it .
this same statement applies even more strikingly to the corresponding optical quantities .
for the imaginary part of the memory function , optical mass renormalization @xmath134% -1\}$ ] , black dashed curve for the normal state and black dash - dotted curve for the superconducting state , the boson structure is much greater in the dash - dotted curve . comparing the normal state memory function ( black dashed curve ) with the self energy ( black dotted curve )
shows that the structure in the memory function is indeed at @xmath57 rather than at @xmath7 although there is a background to our boson spectrum besides the @xmath114-function .
this square root of two shift becomes critical in serious comparison arpes and optical results and has not been appreciated in the past.@xcite note that for the superconducting case the structure in the memory function is greatly enhanced over its normal state value and that it is also shifted upwards in energy as for the self energy .
its exact position depends on the details of the @xmath4-wave gap structure and a complete eliashberg calculation is required to determine it .
the size of the peak in the underlying electron - boson spectral density @xmath3 is not related simply to the size of the structure seen in the self energy or memory function . to produce sharp , easily identifiable , structures in arpes or optical self energies ,
it is necessary to have correspondingly sharp structures in @xmath3 .
this fact is well illustrated in the bottom frame of fig .
[ fig:4 ] where we show results based on a simple mmp model , eq . ,
spectral density @xmath3 ( solid curve in the top frame of fig .
[ fig:3 ] ) . direct comparison of the curves in the bottom frame of fig .
[ fig:4 ] with the corresponding curves in the top frame shows that now the boson structures are much reduced .
although there is a broad peak at @xmath118mev in @xmath3 for the pure mmp spectrum , this does not translate into a discernable peak in any of our results in the bottom frame of fig .
[ fig:4 ] . from this comparison
we conclude that a broad peak in @xmath3 is difficult to detect as a clear signature in either the real part of the quasiparticle self energy @xmath42 or in the optical mass renormalization @xmath51 in the normal as well as in the superconducting state .
a detailed eliashberg analysis is needed to extract @xmath3 from such data .
a method for achieving this has been described in refs . and .
an appropriate derivative of @xmath11 , namely @xmath135/% d\omega^2 $ ] which is closely related to @xmath3 , see eq .
, is used to get a first modification to the mmp form , eq .
, that fits the optical data better .
the procedure can be continued until a reasonable fit with the optical data is obtained from the model spectral density .
however , no high accuracy fit has so far been achieved . in the absence of such an analysis experimentalists have tried to get information on structure in @xmath3 through comparison between normal and superconducting state data . to this end
we show in the bottom frame of fig .
[ fig:4 ] results for @xmath51 in the normal state at @xmath136k ( black dash - double dotted curve ) .
a subtraction of this data from the dash - dotted curve in the superconducting state gives the difference ( gray dash - dotted curve ) .
on comparing this curve with the input @xmath3 ( solid line , top frame of fig .
[ fig:3 ] ) we see little detail correspondence .
the peak in the difference curve is not at @xmath118mev and the mmp tails at larger values of @xmath1 are not picked up .
this holds as well if the normal state data at @xmath121k ( black dashed curve ) , albeit not accessible to experiment , is used in the subtraction ( gray solid curve ) or if @xmath42 is considered ( gray dotted curve ) .
this subtraction procedure does not give a reliable way to relate boson structure in experimentally measured quantities directly to boson structure in the electron - boson spectral density and is not recommended as an analysis procedure .
this is also seen in the middle frame of fig .
[ fig:4 ] which applies to the case of the spectrum fit to optical data on ybco@xmath124 twinned single crystals.@xcite it is shown as the dotted curve in the top frame of fig . [ fig:3 ] .
it has a @xmath117 and corresponds to an mmp form modified with an optical resonance peak at @xmath137mev and there is zero weight at small @xmath1 .
this spectrum is intermediate in
sharpness of boson structure " between the @xmath114-function and the pure mmp case .
we see that this reflected itself in how structured @xmath42 and @xmath51 are .
note in particular that the large peaks in the black dash - dotted curve for @xmath51 is now shifted to higher frequencies in the middle frame when compared to the top frame although the optical resonance peak is at the same frequency as is the @xmath114-function in fig .
[ fig:3 ] .
( this is because the gap @xmath131 is bigger now . )
this demonstrates once more that no exact correspondence between size and position of the structure in @xmath51 and @xmath3 is possible without a detailed analysis , particularly for extended spectra .
also these functions strongly reflect the energy dependence of the underlying quasiparticle density of states that exists in the superconducting state ( dotted curve , bottom frame of fig .
[ fig:3 ] ) and it is not possible to , so to speak , subtract out the effects of these modulations from the data .
formally the subtraction procedure described above between superconducting and normal state quantities when applied to the middle frame of fig .
[ fig:4 ] again fails to give reliable information on the shape of the boson spectrum . in the top frame of fig .
[ fig:5 ] we show results for the vs @xmath1 for the three quasiparticle - boson spectral densities used in fig . . in each of the three frames the solid black curve is for the superconducting state at @xmath121k , the black dotted curve for the normal state at the same temperature , and the black dashed curve for the normal state at @xmath136k .
the gray curves are the difference curves , solid curve for @xmath121k and dashed curve for @xmath136k . , width=340 ] optical scattering rate @xmath11 based on the model @xmath3 which consists of an mmp form plus a @xmath114-function ( dashed curve in the top frame of fig .
[ fig:3 ] ) at @xmath115mev .
the solid black curve is in the superconducting state at @xmath121k while the black dotted curve is in the normal state at the same temperature .
while the underlying @xmath3 is the same in both cases yet the amount of structure seen in the solid curve is much more pronounced than it is for the dotted curve .
this is due entirely to the energy dependence of the quasiparticle density of states ( dashed line , bottom frame of fig .
[ fig:3 ] ) which arises in a @xmath4-wave superconductor .
on comparing the solid and the dotted curve we note a shift upward in the initial sharp rise in the scattering rate as we go to the superconducting case .
this corresponds to the opening of the gap .
the black dashed curve is in the normal state at @xmath136k .
temperature smears the sharp rise at small @xmath1 . only the @xmath136k data
is available to experiment . if , as we did in fig .
[ fig:4 ] , we take the difference between superconducting and normal state optical scattering rate we get the gray solid line for @xmath121k and the gray dashed curve for @xmath136k .
the minimum in each of these curves is around @xmath138mev which is @xmath7 shifted by approximately @xmath131 .
it is obviously hard to get reliable information on @xmath7 from such data .
the middle and bottom frames of fig .
[ fig:5 ] are for less structured @xmath3 and show a progressive decrease in the corresponding structures in @xmath11 .
we turn next to the exact correspondence between @xmath11 and the quasiparticle scattering rates given as twice the imaginary part of the self energy . in the previous section we have already examined this relation but based our discussion on approximations for the conductivity formula .
next we use the more exact eq . and solutions of the full eliashberg equations .
we return to the model function @xmath83 defined in sec .
[ sec:3 ] .
it can be constructed from the numerical data shown in fig .
[ fig:5 ] for @xmath11 defined as @xmath139 and @xmath140 we note in passing , that for small @xmath1 the optical effective mass formula is dominated by the imaginary part of the optical conductivity @xmath141 in the superconducting state because it diverges as @xmath142 for small @xmath1 with coefficients proportional to the inverse square of the penetration depth .
this is not related to a quasiparticle effective mass , but is rather a property of the condensate itself .
nevertheless , this is what has been done traditionally and we need to follow this procedure here to make contact with the literature .
an alternate approach would be to subtract out of the imaginary part of the optical conductivity the condensate contribution before forming the ratios indicated in eqs . and .
in this alternate approach the resulting @xmath143 and @xmath144 would refer directly only to the normal fluid part of the optical conductivity and so would be more closely related to quasiparticle properties .
but this is not what is done in the literature . for comparison with the quasiparticle self energy it is useful to use @xmath81 given by @xmath145\ ] ] than @xmath2 .
its imaginary part is given by the kk - transform @xmath146 in fig .
[ fig:6 ] we show results for @xmath83 ( dashed lines ) , quasiparticle , @xmath147 ( solid lines ) , and model @xmath83 ( dotted lines ) scattering rates vs @xmath1 for a temperature of @xmath121k . for easier comparison @xmath148 and @xmath149 are shown .
the two top frames compare our results for an mmp form with @xmath112mev in the normal ( left hand frame ) and the superconducting state ( right hand frame ) .
the two bottom frames give the results for the model @xmath3 determined from optics for ybco@xmath124 as shown in the top frame of fig .
( dotted curve ) .
, width=340 ] based on the numerical results of fig .
[ fig:5 ] and compare directly with the imaginary part of the self energy @xmath150 obtained directly from eliashberg calculations , namely @xmath151 of eq . .
by definition of the self energy @xmath152 . in the two top frames of fig .
[ fig:6 ] we show results for the mmp form ( solid line in the top frame of fig . [ fig:3 ] ) and in the two bottom frames the ybco@xmath124 form ( dotted curve in the top frame of fig . [ fig:3 ] ) for @xmath3 is used . in each case
we present two frames , the left hand frame is for the normal state and the other frame for the superconducting state .
all results are at temperature @xmath121k . for ease of comparison with @xmath11 ( dashed lines )
it is @xmath148 ( solid lines ) and @xmath149 ( dotted lines ) that is shown .
even when a factor of one half is included to make magnitude more comparable , the optical scattering rates differ significantly from the quasiparticle scattering rates .
on the other hand , in the normal state the model scattering rate agrees almost perfectly with the quasiparticle scattering rate ( top left hand frame ) .
it is clear that @xmath83 should be used in comparison with arpes data and not @xmath11 .
the close correspondence between model and optical rates is , however , lost when the superconducting state is considered ( top right hand frame ) .
the solid curve is significantly different from the dotted curve . in particular , @xmath83 is smaller at small @xmath1 and rises faster around @xmath153mev after which it stays significantly above its quasiparticle counterpart up to almost @xmath154mev where the two curves cross again .
these differences have their origin in the energy dependence of the superconducting quasiparticle density of states ( solid line , bottom frame of fig . [ fig:3 ] ) and have nothing to do with the boson structure . for a comparison of @xmath11 ( optics ) and @xmath147 ( arpes ) based completely on experiment ,
the reader is refered to kaminski _
et al._@xcite when the underlying @xmath3 is more structured , as is the case for the ybco@xmath124 spectrum ( dotted curve in fig .
[ fig:3 ] ) , the correspondence between @xmath83 and @xmath147 is not as good as is seen in the two bottom frames of fig .
[ fig:6 ] .
there is a small additional oscillation in the model case , not present in the self energy . nevertheless , the agreement between them is still much closer than is the case for the optical rate ( dashed curve ) .
note , however , that for the superconducting state the disagreement between @xmath147 ( solid curve ) and @xmath83 ( dotted curve ) is now much greater and , in particular , a large peak is seen in the model curve which is not there in the dotted curve .
this peak can be related to the broad shoulder in the quasiparticle density of states which follows the logarithmic singularity ( dotted line , bottom frame of fig . [ fig:3 ] ) .
we have analyzed the relationship between the boson structure seen in the quasiparticle self energy ( be it real or imaginary part ) measured in arpes experiments and the corresponding structure seen in the optical memory function determined in the infrared conductivity measurements .
starting first with the real part of the self energy a @xmath114-function peak at @xmath43 in the quasiparticle - boson spectral density @xmath3 shows up as a logarithmic - like peak at @xmath43 in the normal state . in the superconducting state a gap develops in the quasiparticle density of states and
this introduces further structures in quasiparticle quantities .
this effectively shifts to higher energy the normal state boson structure , broadens it , and can make it appear more prominent . for an @xmath132-wave gap
the shift would be @xmath131 ( gap amplitude ) but for @xmath4-wave it is somewhat less because a distribution of gap values is involved .
by contrast , for the memory function the boson structure in the optical mass renormalization is at @xmath57 in the normal state rather than at @xmath7 and it is much less prominent . only a very small peak results which greatly reduces the value of such measurements for determining boson structure as was attempted recently in ref . .
again , the boson structure shifts in the superconducting state and can also appear more prominent as a result of the additional modulation brought about by the opening of a gap . these modifications , however , have nothing to do with the structures in @xmath3 . the two effects are not additive and require a full non linear eliashberg analysis to disentangle in detail as we have provided here . when extended rather than @xmath114-function spectra are used
, the situation is even more complex .
for example , a rather broad peak in @xmath3 , as in the mmp form for spin fluctuations produces no peak at all , even in the superconducting state .
so far we have described only the real part of the self energy and equivalent optical quantity .
much the same can be said about scattering rates . for a @xmath114-function @xmath3 ,
the quasiparticle scattering rate jumps from zero to a finite value at @xmath43 and remains unchanged after that . on the other hand ,
the optical scattering rate starts from zero at @xmath43 and increases gradually towards the same finite value which it only attains at very high @xmath1 .
thus , the signature of a @xmath114-function is not singular as it is for the quasiparticle case .
we have also found that in the normal state a new model scattering rate can be introduced which we denote @xmath81 and define as @xmath155/d\omega$ ] .
it is completely determined from optical data and has the advantage that it follows much more closely the @xmath1 dependence of the quasiparticle scattering rate than does @xmath2 itself .
its kk - transform is very close to the real part of the quasiparticle self energy .
we propose that it is this quantity that should be used in comparison between arpes and optical experimental data
. the comparison will be close in isotropic systems and in the anisotropic case , the model quantity gives an average over all directions in the brillouin zone . in the superconducting state
no such simple comparison between arpes and optics is possible .
this needs to be kept in mind when analyzing experimental data , particularly in studies aimed at deriving boson structure from such data .
this fact does not seem to have always been appreciated in analysis of experimental data , where features , at least partially associated with the superconducting quasiparticle density of states , has been assigned to boson structure .
the model spectral density @xmath3 consisting of an mmp form with superimposed the @xmath114-peak was used to demonstrate that optics as well as arpes just pick up the sharp structures and do not give direct information on a possible coupling to a smooth background .
nevertheless , it may well be that it is precisely this coupling of the quasiparticles to such a background which is primarily responsible for superconductivity .
research supported by the natural sciences and engineering research council of canada ( nserc ) and by the canadian institute for advanced research ( ciar ) .
tanner and t. timusk , in _ physical properties of high temperature superconductors iii _ , edited by d.m .
ginsberg ( world scientific , singapore , 1992 ) , p. 363 .
puchkov , d.n .
basov , and t. timusk , j. phys . : condens .
matter * 8 * , 10049 ( 1996 ) .
campuzano , m.r .
norman , and m. randeria , in _ the physics of superconductivity vol .
ii : superconductivity in nanostructures , high - tc and novel superconductors _ , edited by k .- h . bennemann and j.b .
ketterson ( springer - verlag , berlin , 2003 ) , vol .
2 , p. 167 .
a. damascelli , zahid hussain , and z .- x .
shen , rev .
phys * 75 * , 473 ( 2003 ) .
norman and c. pepin , rep .
phys * 66 * , 1547 ( 2003 ) .
f. marsiglio and j.p .
carbotte , in _ the physics of superconductivity vol .
i : conventional and high - tc superconductors _ , edited by k .- h .
bennemann and j.b .
ketterson ( springer - verlag , berlin , 2003 ) p. 233 .
e. schachinger and j.p .
carbotte , in : _ models and methods of high - tc superconductivity : some frontal aspects _ , edited by j.k .
srivastava and s.m .
rao , ( nova science , hauppauge , ny , 2003 ) , vol .
ii , pp . 73
j. hwang , t. timusk , and g.d .
gu , nature ( london ) * 427 * , 714 ( 2004 ) .
quijada , d.b .
tanner , r.j .
kelley , m. onellion , h. berger , and g. margaritondo , , 14917 ( 1999 ) .
allen and j.c .
mikkelsen , , 2952 ( 1977 ) .
a. lanzara , p.v .
bogdanov , x.j .
zhou , s.a .
kellar , d.l .
feng , e.d .
lu , t. yoshida , h. eisaki , a. fujimori , k. kishio , j .- i .
shinoyama , t. modall , s. uchida , z. hussain , and z .- x .
shen , nature ( london ) * 412 * , 510 ( 2001 ) .
zhou , t. yoshida , a. lanzara , p.v .
bogdanov , s.a .
kellar , k.m .
shen , w.l .
yang , f. ronning , t. sasagava , t. kakeshita , t. noda , h. eisaki , s. uchida , c.t .
lin , f. zhou , j.w .
xiong , w.x .
ti , z.x .
zhao , a. fujimori , z. hussain , and z .- x .
shen , nature ( london ) * 423 * , 398 ( 2003 ) ; e. cappeluti and l. pietronero , cond - mat/0309080 ( unpublished ) . f. marsiglio , t. startseva , and j.p .
carbotte , phys .
lett . a * 245 * , 172 ( 1998 ) .
s. verga , a. knigavko , and f. marsiglio , , 054503 ( 2003 ) .
e. schachinger , j.j .
tu , and j.p .
carbotte , , 214508 ( 2003 ) .
e. schachinger , j.p .
carbotte , and d.n .
basov , europhys . lett . * 54 * , 380 ( 2001 ) .
carbotte , e. schachinger , and d.n .
basov , nature ( london ) * 401 * , 354 ( 1999 ) .
e. schachinger and j.p .
carbotte , , 9054 ( 2000 ) .
chubukov , d. pines , and j. schmalian , in _ the physics of superconductivity vol .
i : conventional and high - tc superconductors _ , edited by k .- h .
bennemann and j.b .
ketterson ( springer - verlag , berlin , 2003 ) vol . 1 ,
carbotte , rev .
* 62 * , 1027 ( 1990 ) .
mei - rong li and j.p .
carbotte , , 155114 ( 2002 ) .
b. mitrovi and j.p .
carbotte , can .
j. phys .
* 61 * , 758 ( 1983 ) ; _ ibid .
_ * 61 * , 784 ( 1983 ) ; _ ibid . _ * 61 * , 872 ( 1983 ) .
b. mitrovi , ph .
d. thesis , mcmaster university 1981 ( unpublished ) .
f. marsiglio and j.p .
carbotte , aust . j. phys .
* 50 * , 975 ( 1997 ) ; _ ibid . _ * 50 * , 1011 ( 1997 ) .
g. grimvall , _ the electron - phonon interaction in metals _ , ( north - holland , new york , 1981 ) .
allen , , 305 ( 1971 ) .
shulga , o.v .
dolgov , and e.g. maksimov , physica c * 178 * , 266 ( 1991 ) . w. lee , d. rainer , and w. zimmermann , physica c * 159 * , 535 ( 1984 ) . s. sorella , g.b .
martins , f. becca , c. gazza , l. capriotti , a. parola , and e. dagotto , , 117002 ( 2002 ) .
carbotte and e. schachinger , ( in print ) .
a. millis , a.v .
balatsky , and d. pines , , 167 ( 1990 ) .
bourges , y. sidis , h.f .
fong , b. keimer , l.p .
regnault , j. bossy , a.s .
ivanov , d.l .
milius , and i.a .
aksay , aip conf .
proc . * 483 * , 207 ( 1999 ) .
fong , p. bourges , l.p .
regnault , a. ivanov , g.d .
gu , n. koshizuka , b. keimer , nature ( london ) * 398 * , 588 ( 1999 ) .
h. he , p. bourges , y. sidis , c. ulrich , l.p .
regnault , s. pailhs , n.s .
berzigiarova , n.n .
kolesnikov , b. keimer , science * 295 * , 1045 ( 2002 ) .
e. schachinger and j.p .
carbotte , , 094501 ( 2001 ) .
homes , d.a .
bonn , r. liang , w.n .
hardy , d.n .
basov , t. timusk , and b.p .
clayman , , 9782 ( 1999 ) .
e. schachinger and j.p .
carbotte , , 064514 ( 2002 ) .
a. hosseini , r. harris , s. kamal , p. dosanjh , j. preston , r. liang , w.n .
hardy , and d.a .
bonn , , 1349 ( 1999 ) .
campuzano , h. ding , m.r .
norman , h.m .
fretwell , m. randeria , a. kaminski , j. mesot , t. takeuchi , t. sato , t. yokoya , t. takahashi , t. mochiku , k. kadowaki , p. guptasarma , d.g .
hinks , z. konstantinovic , z.z .
li and h. raffy , , 3709 ( 1999 ) .
a. kaminski , h.m .
fretwell , m.r .
norman , m. randeria , s. rosenkranz , j.c .
campuzano , j. mesot , t. sato , t. takahashi , t. terashima , m. takano , k. kadowaki , z.z .
li , and h. raffy , cond - mat/0404385 ( unpublished ) .
a. kaminski , j. mesot , h. fretwell , j.c .
campuzano , m.r .
norman , m. randeria , h. ding , t. sato , t. takahashi , t. mochiku , k. kadowaki , and h. hoechst , , 1788 ( 2000 ) . | an extended drude form is often used to analyze optical data in terms of an optical scattering rate and renormalized mass corresponding , respectively , to the real and imaginary part of the memory function .
we study the relationship between memory function and quasiparticle self energy for an isotropic system .
we emphasize particularly boson signatures .
we find it useful to introduce a new auxiliary model scattering rate and its kramers - kronig transform determined solely from optics which are much closer to the self energy than is the memory function itself in the normal state . in the superconducting state
the simplification fails because the quasiparticle density of states acquires an essential energy dependence . |
imaging systems are an essential component of quantum gas experiments .
there is an increasing need for high resolution imaging ( @xmath2 ) to extract fine detail , such as topological defects in a bose
einstein condensate @xcite , or a single trapped ion @xcite .
one difficulty faced is that the object of interest is held within a vacuum chamber and must be viewed through a thick glass window .
even a perfect optical flat introduces significant spherical aberration @xcite in a large numerical aperture ( na ) system .
one approach is to house an infinity - corrected objective within the vacuum chamber @xcite .
collimated rays from such objectives are not significantly aberrated because they pass through an optically flat window at near - normal incidence . with this approach
the objective can have a short working distance , allowing for large nas and permitting the use of commercial microscope objectives @xcite .
however _ in vacuo _
optics must be vacuum - compatible and bakeable , and are fixed in positions that limit optical access to the experiment .
optics housed external to the vacuum chamber are particularly desirable , for example , in a bose - einstein condensate apparatus that shares a common source chamber and science chamber @xcite , or in the study of liquid helium using optical cryostats @xcite .
the _ ex vacuo _ objective must have a long working distance if the object is located far ( @xmath3 ) within a vacuum chamber and must also correct for the spherical aberrations introduced by the vacuum window . fortunately in quantum gas experiments the illumination is typically monochromatic with a small field of view ( @xmath4 ) , nevertheless long working distance objectives that correct for large spherical aberrations are not available commercially and remain a design challenge .
published designs that meet these criteria @xcite require manufacturing at least one custom singlet , which is a slow and costly process . because of this difficulty , many experiments employ a single lens , or a single light - gathering objective lens followed by a single image forming lens @xcite .
such imaging systems are not corrected for aberrations and consequently have low resolution .
this article provides an alternative solution to the problem of high resolution imaging with _ ex vacuo _ optics : an objective consisting of entirely catalog singlets that achieves diffraction - limited imaging with an na of 0.36 .
the objective achieves a resolution of 1.3@xmath0 with a field of view ( fov ) of 360@xmath0 for a 5 mm thick window using 780 nm illumination .
the objective is suitable for diffraction - limited imaging of all the alkalis , retaining nas above 0.32 by changing only the aperture diameter and accounting for the chromatic focal shift .
the objective remains well - corrected into the near ultraviolet ; we predict sub - micrometer resolution when imaging yb@xmath5 in ion trapping experiments .
a useful diffraction - limited fov can be retained for glass windows up to 15 mm thick by changing the singlet spacings .
the required spacings can be found through re - optimization of the design in ray - tracing software .
l [email protected] [email protected] l + sur- & & & mat- + face & & & erial + 1 & & 4&00 & bk7 + 2 & 51&50 & 10&50 & air + 3 & 89&47 & 10&36 & bk7 + 4 & -89&47 & 0&76 & air + 5 & 47&90 & 7&30 & bk7 + 6 & 119&30 & 1&12 & air + 7 & 30&30 & 9&70 & bk7 + 8 & 65&80 & 7&30 & air + 9 & & 14&94 & air + 10 & & 5&00 & silica + 11 & & 15&00 & vacuum + l
l l l l l + & @xmath6 & ap & res & na & fov + & ( nm ) & ( mm ) & ( @xmath0 ) & & ( @xmath0 ) + li & 671 & 24.0 & 1.20 & 0.34 & 400 + na & 589 & 22.0 & 1.13 & 0.32 & 560 + k & 767 & 25.0 & 1.33 & 0.35 & 400 + rb & 780 & 25.4 & 1.31 & 0.36 & 360 + cs & 852 & 26.0 & 1.44 & 0.36 & 340 + fr & 718 & 24.0 & 1.29 & 0.34 & 470 + yb@xmath5 & 370 & 16.0 & 0.90 & 0.25 & 520 + + + + + [ table ] our bose
einstein condensate apparatus is similar to ref .
@xcite , but with an optically flat glass cell as our science chamber to maximize optical access .
the magnetic coils surrounding the glass cell accommodate 50.8 mm diameter lenses , while the distance between the 5 mm thick glass window and the condensate demands a working distance greater than 25 mm .
we require a diffraction - limited fov of at least 300@xmath0 for imaging condensates _ in situ_. the objective was designed using zemax 12 se @xcite , a ray - tracing program which varies parameters to optimize the lens performance .
the design of alt @xcite was chosen as a starting point because it achieved a diffraction - limited performance using bk7 glass a material widely available in standard lens catalogs with only four singlets .
our design is distinguished from alt s by a larger na and the use of only catalog lenses . after scaling the design to 50.8 mm diameter singlets , we optimized both on- and 200@xmath0 off - axis for maximal na with the required working distance .
the optimization routine chosen in zemax minimized the squared sum of all wavefront aberrations up to 7th order .
the default merit function was modified to seek a diffraction - limited modulation transfer function ( mtf ) .
this ensures high image contrast for objects with spatial frequencies up to the resolution determined by the na @xcite .
initially all singlet thicknesses , spacings and curvatures were varied until the diffraction limit was reached for the desired working distance and na .
we then replaced the singlet most similar to a catalog lens , fixing its curvature and thickness to the catalog values , and repeated the optimization routine for the remaining lenses .
this process was repeated until all singlets were catalog lenses . during this process
we occasionally made dramatic changes to lens curvatures and spacings to ensure that the optimization routine did not stagnate in a local minimum @xcite .
the final design comprises four catalog singlets : two positive meniscus lenses to enable a high na with minimal addition of spherical aberration , a bi - convex lens , and a rear plano - concave lens to cancel aberrations introduced by the other lenses and by the window @xcite .
this design is shown in figure [ lens_design ] , with the singlet curvatures and spacings listed in table [ table](a ) .
our objective has an na of 0.36 yielding a resolution of 1.3@xmath0 at the design wavelength of 780 nm .
it has a long working distance of 35 mm measured from the front aperture ( surface 9 , fig .
[ lens_design ] ) and an effective focal length of 47 mm .
the objective is infinity - corrected , allowing the image - forming optics to be chosen separately to suit various applications .
the aberration produced by the 5 mm thick window is corrected over a diffraction - limited fov of 360@xmath0 .
here we consider the fov to be ` diffraction - limited ' where the strehl ratio is greater than 0.8 @xcite .
the calculated mtf , both on - axis and at the edge of the fov , is compared to the diffraction - limited mtf in figure [ mtf ] .
the similarities between the curves indicate that the residual aberrations will not significantly degrade image contrast across the 360@xmath0 diameter fov . our objective is applicable to imaging all the alkalis and yb@xmath5 at commonly used wavelengths .
retaining a diffraction - limited fov of a few hundred micrometers only requires changing the aperture diameter ( table ( [ table]b ) ) .
the objective is not corrected for chromatic aberration , so the focal shift must be accounted for when using a different wavelength .
while a full tolerance analysis was not performed , the design appears robust to changes in singlet spacing : rounding the spacings ( thickness of surfaces 2 , 4 and 6 in table [ table ] ) to the nearest 0.5 mm had negligible effect on the predicted performance .
this allowed the housing assembly to be the simple design shown in figure [ lens_design ] .
the singlets are held on - axis inside a smooth aluminium tube and are separated by aluminium spacer rings .
these rings contact each singlet 2 mm from its outer edge , avoiding chamfers which can result in large spacing errors . the tube screws into a mounting bracket to hold the singlets in place and an o - ring moderates the pressure applied to the singlets .
aluminium components are anodized black to reduce reflections .
we measured the mtf , fov and point spread function of the objective using a usaf 1951 resolution target ( edmund optics 58 - 198 ) and a 1@xmath0 pinhole ( edmund optics 39 - 878 ) , both illuminated by collimated 780 nm laser light .
a 5 mm fused silica optical flat was used to mimic a vacuum window .
an @xmath7 mm achromat placed immediately after the objective formed images of the test objects at a magnification of -21.4 on an scmos camera ( andor neo ) .
this camera has a pixel size of 6.50@xmath0 , corresponding to an effective pixel size of 304 nm in the object plane .
including the achromat in our simulation further compensated aberrations , permitting an increase of both the fov to 400@xmath0 and the aperture diameter to 26 mm .
accordingly , this aperture diameter was used in the constructed objective .
the image of the resolution target ( fig . [ target ] ) is undistorted beyond the diffraction - limited fov .
the line pairs in element 6 of group 8 are clearly resolved , corresponding to a resolution of @xmath8 .
the predicted resolution of 1.3@xmath0 would also resolve the elements in group 9 , but the coherent illumination produced diffraction fringes which degraded the image quality .
the point spread function the image of a point source provides another measure of the lens performance .
the measured image of a 1@xmath0 pinhole is similar to the convolution of a 1@xmath0 top - hat function with the simulated point spread function ( fig .
[ pinhole](d ) , inset ) .
the azimuthal averages of these images about the pinhole center ( fig .
[ pinhole ] ) are in agreement , affirming the objective performed as expected .
asymmetry in the on - axis pinhole image is likely due to tilt of the optical flat or objective relative to the camera @xcite .
the pinhole was also used to measure the diffraction - limited fov and to locate the position of the optic axis relative to the camera chip .
we raster - scanned the pinhole across the object plane using two motorized translation stages under automated control @xcite , taking 3600 images on a 10@xmath0 grid in one hour . at each point on the grid we fitted a two dimensional gaussian to the pinhole image ; the geometric mean of the fitted rms diameters ( spot size ) are shown in figure [ pinhole](a ) .
row- and column - wise averages of the spot size revealed distinct minima , which we took to be the position of the optic axis ( cross in fig .
[ pinhole](a ) ) .
the diffraction - limited fov was measured to be 350@xmath0 in diameter , commensurate with a circle inside which the spot size is less than @xmath9 times the on - axis spot size .
the measured fov is smaller than the simulated 400@xmath0 , owing to the asymmetry in spot size about the optic axis .
this asymmetry was likely caused by tilt in the translation stages that moved the pinhole through either side of focus across the object plane .
figures [ pinhole](b ) and [ pinhole](c ) show the diameters of the pinhole image along the column and row that intersect the optic axis respectively .
in addition to the spot size , the constituent @xmath10 and @xmath11 gaussian rms diameters are shown . the variation between the @xmath10 and @xmath11 diameters reveals small astigmatism , suggesting tilt of the optical flat and objective relative to the camera .
the measured spot size compares favorably to simulation both on- and off - axis .
the combination of catalog singlets presented in this paper form a high resolution , long working distance objective suitable for imaging objects far within a vacuum chamber through a thick window .
optical tests confirm the performance predicted using ray - tracing software .
remarkably this objective enables diffraction - limited imaging across a wide fov of all the alkalis and ytterbium ions , by changing only the aperture diameter .
spherical aberration from a range of window thicknesses can be corrected by changing the singlet spacings .
the versatility and simplicity of this design makes it applicable to many experiments in the field of quantum gases .
this work was supported by australian research council grants dp1094399 and dp1096830 , and by a monash university faculty of science equipment grant . | we present a high resolution objective lens made entirely from catalog singlets that has a numerical aperture of 0.36 .
it corrects for aberrations introduced by a glass window and has a long working distance of 35 mm , making it suitable for imaging objects within a vacuum system .
this offers simple high resolution imaging for many in the quantum gas community .
the objective achieves a resolution of 1.3@xmath0 at the design wavelength of 780 nm , and a diffraction - limited field of view of 360@xmath0 when imaging through a 5 mm window .
images of a resolution target and a pinhole show quantitative agreement with the simulated lens performance .
the objective is suitable for diffraction - limited imaging on the d2 line of all the alkalis by changing only the aperture diameter , retaining numerical apertures above 0.32 .
the design corrects for window thicknesses of up to 15 mm if the singlet spacings are modified . 10 y. kawaguchi , m. ueda , `` spinor bose
einstein condensates , '' phys . rep . * 520 * , 253381 ( 2012 ) .
e. w. streed , a. jechow , b. g. norton , and d. kielpinski , `` absorption imaging of a single atom , '' nat . commun . * 3 * , 933 ( 2012 ) .
h. gross , _ handbook of optical systems _ , vol . 1 ( wiley - vch , 2005 ) .
y. r. p. sortais , h. marion , c. tuchendler , a. m. lance , m. lamare , p. fournet , c. armellin , r. mercier , g. messin , a. browaeys , and p. grangier , `` diffraction - limited optics for single - atom manipulation , '' phys .
rev .
a * 75 * , 013406 ( 2007 ) .
w. s. bakr , j. i. gillen , a. peng , s. flling , and m. greiner , `` a quantum gas microscope for detecting single atoms in a hubbard - regime optical lattice , '' nature * 462 * , 7477 ( 2009 ) .
e. a. salim , s. c. caliga , j. b. pfeiffer , and d. z. anderson , `` high - resolution imaging and optical control of bose
einstein condensates in an atom chip magnetic trap , '' arxiv:1208.4897 ( 2012 ) .
y .- j .
lin , a. r. perry , r. l. compton , i. b. spielman , and j. v. porto , `` rapid production of @xmath1rb bose
einstein condensates in a combined magnetic and optical potential , '' phys .
rev .
a * 79 * , 063631 ( 2009 ) .
d. douillet , e. rolley , c. guthmann , and a. prevost , `` an easy - to - build long working distance microscope , '' physica b * 284288 * , part 2 , 20592060 ( 2000 ) .
w. alt , `` an objective lens for efficient fluorescence detection of single atoms , '' optik * 113 * , 142144 ( 2002 ) .
t. b. ottenstein , `` a new objective for high - resolution imaging of bose
einstein condensates , '' diploma thesis , university of heidelberg ( 2006 ) .
k. d. nelson , x. li , and d. s. weiss , `` imaging single atoms in a three - dimensional array , '' nat .
phys . * 3 * , 556560 ( 2007 ) .
r. bcker , a. perrin , s. manz , t. betz , c. koller , t. plisson , j. rottmann , t. schumm , and j. schmiedmayer , `` single - particle - sensitive imaging of freely propagating ultracold atoms , '' new j. phys . * 11 * , 103039 ( 2009 ) .
j. f. sherson , c. weitenberg , m. endres , m. cheneau , i. bloch , and s. kuhr , `` single - atom - resolved fluorescence imaging of an atomic mott insulator , '' nature * 467 * , 6872 ( 2010 ) .
b. zimmermann , t. mller , j. meineke , t. esslinger , and h. moritz , `` high - resolution imaging of ultracold fermions in microscopically tailored optical potentials , '' new j. phys . * 13 * , 043007 ( 2011 ) .
b. p. anderson and m. a. kasevich , `` spatial observation of bose
einstein condensation of @xmath1rb in a confining potential , '' phys .
rev .
a * 59 * , 938941 ( 1999 ) .
the objective can also be modeled using the free software oslo edu , which is limited to ten surfaces , because the first surface is flat and need not be included in the model . h. gross , _ handbook of optical systems _ , vol . 3 ( wiley - vch , 2007 ) .
w. j. smith , _ modern lens design _
( 2^nd^ ed ) ( mcgraw - hill , 2005 ) .
r. e. fischer and k. l. mason , `` spherical aberration some fascinating observations , '' proc .
spie * 0766 * , 5360 ( 1987 ) .
p. t. starkey , c. j. billington , s. p. johnstone , m. jasperse , k. helmerson , l. d. turner , and r. p. anderson , `` a scripted control system for autonomous hardware - timed experiments , '' in preparation . |
molybdenum disulphide ( mos@xmath0 ) is a semiconductor with a bulk bandgap above 1.2 ev,@xcite which can be further manipulated by changing its thickness,@xcite or through application of mechanical strain.@xcite this finite bandgap is a key reason for the excitement surrounding mos@xmath0 as compared to graphene as graphene has a zero bandgap.@xcite because of its direct bandgap and also its well - known properties as a lubricant , mos@xmath0 has attracted considerable attention in recent years.@xcite for example , radisavljevic et al.@xcite demonstrated the application of single - layered mos@xmath0 ( slmos@xmath0 ) as a good transistor . the strain and the electronic noise effects
were found to be important for the slmos@xmath0 transistor.@xcite besides electronic properties , there has been increasing interest in the thermal and mechanical properties of mos@xmath0 .
several recent works have addressed the thermal transport properties of slmos@xmath0 in both ballistic and diffusive transport regimes.@xcite the mechanical behavior of slmos@xmath0 has been investigated experimentally.@xcite for the theoretical part , we have examined the size and edge effects on the young s modulus of the slmos@xmath0 based on the stillinger - weber ( sw ) potential.@xcite quite recently , we derived an analytic formula for the elastic bending modulus of the slmos@xmath0 , where the importance of the finite thickness effect was revealed.@xcite a fundamental property related to the thermal and mechanical behaviors is the lattice dynamics property of mos@xmath0 , i.e the phonon spectrum or phonon modes .
the thermal conductivity in the semiconductor mos@xmath0 is dominated by the lattice thermal transport , which is contributed by the phonon mode .
different mechanical processes are governed by the corresponding phonon modes .
for instance , the nanomechanical resonant oscillation takes advantage of the bending mode ( flexure mode ) .
recently , there have been increasing experimental and theoretical efforts on the lattice dynamics properties of the single - layer or few - layer mos@xmath0 nanosheets.@xcite however , up to date , the lattice dynamics of the single - walled mos@xmath0 nanotube ( swmos@xmath0nt ) was investigated only by few works , which show different lattice properties in the tube structure from the planar sheet structure , because of the special line group in the nanotube.@xcite in this paper , we perform the lattice dynamics study and the molecular dynamics simulation for the lattice properties of the swmos@xmath0nt with arbitrary chiralities . the phonon spectrum is calculated from the dynamics matrix with the force constant matrix obtained from the sw potential .
the helical quantum numbers are used to denote the phonon mode .
the eigen vector of the first twenty lowest - frequency phonon modes are presented .
furthermore , we show that the radial breathing mechanical oscillation is initially disturbed by the three - fold modes and eventually destroyed by the two - fold modes . [ 0.9 ] .
mo atoms are represented by big balls ( gray online ) .
s atoms are presented by small balls ( yellow online ) . above panel
: top view of the two - dimensional mos@xmath0 honeycomb lattice .
three lattice directions are depicted by arrows , @xmath6 , @xmath7 , and @xmath8 .
bottom panels : three swmos@xmath0nts are obtained by rolling up the above mos@xmath0 sheet onto a cylindrical surface with the three corresponding lattice vectors as the circumference.,title="fig:",width=302 ] [ 0.8 ] [ 0.8 ] [ 0.8 ] [ 0.8 ]
@xmath9 in which @xmath10 and @xmath11 are related to each other as : @xmath12 @xmath13 and @xmath14 are reciprocal unit vectors corresponding to @xmath15 . any wave vector in the reciprocal space
can be written as @xmath16 @xmath17 are the two helical quantum numbers . in the first brillouin zone ,
@xmath18 $ ] and @xmath19 is an integer in @xmath20 $ ] .
all screw and rotational symmetry operations in the swmos@xmath0nt form the space group of this tubal system , which is named line group.@xcite according to the irreducible representation of the line group , we have the generalized bloch theory for the vibration displacement of each atom @xmath21 in the phonon mode @xmath22,@xcite @xmath23 where @xmath24 is the branch index .
@xmath25 is the total number of the screw symmetry operation .
@xmath26 is the greatest common divisor of @xmath27 and @xmath28 .
it is the number of pure rotational symmetry operations .
@xmath29 gives the total number of unit cells in the tube .
@xmath30 corresponds to the three atoms s , mo , and s in the unit cell .
@xmath31 is a rotation matrix for rotation around the axial direction for angle @xmath32 .
@xmath33 is the vibration displacement for atom @xmath34 in the phonon mode denoted by @xmath35 .
@xmath36 is the helical quantum state of the phonon mode @xmath35 . applying the generalized bloch theory into the phonon dynamical equations ,
we can obtain the dynamics matrix of the swmos@xmath0nt as , @xmath37 where @xmath38 is the mass of atom @xmath39 .
for mos2 , the dynamical matrix is a @xmath40 matrix , since there are three atoms in each unit cell .
the diagonalization of the dynamical matrix results in nine phonon modes denoted by the helical index @xmath35 , with @xmath41 .
@xmath42 is the force constant matrix . in our calculation , this force constant matrix is obtained using finite differential method based on a newly developed sw potential .
the sw potential is developed for slmos@xmath0 , including two - body and three - body interactions .
we apply this potential to describe the atomic interaction within the swmos@xmath0nt .
all sw potential parameters can be found in our previous work.@xcite fig .
[ fig_dispersion ] shows the phonon spectrum in four swmos@xmath0nts with almost the same diameter of 18.0 .
the spectrum is symmetric about @xmath43 in all tubes , so only curves with @xmath44 $ ] are displayed .
panel ( a ) is the phonon spectrum for armchair tube ( 15 , 15 ) .
there are nine curves for each helical quantum number @xmath19 , corresponding to the nine degrees of freedom in each unit cell .
there are four acoustic branches .
the longitudinal acoustic ( la ) and twisting ( tw ) phonon branches have the same helical quantum number @xmath45 .
the other two acoustic branches are the transverse acoustic ( ta ) phonons with @xmath46 .
one of the ta mode locates at @xmath47 and @xmath48 , which is depicted by a solid arrow on the @xmath49 axis .
the other ta mode locates at @xmath50 and @xmath51 .
the two ta modes are also called flexure modes , and they have a parabolic spectrum due to the quasi - one - dimensional nature of the nanotubes .
panel ( b ) shows the phonon spectrum of the zigzag tube ( 26 , 0 ) .
panels ( c ) and ( d ) show the phonon spectrum for chiral tubes . in tube
( 16 , 13 ) , the quantum number @xmath19 has only one value , i.e @xmath45 , so there is only nine curves in this figure . one significant feature of fig .
[ fig_dispersion ] is the combination of an empirical many - body potential ( sw ) with the helical quantum numbers @xmath17 . in previous studies ,
the helical quantum numbers are used only in dynamical matrix constructed from force field models .
the empirical potential can give more comprehensive information for the phonon spectrum .
in particular , it fulfills the rigid rotational symmetry , leading to the zero frequency of the tw mode .
furthermore , the helical quantum numbers correspond to the actual symmetric operations ( screw and pure rotational symmetric operations ) in the swmos@xmath0nt . hence they are the good quantum numbers for the phonon modes in the system.@xcite the phase ` good ' means that these quantum numbers are conserved during the phonon - assisted physical process .
for instance , it has been shown that these helical quantum numbers give a natural ( the simplest ) selection rule for the phonon - phonon scattering process in the lattice thermal transport of the single - walled boron nitride nanotube.@xcite here , we have successfully combined the helical quantum numbers with empirical potential ; thus take advantage of both parts .
hence , fig .
[ fig_dispersion ] provides a ` good ' recipe for further investigations of the phonon - assisted physical processes in swmos@xmath0nts .
we now pay attention to the low - frequency modes in swmos@xmath0nts , because these modes are easier to be excited in practice .
[ fig_u_arm ] shows the first twenty lowest - frequency modes in the armchair tube ( 15 , 15 ) .
three big translational cells are used in this calculation .
the periodic boundary condition is applied in the axial direction .
these modes locate in the @xmath52 point in the brillouin zone ; i.e , the linear quantum number is zero .
the linear quantum number is the wave vector corresponding to the big translational operation .
numbers in the figure are the frequency of each mode in the unit of @xmath5 .
the figure is plotted with xcrysden.@xcite the arrow attached to each atom represents the component of this atom in the eigen vector of the phonon mode . [ 1.0 ] , for all data.,title="fig:",width=302 ] [ 0.9]nts of different chiralities .
tubes oscillate like a breather .
( a ) armchair ( 15 , 15 ) .
( b ) zigzag ( 26 , 0 ) .
( c ) chiral ( 20 , 10 ) .
arrows depict the life time , at which the radial breathing oscillation starts to be disturbed .
the inset in each panel shows the life time versus actuation parameter @xmath53 .
the solid line is the exponential fitting to the calculated data.,title="fig:",width=302 ] [ 0.8 ] the first four modes are the acoustic phonon modes with zero frequency .
their helical quantum numbers are @xmath54 , @xmath55 , @xmath56 , and @xmath57 .
in particular , the fourth mode is the tw mode , resulting from the particular hollow cylindrical structure of the swmos@xmath0nt .
these four acoustic modes have important contribution to the thermal transport of swmos@xmath0nts .
the fifth and sixth modes have the two - fold symmetry , which are denoted by @xmath58 . from their peculiar vibration morphology and their pretty low frequencies
, it can be easily imagined that the vibration of these modes is likely to squash the tube configuration , leading to possible instability of the swmos@xmath0nt .
the seventh and eighth modes have a three - fold symmetry , which are denoted by @xmath59 .
the thirteenth mode is the radial breathing mode ( rbm ) , which is denoted by @xmath60 .
the frequency of the rbm is sensitive to the tube diameter , due to its vibration morphology and this mode is raman active , so its frequency is usually used to estimate the tube diameter in the experiment .
[ fig_u_zig ] and [ fig_u_chiral ] show the first twenty lowest - frequency modes in tubes ( 26 , 0 ) and ( 20 , 10 ) , respectively .
in particular , the rbm has close frequency among these three tubes of quite different chiralities , since the chirality only takes effect on the order of@xcite @xmath61 .
in the above , we have studied the full phonon spectrum for swmos@xmath0nts .
the rest of this paper is devoted to the discussion of the rbm , since this mode plays an important role in the tubal structure .
we will study the diameter dependence and nonlinear properties of the rbm .
[ fig_frequency ] shows the frequency of the rbm in armchair , zigzag , and chiral tubes , i.e @xmath62 , @xmath63 , and @xmath64 .
the frequency of all tubes can be fitted to function @xmath4 @xmath5 , where @xmath3 is the diameter .
the frequency of the rbm in swmos@xmath0nts is much lower than that in the single - walled carbon nanotube with the same diameter .
the coefficient here 665.3 is about one third of the value of 2295.6 in the single - walled carbon nanotube.@xcite it is because the frequency of the rbm is related to the tensile mechanical properties,@xcite and the mechanical strength is weaker in the swmos@xmath0nt.@xcite furthermore , mo and s atoms are heavier than the c atom in carbon nanotubes .
[ fig_frequency ] can be useful in the estimation of the diameter for swmos@xmath0nts in the experiment .
we further investigate the nonlinear effect on the rbm .
the molecular dynamics is performed to simulate the radial breathing oscillation of swmos@xmath0nts ( 15 , 15 ) , ( 26 , 0 ) , and ( 20 , 10 ) .
the periodic boundary condition is applied in the axial direction .
the tube is optimized to the energy minimum configuration . at the optimized configuration ,
the initial velocity distributions are set according to the vibration morphology of the rbm .
the total kinetic energy corresponding to this initial velocity distribution is @xmath65 , where @xmath53 is the energy actuation parameter.@xcite @xmath26 is the total atom number .
mev is a small piece of energy quantum for a single atom .
the tube is allowed to oscillate within the nve ensemble with the initial velocity distribution .
the total kinetic energy and the potential energy exchange between each other during this radial breathing oscillation . in the initial oscillation stage , there is only a single oscillation mode , i.e the radial breathing oscillation .
if the oscillation is in the linear regime , i.e it has a small amplitude , this radial breathing oscillation can be preserved for a very long time and other vibration modes are seldom excited . however , if the oscillation is in the nonlinear regime , i.e with larger actuation parameter @xmath53 , then the other oscillation modes will be excited quickly . as a result
, the radial breathing oscillation decays quickly , owning to the nonlinear induced phonon - phonon scattering between the rbm and the other excited modes .
[ fig_energy ] shows the kinetic energy time history during the radial breathing oscillation of the three tubes . to study the nonlinear effect ,
we have shown only simulations with large actuation parameters @xmath67 , 30 , and 50 in the figure .
the radial breathing oscillation will not decay if very small actuation parameter is used , eg .
@xmath68 , which will be similar as the graphene nanomechanical resonator.@xcite fig .
[ fig_energy ] shows that in the initial stage the total kinetic energy oscillates between zero and a maximum value , which reflects the good resonant oscillation of the swmos@xmath0nt . after some time , the radial breathing oscillation starts to be disturbed by some other vibration modes , which are excited by the nonlinear effect relating to the large actuation energy .
arrows in the figure depict the life time , at which the radial breathing oscillation starts to be disturbed .
the inset in each panel shows the life time versus actuation parameter @xmath53 .
the inset shows that the life time decays exponentially with increasing actuation parameter @xmath53 .
the solid line is the exponential fitting to the calculated data .
this set of simulations display that it is important to use small actuation parameter for the radial breathing oscillation , if this oscillation is used as the nanomechanical resonator.@xcite in all of the three tubes , for very large actuation parameters such as @xmath69 and 50 , the kinetic energy increases sharply after the radial breathing oscillation is disturbed .
it indicates that these tubes undergo a structure transition , which releases lots of potential energy .
this potential energy is converted into the kinetic energy in the system . fig .
[ fig_cfg_md ] illustrates the structure evolution during the radial breathing oscillation in the swmos@xmath0nt ( 20 , 10 ) with actuation parameter @xmath70 .
these snap - shots are produced by ovito.@xcite similar phenomena have been observed in the molecular dynamics simulation of tubes ( 15 , 15 ) and ( 26 , 0 ) .
the arrow attached to each atom represents the velocity of the atom at that moment . at @xmath71
ps , the radial breathing oscillation is actuated by adding an initial velocity distribution , which follows the eigen vector of the rbm ( 13rd mode ) in fig .
[ fig_u_chiral ] .
configurations at @xmath72 and 16.0 ps show that the radial breathing oscillation is disturbed gradually by the seventh and eighth modes ( with three - fold symmetry ) in fig .
[ fig_u_chiral ] due to the nonlinear effect relating to large actuation energy .
the radial breathing oscillation is completely replaced by these three - fold modes after @xmath73 ps .
these three - fold oscillation is clearly demonstrated by the two configurations at @xmath74 and 22.3 ps , which exactly follow the eigen vector of the two three - fold modes .
this shows that the three - fold mode is also a type of stable oscillation , i.e this oscillation does not destroy the tubal structure .
however , after @xmath75 ps , the fifth and sixth modes ( with two - fold symmetry ) in fig .
[ fig_u_chiral ] start to be excited .
these two - fold modes are unstable due to their special vibration morphology and their low frequencies .
a low frequency indicates that these modes are able to deform the tubal structure quite a lot with small amount of energy .
indeed , the tube starts to be squashed by these two - fold modes after @xmath76 ps .
in conclusion , we study the lattice dynamics properties of the swmos@xmath0nt with arbitrary chiralities . in the construction of the @xmath77 dynamical matrix , the force constant matrix is calculated from the sw potential and the phonon modes are denoted by helical quantum numbers , which correspond to the line group in the system .
the frequency and eigen vector of the low - frequency phonon modes are analyzed .
the frequency of the rbm is found to be inversely proportional to the tube diameter as @xmath4 @xmath5 , where @xmath3 is the tube diameter .
we perform the molecular dynamics simulation to investigate the radial breathing mechanical oscillation , and find that the life time of the radial breathing oscillation decays exponentially with increasing oscillation energy .
more specifically , in the initial stage , this radial breathing oscillation is disturbed by the three - fold modes .
the swmos@xmath0nts are squashed by the two - fold modes in the final stage , if the oscillation is actuated into nonlinear regime .
48ifxundefined [ 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 * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) | we study the phonon modes in single - walled mos@xmath0 nanotubes via the lattice dynamics calculation and molecular dynamics simulation .
the phonon spectra for tubes of arbitrary chiralities are calculated from the dynamical matrix constructed by the combination of an empirical potential with the conserved helical quantum numbers @xmath1 .
in particular , we show that the frequency ( @xmath2 ) of the radial breathing mode is inversely proportional to the tube diameter ( @xmath3 ) as @xmath4 @xmath5 .
the eigen vectors of the first twenty lowest - frequency phonon modes are illustrated .
based on these eigen vectors , we demonstrate that the radial breathing oscillation is disturbed by phonon modes of three - fold symmetry initially , and the tube is squashed by the modes of two - fold symmetry eventually .
our study provides fundamental knowledge for further investigations of the thermal and mechanical properties of the mos@xmath0 nanotubes . |
information flow , or information transfer as it may be referred to in the literature , has long been recognized as the appropriate measure of causality between dynamical events@xcite .
it possesses the needed asymmetry or directionalism for a cause - effect relation , and , moreover , provides a quantitative characterization of the otherwise statistical test , e.g. , the granger causality test@xcite .
for this reason , the past decades have seen a surge of interest in this arena of research .
measures of information flow proposed thus far include , for example , the time - delayed mutual information@xcite , transfer entropy@xcite , momentary information transfer@xcite , causation entropy@xcite , etc . , among which transfer entropy has been proved to be equivalent to granger causality up to a factor 2 for linear systems@xcite .
recently , liang and kleeman find that the notion of information flow actually can be put on a rigorous footing within a given deterministic system@xcite .
the basic idea can be best illustrated with a system of two components , say , @xmath0 and @xmath1 .
the problem here essentially deals with how the marginal entropies of @xmath0 and @xmath1 , written respectively as @xmath2 and @xmath3 , evolve .
take @xmath2 for an example
. its evolution could be due to @xmath0 its own and/or caused by @xmath1 .
that is to say , @xmath4 can be split exclusively into two parts : @xmath5 if we write the contribution from the former mechanism as @xmath6 and that from the latter as @xmath7 .
this @xmath7 is the very time rate of information flowing from @xmath1 to @xmath0 . to find the information flow @xmath7
, it suffices to find @xmath8 , since , for each deterministic system , there is a liouville equation for the density of the state and , accordingly , @xmath4 can be obtained . in ref .
@xcite , @xmath8 is acquired through an intuitive argument based on an entropy evolutionary law established therein .
the same result is later on rigorously proved ; see @xcite for a review . for stochastic systems which we will be considering in this study , the trick ceases to work , but in @xcite liang manages to circumvent the difficulty and find the result , which we will be briefly reviewing in the following .
consider a two - dimensional ( 2d ) stochastic system @xmath9 where @xmath10 is the vector of drift coefficients ( differentiable vector field ) , @xmath11}}$ ] the matrix of stochastic perturbation coefficients , and @xmath12 a 2d standard wiener process .
let @xmath13 , and @xmath14 be the marginal probability density function of @xmath15 .
liang ( 2008)@xcite proves that the time rate of information flowing from @xmath1 to @xmath0 is @xmath16 where @xmath17 signifies the operator mathematical expectation .
this measure of information flow is asymmetric between the two parties and , particularly , if the process underlying @xmath0 does not depend on @xmath1 , then the resulting causality from @xmath1 to @xmath0 vanishes .
this is the so - called _ property of causality _ , which asserts , in the above language , that @xmath7 vanishes if @xmath18 and @xmath19 are independent of @xmath20 .
when @xmath7 is nonzero , it may take positive or negative values .
a positive @xmath7 means @xmath1 causes @xmath0 to be more uncertain , while a negative @xmath7 reduces the entropy of @xmath0 , and hence functions to stabilize the latter . for more details ,
referred to ref .
@xcite .
the above theorem is recently applied to time series analysis . under the assumption of a linear model with additive noise
, liang@xcite shows that the maximal likelihood estimate of the information flow in ( [ eq : t21 ] ) turns out to be very tight in form , involving only the common statistics namely sample covariances .
take two series @xmath0 and @xmath1 for example .
the rate of information flowing ( units : nats per unit time ) from @xmath1 to @xmath0 is shown to be @xmath21 where @xmath22 is the sample covariance between @xmath23 and @xmath24 , and @xmath25 is that between @xmath23 and @xmath26 , @xmath26 being the difference approximation of @xmath27 using the euler forward scheme : @xmath28 here @xmath29 is usually 1 , but for highly chaotic and densely sampled series , @xmath30 should be chosen to avoid getting spuriously large @xmath26 due to possible shock structures that make the differencing highly sensitive to the error in @xmath24 .
this formula involves only sample covariances , and is hence very convenient to evaluate .
in addition , it is easy to see that if @xmath31 then @xmath32 , but when @xmath32 the correlation @xmath33 does not need to vanish .
that is to say , contrapositively , _
causation implies correlation , but correlation does not imply causation_. in an explicitly quantitative way , this corollary resolves the long - standing debate over causation versus correlation . and @xmath1 ) of the autoregressive process ( [ eq : ar1 ] ) for ( a ) @xmath34 , @xmath35 , and ( b ) @xmath36 @xmath35
. , scaledwidth=75.0% ] in general , the rates of information flow differ from case to case .
one would like to normalize them in real applications , just as that with correlation coefficients , in order to assess the importance of the flow identified , if any .
for example , consider two series generated from two autoregressive processes : [ eq : ar1 ] @xmath37 where the errors @xmath38 and @xmath39 are independent .
generate a pair of series with 80000 steps on matlab , and perform the causality analysis ( with @xmath40 ) , one obtains ( units are in nats per iteration ; same below in this section ) * for @xmath41 , @xmath42 , @xmath43 ; * for @xmath44 : @xmath45 , @xmath46 .
( the results may differ slightly with different series due to the random number generation . ) for the case @xmath34 , @xmath47 , one may then conclude that this is a one - way causality from @xmath1 to @xmath0 , as is indeed true . for the latter
, however , one actually can not say much from the numbers .
although they are small , they tell no more than that the information flows in both directions are of equal importance . from this particular example , one sees that , though theoretically the information flows in both directions should be precisely zero , in reality the rates evaluated from two time series , albeit very small , generally do not precisely vanish .
one then can not tell whether the causality indeed exists .
we need to normalize the obtained information flow for one to see the relative magnitude .
of course , one may perform statistical test for the results . in the first example , at a 90% significance level , @xmath48 @xmath49
so @xmath50 is insignificant . for the second example , @xmath51 @xmath52 that is to say , at a 90% level , these flow rates are not significantly different from zero . in this sense
, it seems that we could indeed infer rather accurately the true causality .
however , a statistical test just tells how precise the estimate is ; it does not tell how the information flow may weigh in the entropy balance of the series .
besides , it depends on the length of the series which is irrelevant to the parameter to be estimated . to see this more clearly ,
consider the following case : @xmath53 obviously , the information flows , albeit existing , make only tiny contributions to their respective series , as the coupling coefficients are over an order smaller ; in classical perturbation analysis , they can be dropped to the first order approximation .
the computed information flows , at a 90% significance level , are @xmath54 @xmath55 these results indicate that they are significantly different from zero this from one aspect testifies to the success of the formalism .
however , the small numbers can not tell how important they are , since , with a slowly varying series , even the dominant flow could be very small . on the other hand , if we cut the series by half and pick the first 40000 points for the analysis , then the result will be @xmath56 @xmath57 so one finds that @xmath7 is insignificant while @xmath50 is .
( again , these small numbers may show large fluctuation if series of different lengths are used , because in reality they are insignificant . )
can one thus conclude that there is a one - way causality , or can he / she thus asserts that this shortened series yields a more reliable estimation ?
surely this is absurd .
the problem here is that we do need a normalized flow to evaluate its importance relative to other factors .
the normalization is by no means as simple as it appears with correlation analysis , which is in an inner product form based on the cauchy - schwarz inequality . in the following
we will see that we need to get down to the fundamentals of information flow before arriving at a logically and physically sound normalizer .
as what we did before in @xcite , this normalizer is estimated with the method of maximal likelihood estimation , using the given time series ( section [ sect : estimation ] ) .
the resulting formula is then validated ( section [ sect : validation ] ) with the autoregression example as shown above .
to demonstrate its diverse applicability , presented subsequently are two real world examples , one from climate science ( section [ sect : climate ] ) , another from financial economics ( section [ sect : finance ] ) .
this study is summarized in section [ sect : summary ] .
as mentioned in the introduction , the normalization is not as simple as it seems to be . a natural normalizer that comes to mind , at the hint of correlation coefficient ,
might be the information of a series transferred from itself .
a snag is , however , that this quantity may turn out to be zero , just as that in the hnon map , a benchmark problem we have examined before ( see the references in @xcite ) .
another snag is , the two way causality actually can not be normalized together , as that in correlation analysis based on the cauchy - schwarz inequality .
that is to say , two information flows of equal size may have different relative importances in their respective series .
since in our framework the information flow from @xmath1 to @xmath0 is the contribution of @xmath1 that makes to the time rate of change of the marginal entropy of @xmath0 , written @xmath2 , one may ask whether the rate of marginal entropy change @xmath4 can be the normalizer .
this might be appealing , but there is a third snag . as information flow
can be positive or negative , @xmath4 may turn out to be smaller than the flow in absolute value the so - obtained relative flow would exceed 100% , a case which we do not want to see .
all the above tells that information flow normalization is by no means a trivial task .
we need to get to the basics and analyze how an information flow within a system is derived . by ref .
@xcite , the time rate of change of the marginal entropy of @xmath0 is @xmath58 it is actually a result of two mutually exclusive mechanisms : the first is the information flow @xmath7 as shown in ( [ eq : t21 ] ) ; the second is the complement , i.e. , the rate of entropy increase without taking into account of the effect of @xmath1 .
denote this latter as @xmath59 , liang ( 2008)@xcite has proved that @xmath60 the right hand side has three terms .
the first term is precisely the time rate of change of @xmath2 due to @xmath0 itself in the absence of stochasticity .
this is the starting point which we have shown in 2005@xcite in establishing the rigorous formalism and proved later on ( cf .
hence through a careful analysis , the increase in the marginal entropy @xmath2 is decomposed into three parts : @xmath61 as well as @xmath7 in ( [ eq : t21 ] ) , which correspond to , respectively , the contribution due to @xmath0 itself , the stochastic effect , and the information flowing from @xmath1 .
note this decomposition does not appear explicitly in the marginal entropy evolution equation ( [ eq : dh1 ] ) , as the two stochastic terms cancel out .
the normalization is now made easy .
let @xmath62 obviously it is no less than @xmath7 in magnitude , and can not be zero unless @xmath0 does not change , a situation that is excluded in time series analysis .
we may therefore pick @xmath63 as the normalizer , and define @xmath64 this way if @xmath65 , the variation of @xmath2 is 100% due to the information flow from @xmath1 ; if @xmath66 is approximately zero , @xmath1 is not causal
. therefore , @xmath66 assesses the importance of the influence of @xmath1 to @xmath0 relative to other processes .
it should be pointed out that , the above normalizer applies to @xmath7 only . for @xmath50 ,
it is @xmath67 which may be quite different in value .
this from another aspect reflects the asymmetry between @xmath7 and @xmath50 .
as in ref .
@xcite , consider a linear version of the stochastic differential equation ( sde ) ( [ eq : sde ] ) @xmath68 where @xmath69 is a constant vector , and @xmath70 and @xmath71 are constant matrices .
initially if @xmath72 obeys a gaussian distribution , then it is a gaussian for ever , i.e. , @xmath73 with the mean @xmath74 and covariance matrix @xmath75 governed by equations @xmath76 so eqs .
( [ eq : dh1star ] ) and ( [ eq : dh1noise ] ) can be explicitly evaluated : @xmath77 and @xmath78 } -\frac12 \iint_{\r^2 } \rho_{2|1 } { \frac { { \partial}^2 g_{11}\rho_1 } { { \partial}x_1 ^ 2 } } dx_1dx_2 \cr & = & \frac12 \frac { g_{11 } } { \sigma_1 ^ 2 } - \frac12 \int_\r { \frac { { \partial}^2 g_{11}\rho_1 } { { \partial}x_1 ^ 2 } } { \left(\int_\r \rho_{2|1}(x_2 | x_1 ) dx_2\right ) } dx_1 , \end{aligned}\ ] ] since neither @xmath19 nor @xmath79 depends on @xmath20 . but @xmath80 , and @xmath79 is compactly supported , the whole second term on the right hand side then vanishes . hence @xmath81 where for notational symmetry @xmath82 has been written as @xmath83 .
these , together with the information flow from @xmath1 to @xmath0 as we have obtained before@xcite@xcite , @xmath84 form the three constituents that account for the evolution of the marginal entropy of @xmath0 .
an observation about @xmath85 , where @xmath86 is that it is always positive . that is to say
, the noise always contributes to increase the marginal entropy of @xmath0 , conforming to our common sense . in financial economics ,
this reflects the volatility of , say , a stock . on the other hand , for a stationary series , i.e. , when @xmath87 , the balance on the right hand side of eq .
( [ eq : sigma ] ) requires that @xmath88 .
so this quantity is also related to the noise - to - signal ratio .
the above results need to be estimated if what we are given are just a pair of time series .
that is to say , what we know is a single realization of some unknown system , which , if known , can produce infinitely many realizations . the problem now is turned into estimating ( [ eq : dh1star_lin ] ) and ( [ eq : dh1noise_lin ] ) with the available statistics of the given time series . we use maximum likelihood estimation ( e.g. , @xcite ) to achieve the goal .
the procedure follows precisely that of @xcite , which for easy reference we briefly summarize here . as established before , a further assumption that @xmath89 , and hence @xmath90 , will much simplify the result , while in practice this is quite reasonable .
suppose that the series are equal - distanced with a time stepsize @xmath91 , and let @xmath92 be the sample size .
consider an interval @xmath93 $ ] , and let the transition probability function ( pdf ) be @xmath94 , where
@xmath95 stands for the vector of parameters to be estimated .
so the log likelihood is @xmath96 as @xmath92 is usually large , the term @xmath97 can be dropped without causing much error .
the transition pdf is , with the euler - bernstein approximation ( see @xcite ) , @xmath98^{1/2 } } \times e^{-\frac12 ( { { \bf x}}_{n+1 } - { { \bf x}}_n - { { \bf f}}{{\delta t}})^t ( { { \bf { b}}}{{\bf { b}}}^t{{\delta t}})^{-1 } ( { { \bf x}}_{n+1 } - { { \bf x}}_n - { { \bf f}}{{\delta t } } ) } , \end{aligned}\ ] ] where @xmath99 .
this results in a log likelihood functional @xmath100 where @xmath101 and @xmath102 is the euler forward differencing approximation of @xmath103 : @xmath104 with @xmath105 .
usually @xmath40 should be used to ensure accuracy , but in some cases of deterministic chaos and the sampling is at the highest resolution , one needs to choose @xmath30 . maximizing @xmath106 , we find@xcite that the maximizer @xmath107 satisfies the following algebraic equation : @xmath108 } } { { \left[\begin{array}{l } \hat f_1 \\
\hat a_{11 } \\
\hat a_{12 } \end{array}\right ] } } = { { \left[\begin{array}{l } \overline{\dot x_1 } \\ \overline{x_1\dot x_1 } \\
\overline{x_2 \dot x_1 } \end{array}\right ] } } , \end{aligned}\ ] ] where the overline signifies sample mean .
after some manipulations ( see @xcite ) , this yields the mle estimators : @xmath109 where @xmath110 are the sample covariances , and @xmath111}^2\cr & = & \sum_{n=1}^n { \left [ ( \dot x_{1,n } - \overline{\dot x_{1,n } } ) - \hat a_{11 } ( x_{1,n } - \bar x_1 ) - \hat a_{12 } ( x_{2,n } - \bar x_2 ) \right]}^2 \cr & = & n ( c_{d1,d1 } + \hat a_{11}^2 c_{11 } + \hat a_{12}^2 c_{22 } - 2\hat a_{11 } c_{d1,1 } - 2\hat a_{12 } c_{d1,2 } + 2\hat a_{11 } \hat a_{12 } c_{12 } ) .
\end{aligned}\ ] ] on the other hand , the population covariance matrix @xmath75 can be rather accurately estimated by the sample covariance matrix @xmath112 .
so ( [ eq : dh1star_lin])-([eq : dh1noise_lin ] ) become @xmath113 as that in ref .
@xcite with @xmath7 , here @xmath114 and @xmath115 should bear a hat , since they are the corresponding estimators .
we abuse the notation a little bit to avoid notational complexity ; from now on they should be understood as their respective estimators .
with these the normalizer is @xmath116 and hence we have the relative information flow from @xmath1 to @xmath0 : @xmath117
back to the autoregressive process exemplified in the beginning . when @xmath118 , @xmath35 , the computed relative information flow rates are : @xmath119 clearly both are negligible in comparison to the contributions from the other processes in their respective series .
this is in agreement with what one would conclude based on the absolute information flow computation and statistical testing . for the case
@xmath120 , in which one may encounter difficulty due to the ambiguous small numbers , the computed relative information flow rates are : @xmath121 again they are essentially negligible , just as one would expect . on the other hand , when @xmath34 , @xmath35 , @xmath122 to @xmath0 , the influence from @xmath1 is large , contributing to more than 1/6 of the total entropy change .
in contrast , the influence from @xmath0 to @xmath1 is negligible .
it should be pointed out that the relative information flow , say , @xmath66 , makes sense only with respect to @xmath0 , since the comparison is within the series itself .
here comes the following situation : for a two - way causal system with absolute information flows @xmath7 and @xmath50 of equal importance , their relative importances within their respective series could be quite different .
for example , @xmath123 where @xmath124 and @xmath125 are identical independent normals ( @xmath126n(0,1 ) ) .
initialize them with random values between [ 0,1 ] and generate 80000 data points on matlab .
the computed information flow rates ( in nats per iteration ) @xmath127 which are almost the same .
the relative information flows , however , are quite different : @xmath128 in terms of relative contribution in their respective series , the former is way more below the latter . generally speaking , the above imbalance is a rule , not an exception , reflecting the asymmetry of information flow .
one may reasonably imagine that , in some extreme situation , a flow might be dominant while its counterpart is negligible within their respective series , although the two are of the same order in absolute value .
el nio , also known as el nio - southern oscillation , or enso for short , is a long known and extensively studied climate mode in the tropical pacific ocean due to its relation to the global disasters like the droughts in southeast asia , southern africa , and northern australia , the floods in ecuador , the increasing number of typhoons , the death of birds and dolphins in peru , and the famine and epidemic diseases in far - flung regions of the world@xcite . a correct forecast of an el nio ( or its cold counterpart , la nia ) a few months earlier will not only help issue in - advance warnings of potential disastrous impacts , but also make the subsequent seasonal weather forecasting much easier .
however , this aperiodic leading mode in the tropical pacific seems to be extraordinarily difficult to predict .
a good example is the latest `` super el nio '' or `` monster el nio '' , which has been predicted to arrive in 2014 in a lot of portentous forecasts , turns out to be a computer artifact . for more reliable predictions
, it is imperative to clarify the source of its uncertainty or unpredictability . in ref .
@xcite , we have presented an application of eq .
( [ eq : t21_est ] ) to the relation study between el nio and the indian ocean dipole ( iod ) , another major climate mode in the indian ocean@xcite , and found that the indian ocean is a source of uncertainty that keeps enso from being accurately predicted . since in that study
there is no relative importance assessment , we do not know whether the information flows , albeit significant , do weigh much in the modal variabilities .
we hence redo the computation using the relative information flow formula ( [ eq : tau21 ] ) .
we use for this study the same data as that used in @xcite , which include the nio4 index series and the sea surface temperature ( sst ) series downloaded from the noaa esrl physical sciences division@xcite , and the iod index namely dmi series from the jamstec site@xcite . shown in fig .
[ fig : indian ] is the relative information flow rate from the indian ocean sst to el nio , @xmath129 . from it
one can see that the information flow accounts for more than 10% of the uncertainties of nio4 , the maximum reaching 27% .
this number is very large .
besides , all the values are positive , indicating that the indian ocean sst functions to make el nio more uncertain .
no wonder recently researchers find that assimilation of the indian ocean data helps the prediction of el nio ( e.g. , @xcite ) , although traditionally the indian ocean is mostly ignored in el nio modeling . besides the relative importance we have just obtained , fig .
[ fig : indian ] also reveals some difference in structure from its counterpart , i.e. , the fig .
5b of @xcite .
a conspicuous difference is that now there are clearly two centers , residing on either side of indian .
note this structure is different from the traditional dipolar pattern as one would expect ; here both centers are positive .
this means that the northern indian ocean sst anomalies , both the positive phase and negative phase , as an integral entity influence the el nio variabilities , and , in particular , make el nio more unpredictable .
the dipolar structure implies that most probably this entity is iod , not others like iobm ( indian ocean basin mode ; see @xcite ) . .
, scaledwidth=60.0% ] to see more about this , we look at the information flow from the index dmi to the tropic pacific sst .
the absolute rates are referred to the fig .
4a of @xcite ; shown in fig .
[ fig : pacific ] are @xmath130 .
indeed the computed flow rates are significant , and all are positive .
the largest @xmath131 , which occupies a large swathe of the equatorial region between @xmath132 through @xmath133 , reaches 10% .
moreover , the structure reminds one of the el nio pattern .
it is generally the same as that in its counter part , i.e. , fig .
4a of @xcite , save for two changes : ( 1 ) the maximum center moves westward ; ( 2 ) the small center of a secondary maximum near @xmath134 at the equator disappears .
this clear el nio - like structure attests to the above conjecture that iod is indeed a major source of uncertainty for the el nio forecast . .
, scaledwidth=100.0% ] we have also computed the relative information flows from el nio to the indian ocean sst , and that from the pacific sst to the iod , using the same datasets .
the results are also significant , though only approximately half as shown above .
we now look at the causal relations between several financial time series . here
it is not our intention to conduct an financial economics research or study the market dynamics from an econophysical point of view ; our purpose is to demonstrate a brief application of the aforementioned formalism for time series analysis .
nonetheless , this topic is indeed of interest to both physicists and economists in the field of macroscopic econophysics ; see , for example , @xcite .
we pick nine stocks in the united states and download their daily prices from @xmath135 .
these stocks are : msft ( microsoft corporation ) , aapl ( apple inc . ) , ibm ( international business machines corporation ) , intc ( intel corporation ) , ge ( general electric company ) , wmt ( wal - mart stores inc . ) , xom ( exxon mobil corporation ) , cvs ( cvs health corporation ) , f ( ford motor corporation ) . among these are high - tech companies ( msft , aapl , ibm , intc ) , retail trade companies
[ e.g. , the drugstore chains ( cvs ) and discount stores ( wmt ) ] , automotive industry ( f ) , oil and gas industry ( xom ) , and the multinational conglomerate corporation ge which operates through the segments of energy , technology infrastructure , capital finance , etc . here by `` daily '' we mean on a trading day basis ,
excluding , say , holidays and weekends .
since stock prices are generally nonstationary , we check the series of daily return , i.e. , @xmath136 / p(t),\ ] ] or log - return @xmath137 where @xmath138 are the adjusted closing prices in the yahoo spreadsheet , and @xmath139 is one trading day .
following most people we use the series of log - returns @xmath140 for our purpose .
in fact , return and log - return series are approximately equivalent , particularly in the high - frequency regime , as indicated by @xcite .
since the most recent stock msft started on march 13 , 1986 , all the series are chosen from that date through december 26 , 2014 , when we started to examine these series .
this amounts to 7260 data points , and hence 7259 points for the log - return series . using eq .
( [ eq : t21_est ] ) , we compute the information flows between the nine stocks and form a matrix of flow rates ; see table [ tab : stocks ] . a flow direction is represented with the matrix indices ; more specifically , it is from the row index to the column index .
for example , listed at the location ( 2,4 ) is @xmath141 , i.e. , @xmath142 , the flow rate from apple to intel , while ( 4,2 ) stores the rate of the reverse flow , @xmath143 . also listed in the table
are the respective confidence intervals at the 90% level . from table [ tab : stocks ] , most of the information flow rates are significant at the 90% level , as highlighted .
their values vary from 4 to 22 ( units : @xmath144 nats / day ; same below in this section ) .
the maximum is @xmath145 , and second to it are @xmath146 and @xmath147 , both being @xmath148 .
the influence of ibm to exxon is not a surprise , considering the dependence of the oil industry on high - tech equipments .
the mutual causality between the retail stores wmt and cvs are also understandable .
the information flow from cvs to ge could be through the sales of ge products ; after all , ge makes household appliances . for the rest in the table
, they can be summarized from the following two aspects .
* companies as sources .
+ look at the table row by row .
perhaps the most conspicuous feature is that the whole cvs row is significant .
next to it is xom , with only three entries insignificant .
that is to say , cvs has been found causal to all other stocks , though the causality magnitudes are yet to be assessed ( see below ) .
this does make sense . as a chain of convenience stores ,
cvs connects most of the general consumers and commodities and hence the corresponding industries . for xom , it is also understandable that why it makes a source of causality .
oil or gas is for sure one of the most fundamental component in the american economy .
* companies as recipients .
+ examining column by column , the most outstanding stock is again cvs , with only one entry insignificant .
that is to say , cvs is influenced by all other stocks except xom .
following cvs is xom , wmt , and intc .
the ibm and msft columns form the third tier .
a few words regarding the stock f. as a cause to other stocks ( though causality maybe tiny ) , xom has not been identified to be causal to f. in fact , f has not been found causal to xom , either . this is a little surprising ; the reason(s ) can be found only after a careful analysis of ford , which is beyond the scope here .
( in fact , computation does reveal information flows between xom and toyota . )
interestingly , @xmath149 .
this is easy to understand , as we rely on our motor vehicles to shop at wal - mart , while cvs stores could be just somewhere in the neighborhood !
.the rates of absolute information flow between the 9 chosen stocks ( in @xmath144 nats per trading day ) . at each entry
the direction is from the row index to the column index of the matrix .
also listed are the standard errors at a 90% significance level , and highlighted are the significant flows . [ cols="^,^,^,^,^,^,^,^,^,^",options="header " , ] it should be noted that the causal relations generally change with time .
if the series are long enough , we may look at how these information flows may vary from period to period . pick the pair ( ibm , ge ) as an example . for the duration ( march 1986 through present )
considered above , @xmath150 , while @xmath151 is not significant .
neither @xmath152 nor @xmath153 reaches 1% . since from the yahoo site both ge and ibm
can be dated back to january 2 , 1962 , we can extend the time series a lot up to 13338 data points . shown in fig .
[ fig : ibm_ge]a are the series of their historic prices , and in fig .
[ fig : ibm_ge]b and [ fig : ibm_ge ] c are the corresponding log - returns . computation of information flows with the whole series ( 13338 points ) results in @xmath154 and @xmath155 , and @xmath156 , @xmath157 nats / day , both being significant at the 90% level .
this is very different from what are shown in tables [ tab : stocks ] and [ tab : stocks_norm ] , with the causal structure changed from a weak two - way causality to a stronger and more or less one - way causality .
since in the above only the data of the recent 30 years are used , we expect that in the early years this causal structure could be much enhanced .
choose the first 7000 points ( from january 1962 through november 1989 ) , the computed relative information flow rates are : @xmath158 where the units for the latter pair are in @xmath144 nats / day , same below in this section .
further narrow down the period to 2250 - 3250 ( corresponding to the period 1971 - 1975 ) , then @xmath159 attaining the maximum of @xmath153 , in contrast to the insignificant flow in table [ tab : stocks ] .
obviously , during this period , the causality can be viewed as one - way , i.e. , from ibm to ge .
and the relative flow makes more than 5% , much larger than those in table [ tab : stocks_norm ] . the above remarkable causal structure for that particular period actually can trace its reason back in the history of ge@xcite .
there is such a period in 1960 s when `` seven dwarfs '' ( burroughs , sperry rand , control data , honeywell , general electric , rca and ncr ) competed with ibm the giant for computer business , and , particularly , to build mainframes . in 1965
, ge had only a 3.7-percent market share of the industry , though it was then dubbed as the `` king of the dwarfs '' , while ibm had 65.3% share .
historically ge was once the largest computer user outside the us federal government ; it got into computer manufacturing to avoid dependency on others . and , indeed ,
throughout the 60s , the causalities between ge and ibm are not significant . then , why , as time entered 70s , was the information flow from ibm to ge suddenly increased to its highest level ?
it turned out that ge sold its computer division to honeywell in 1970 ; in the following years ( starting from 1971 ) , it relied much on the ibm products .
this ge computer era , which has almost gone to oblivion , does substantiate the existence of a causation between ge and ibm , and , to be more precise , an essentially one - way causation from ibm to ge . in this sense
, our formalism is well validated .
to assess the importance of a flow of information from a series , say , @xmath160 , to another , say , @xmath161 , it needs to be normalized .
the normalization can not follow a way as that in computing a correlation coefficient , since there is no such a theorem like the cauchy - schwarz inequality for it to base .
getting down to the fundamentals , we were able to distinguish three types of mechanisms that contribute to the evolution of the marginal entropy of @xmath0 , @xmath2 , similarly there are three such quantities for @xmath1 , as schematized in fig .
[ fig : schem_info_flow ] .
we hence proposed that the normalization can be fulfilled as follows : @xmath163 obviously , a normalized flow tells its importance relative to other mechanisms within its own series .
in other words , the two flows are normalized differently , echoing the property of asymmetry which makes information flow analysis distinctly different from those such as correlation analysis or mutual information analysis .
the above normalizer can be accurately obtained in the framework of a dynamical system .
when only two equi - distanced series , say , @xmath0 and @xmath1 , are given , with a linear model its constituents can be estimated as follows , in a nutshell , @xmath164 } , \\ & & t_{2\to1 } = \frac { c_{11}c_{12}c_{2,d1 } - c_{12}^2c_{1,d1 } } { c_{11}^2 c_{22 } - c_{11 } c_{12}^2 } , \end{aligned}\ ] ] where @xmath91 is the time stepsize , @xmath165 @xmath22 the sample covariance between @xmath23 and @xmath24 , @xmath25 the sample covariance between @xmath23 and @xmath26 , and @xmath166 ( @xmath40 ; but for chaotic series sampled at high resolution , @xmath30 may be needed ) .
it should be noted that a relative information flow is for the comparison purpose within its own series .
the two reverse flows between two series can only be compared in terms of absolute value , since they belong to different series . in this sense ,
absolute and relative information flows should be examined simultaneously .
this is clarified in the schematic diagram in fig .
[ fig : schem_info_flow ] , and has been exemplified in the validations with two autoregressive processes .
it is quite normal that two identical information flows may differ a lot in relative importance with respect to their own series , as testified in our realistic applications . in some extreme situation
, a pair of equal flows may find one dominant but another negligible in their respective entropy balances .
partly for demonstration and partly for verification , we have presented two applications .
the first is a re - examination of the climate science problem previously studied in ref .
@xcite . considering the fadeout of the recent portentous predictions of a `` super '' or `` monster '' el nio , we have particularly focused on the predictability of el nio .
our result reconfirmed that the indian ocean sst is a source of uncertainty to the el nio prediction .
we further clarified that the information flow from the indian ocean is mainly through the indian ocean dipole ( iod ) .
another realistic problem we have examined regards the causation between a few randomly picked american stocks .
it is shown that many flows ( and hence causalities ) , though significant at a 90% level , their respective importances relative to other mechanisms are mostly negligible .
the resulting matrices of absolute and relative information flows provide us a pattern of causality mostly understandable using our common sense .
for example , ford has a larger influence on wal - mart than on cvs because people rely on motor vehicles to shop at wal - mart , while cvs could be just somewhere within a walking distance .
a particularly interesting case is that we have identified a strong one - way causality from ibm to ge during the early stage of these companies .
this has revealed to us the story of `` seven dwarfs '' competing ibm the giant for computer market . in an era
when this story has almost gone to oblivion ( one even can not find it from ge s website ) , and ge may have left us an impression that it never built any computers , let alone a series of mainframes , this finding is indeed remarkable . * acknowledgments .
* this study was partially supported by jiangsu provincial government through the `` specially - appointed professor program '' ( jiangsu chair professorship ) to xsl , and by the national science foundation of china ( nsfc ) under grant no .
41276032 .
e.g. , k. hlav@xmath167kov - schindler , m. palu@xmath168 , m. vejmelka , j. bhattacharya , physics reports , 441(1 ) , 1 - 46 ( 2007 ) ; j. pearl , _ causality : models , reasoning , and inference_. mit press , cambridge , ma , 2nd edition ( 2009 ) ; m. lungarella , k. ishiguro , y. kuniyoshi , n. otsu , international journal of bifurcation and chaos , 17(3 ) , 903 - 921 ( 2007 ) . | recently , a rigorous yet concise formula has been derived to evaluate the information flow , and hence the causality in a quantitative sense , between time series . to assess the importance of a resulting causality
, it needs to be normalized .
the normalization is achieved through distinguishing three types of fundamental mechanisms that govern the marginal entropy change of the flow recipient . a normalized or relative flow measures its importance relative to other mechanisms . in analyzing realistic series , both absolute and relative information flows need to be taken into account , since the normalizers for a pair of reverse flows belong to two different entropy balances ; it is quite normal that two identical flows may differ a lot in relative importance in their respective balances .
we have reproduced these results with several autoregressive models .
we have also shown applications to a climate change problem and a financial analysis problem .
for the former , reconfirmed is the role of the indian ocean dipole as an uncertainty source to the el nio prediction .
this might partly account for the unpredictability of certain aspects of el nio that has led to the recent portentous but spurious forecasts of the 2014 `` monster el nio '' .
for the latter , an unusually strong one - way causality has been identified from ibm ( international business machines corporation ) to ge ( general electric company ) in their early era , revealing to us an old story , which has almost gone to oblivion , about `` seven dwarfs '' competing a giant for the mainframe computer market . |
this paper investigates the fractional chromatic number of a triangle - free graph with maximum degree at most @xmath5 . for a simple ( finite ) graph @xmath0 , the fractional chromatic number of @xmath0 is the linear programming relaxation of the chromatic number of @xmath0 .
let @xmath6 be the family of independent sets of @xmath0 .
a mapping @xmath7 $ ] is called an @xmath8-fractional coloring of @xmath0 if @xmath9 and @xmath10 for each @xmath11 .
the _ fractional chromatic number _ @xmath12 of @xmath0 is the least @xmath8 for which @xmath0 has an @xmath8-fractional coloring .
alternatively , the fractional chromatic number can also be defined through multiple colorings .
a @xmath13-_fold coloring _ of @xmath0 assigns a set of @xmath13 colors to each vertex such that any two adjacent vertices receive disjoint sets of colors .
we say a graph @xmath0 is @xmath14:@xmath13- _ colorable _ if there is a @xmath13-fold coloring of @xmath0 in which each color is drawn from a palette of @xmath14 colors .
we refer to such a coloring as an @xmath14:@xmath13-coloring .
the @xmath13-_fold coloring number _ , denoted as @xmath15 , is the smallest integer @xmath14 such that @xmath0 has a @xmath14:@xmath13-coloring .
note that @xmath16 .
it is known @xcite that @xmath15 ( as a function of @xmath13 ) is sub - additive and so the @xmath17 always exists , which turns out to be an alternative definition of @xmath12 .
( moreover , @xmath12 is a rational number and the limit can be replaced by minimum . )
let @xmath18 be the chromatic number of @xmath0 and @xmath19 be the clique number of @xmath0 .
we have the following simple relation , @xmath20 now we consider a graph @xmath0 with maximum degree @xmath21 at most three . if @xmath0 is not @xmath22 , then @xmath0 is @xmath5-colorable by brooks theorem . if @xmath0 contains a triangle , then @xmath23
. equation ( [ eq:1 ] ) implies @xmath24 .
one may ask what is the possible value of @xmath12 if @xmath0 is triangle - free and @xmath21 is at most 3 ; this problem is motivated by a well - known and solved problem of determining the maximum independence number @xmath25 for such graphs .
staton @xcite showed that @xmath26 for any triangle - free graph @xmath0 on @xmath27 vertices with maximum degree at most @xmath5 .
actually , staton s bound is the best possible since the generalized petersen graph @xmath28 has 14 vertices and independence number 5 as noticed by fajtlowicz @xcite .
griggs and murphy @xcite designed a linear - time algorithm to find an independent set in @xmath0 of size at least @xmath29 , where @xmath30 is the number of components of @xmath0 that are 3-regular .
heckman and thomas @xcite gave a simpler proof of staton s bound and designed a linear - time algorithm to find an independent set in @xmath0 with size at least @xmath31 .
in the same paper @xcite , heckman and thomas conjectured @xmath32 for every triangle - free graph with maximum degree at most 3 .
note that @xcite @xmath33 provided @xmath0 is vertex transitive .
equation ( [ eq:3 ] ) implies that the generalized petersen graph @xmath28 has the fractional chromatic number @xmath34 .
thus , the conjecture is tight if it holds .
recently , hatami and zhu @xcite proved that @xmath35 , provided @xmath0 is triangle - free with maximum degree at most three .
their idea is quite clever .
for some independent set @xmath36 , the graph obtained by identifying all neighbors of @xmath36 into one fat vertex is @xmath5-colorable .
now each vertex in @xmath36 has freedom of choosing two colors .
this observation results the improvement of @xmath12 . here
we improve their result and get the following theorem .
[ t1 ] if @xmath0 is triangle - free and has maximum degree at most 3 , then @xmath37 . comparing to hatami and zhu s
result , here we only shrink the gap ( to the conjectured value @xmath38 ) by @xmath39 .
however , it is quite hard to obtain this improvement . the main idea is to extend the independent set @xmath36 of @xmath40 in lemma 12 of [ 5 ] to the admissible set @xmath41 , where @xmath42 , @xmath43 , and @xmath44 are three independent sets in @xmath40 .
however , this extension causes fundamental difficulty ( see the proof of lemma [ l:3colorable ] ) ; the reason is that in general @xmath45 ( in the proof of lemma [ l:3colorable ] ) could be 4-chromatic as shown in figure [ fig:0 ] .
[ fig:0 ] to get over the difficulty , we develop a heavy machinery of fractionally- critical graphs and use it to prove lemma [ c5 ] and [ c7 ] .
the contribution of this paper is not just an improvement on @xmath12 for triangle - free graph @xmath0 with @xmath46 .
more importantly , the theory developed in section 2 can be applied to more general scenarios concerning the fractional chromatic numbers of graphs . using the tools developed in this paper
, king - lu - peng @xcite classified all connected graphs with @xmath47 .
they @xcite further proved that @xmath48 for all graphs @xmath0 such that @xmath0 is @xmath49-free , @xmath50 , and @xmath0 is neither @xmath51 ( the square of @xmath52 ) nor @xmath53 ( the strong product of @xmath54 and @xmath55 ) as shown in figure [ fig:2 ] very recently , edwards and king @xcite improve the lower bound on @xmath56 for all @xmath57 .
the heckman - thomas conjecture is a special case at @xmath58 .
we also notice that a better bound @xmath59 ( toward heckman - thomasconjecture ) was proved in @xcite very recently .
@xmath51 @xmath53 the rest of the paper is organized as follows . in section 2
, we will study the convex structure of fractional colorings and the fractionally - critical graphs . in section 3
, we will prove several key lemmas .
in last section , we will show that @xmath0 can be partitioned into @xmath60 admissible sets and present the proof of the main theorem .
in this section , we introduce an alternative definition of `` fractional colorings . ''
the new definition highlights the convex structure of the set of all fractional colorings .
the extreme points of these `` fractional colorings '' play a central role in our proofs and seem to have independent interest .
our approach is analogous to defining rational numbers from integers . in this paper
, we use bold letter @xmath61 to represent a coloring . recall that a @xmath13-fold coloring of a graph @xmath0 assigns a set of @xmath13 colors to each vertex such that any two adjacent vertices receive disjoint sets of colors . given a @xmath13-fold coloring @xmath61 ,
let @xmath62 be the set of all colors used in @xmath61 .
two @xmath13-fold colorings @xmath63 and @xmath64 are _ isomorphic _ if there is a bijection @xmath65 such that @xmath66 . in this case
we write @xmath67 .
the isomorphic relation @xmath68 is an equivalence relation .
we use @xmath69 to denote the isomorphic class in which @xmath61 belongs to .
whenever clear under the context , we will not distinguish a @xmath13-fold coloring @xmath61 and its isomorphic class @xmath70 .
for a graph @xmath0 and a positive integer @xmath13 , let @xmath71 be the set of all ( isomorphic classes of ) @xmath13-fold colorings of @xmath0 .
for @xmath72 and @xmath73 , we can define @xmath74 as follows : for any @xmath75 , @xmath76 i.e. , @xmath77 assigns @xmath78 the disjoint union of @xmath79 and @xmath80 .
let @xmath81 .
it is easy to check that `` @xmath82 '' is commutative and associative . under the addition above , @xmath83 forms a commutative monoid with the unique @xmath84-fold coloring ( denoted by @xmath84 , for short ) as the identity . for a positive integer @xmath85 and @xmath86
, we define @xmath87 to be the new @xmath88-fold coloring by duplicating each color @xmath85 times . for @xmath72 and @xmath73 , we say @xmath63 and @xmath64 are _ equivalent _ , denoted as @xmath89 , if there exists a positive integer @xmath90 such that @xmath91 .
( this is an analogue of the classical definition of rational numbers with a slight modification .
the multiplication by @xmath90 is needed here because the division of a fractional coloring by an integer has not been defined yet .
the proofs of lemma [ l : equ ] and [ l : sim ] are straightforward and are omitted here . )
[ l : equ ] the binary relation @xmath92 is an equivalence relation over @xmath83 .
let @xmath93 be the set of all equivalence classes .
each equivalence class is called a fractional coloring of @xmath0 . for any @xmath86
, the equivalence class of @xmath61 under @xmath92 is denoted by @xmath94 .
* remark : * the notation @xmath95 makes sense only when @xmath61 is a @xmath13-fold coloring
. given any rational number @xmath96 $ ] ( with two positive integers @xmath97 and @xmath98 ) and two fractional colorings ( two equivalence classes ) @xmath99 and @xmath100 , we define the linear combination as @xmath101 the following lemma shows the definition above is independent of the choices of @xmath63 and @xmath64 , and so @xmath102 is a fractional coloring depending only on @xmath103 , @xmath99 , and @xmath100 .
[ l : sim ] for @xmath104 , let @xmath105 .
suppose that @xmath106 and @xmath107 .
for any non - negative integers @xmath97 , @xmath98 , @xmath108 , and @xmath109 satisfying @xmath110 $ ] , we have @xmath111 .
define a function @xmath112 by @xmath113 . if the graph @xmath0 is clear under the context , then
we write it as @xmath114 for short .
it is easy to check that @xmath115 does not depend on the choice of @xmath61 and so @xmath115 is well - defined .
for any @xmath116 , we define @xmath117 a fractional coloring @xmath61 is called _ extremal _ in @xmath118 if it can not be written as a linear combination of two ( or more ) distinct fractional colorings in @xmath118 .
[ tphi ] for any graph @xmath0 on @xmath27 vertices , there is an embedding @xmath119 such that @xmath120 keeps convex structure .
moreover , for any rational number @xmath121 , @xmath118 is the convex hull of some extremal fractional colorings .
* proof : * we would like to define @xmath122 as follows . given a fractional coloring @xmath95 , we can fill these colors into the regions of the general venn diagram on @xmath27-sets . for @xmath123 , we can write @xmath124 as a binary string @xmath125 such that @xmath126 for all @xmath127 . write @xmath128 and @xmath129 ( the complement set of @xmath130 ) ; then the number of colors in @xmath124-th region of the venn diagram can be written as @xmath131 by the definition , @xmath132 is additive , i.e. @xmath133 thus @xmath134 depends only on the fractional coloring @xmath95 but not on @xmath61 itself .
the @xmath124-th coordinate of @xmath135 is defined to be @xmath136 it is easy to check that @xmath137 is a well - defined function on @xmath138 . moreover , for any @xmath96 $ ] and any two fractional colorings @xmath99 and @xmath100 , we have @xmath139 thus @xmath120 keeps the convex structure .
it remains to show @xmath120 is a one - to - one mapping .
assume @xmath140 .
we need to show @xmath89 .
let @xmath141 and @xmath142 . both @xmath143 and @xmath144 are @xmath145-fold colorings .
note @xmath146 . for @xmath147 and @xmath148 ,
we denote the set of colors in the @xmath124-th venn diagram region of @xmath149 by @xmath150 .
since @xmath146 , we have @xmath151 for @xmath152 there is a bijection @xmath153 from @xmath154 to @xmath155 .
note that for @xmath156 , we have a partition of @xmath149 : @xmath157 define a bijection @xmath158 from @xmath159 to @xmath160 be the union of all @xmath153 ( @xmath148 ) .
we have @xmath161 thus @xmath162 .
note that @xmath163 and @xmath164 . we conclude that @xmath89 . under the embedding ,
@xmath165 consists of all rational points in a polytope defined by the intersection of finite number of half spaces .
note that all coefficients of the equations of hyperplanes are rational .
each rational point in the polytope corresponds to a fractional coloring while each vertex of the polytope corresponds to an extremal fractional coloring .
@xmath166 * remark : * it is well - known that for any graph @xmath0 there is a @xmath14:@xmath13-coloring of @xmath0 with @xmath167 . in our terminology
, we have @xmath168 let @xmath169 be a subgraph of @xmath0 .
a @xmath13-fold coloring of @xmath0 is naturally a @xmath13-fold coloring of @xmath169 ; this _ restriction _ operation induces a mapping @xmath170 .
it is easy to check that @xmath171 keeps convex structure , i.e. , for any @xmath172 and @xmath173\cap \mathbb{q}$ ] , we have @xmath174 it is also trivial that @xmath175 now we consider a reverse operation .
we say a fractional coloring @xmath176 is _ extensible _ in @xmath177 if there is a fractional coloring @xmath178 satisfying @xmath179 we say a fractional coloring @xmath176 is _ fully extensible _ in @xmath138 if it is extensible in @xmath177 , where @xmath180 .
( it also implies that @xmath99 is extensible in @xmath177 for all @xmath181 . )
[ fully ] let @xmath169 be a subgraph of @xmath0 and @xmath182 for @xmath183 .
assume that for @xmath184 , @xmath185 is fully extensible in @xmath138 .
for any @xmath186 $ ] , we have @xmath187 is fully extensible in @xmath138 .
* proof : * let @xmath188 for @xmath189 .
note that there are fractional colorings @xmath190 such that @xmath191 for @xmath184 .
let @xmath192 .
we have @xmath193 and , @xmath194 note that @xmath195 therefore , @xmath196 is fully extensible in @xmath138 by the definition .
@xmath166 we say @xmath197 if @xmath198 and @xmath199 . similarly , we say @xmath200 if @xmath201 and @xmath202 .
[ glue ] let @xmath0 be a graph .
assume that @xmath203 and @xmath204 are two subgraphs such that @xmath205 and @xmath206 .
if two fractional colorings @xmath207 and @xmath208 satisfy @xmath209 , then there exists a fractional coloring @xmath210 satisfying @xmath211 for @xmath184 and @xmath212 .
* proof : * without loss of generality , we can assume @xmath213 ( by taking the least common multiple if it is necessary . )
we also assume @xmath214 , then we have @xmath215 , there is a bijection @xmath120 from @xmath216 to @xmath217 .
extend @xmath120 as an one - to - one mapping from @xmath218 to @xmath219 in an arbitrary way .
now we define a @xmath13-fold coloring @xmath61 of @xmath0 as follows .
@xmath220 since @xmath203 and @xmath204 cover all edges of @xmath0 , @xmath61 is well - defined .
note @xmath221 and @xmath222 .
thus for @xmath184 , we have @xmath223 we also have @xmath224 . @xmath166
[ extension ] let @xmath0 be a graph . assume that @xmath203 and @xmath204 are two subgraphs such that @xmath205 and @xmath206 .
if @xmath225 and any extreme fractional coloring in @xmath226 is extensible in @xmath227 , then we have @xmath228 .
* proof : * there is a fractional coloring @xmath229 satisfying @xmath230 . since every extreme fractional coloring of @xmath226 is extensible in @xmath227 ,
there exist fractional colorings @xmath231 such that @xmath232 are all extreme fractional colorings in @xmath226 .
the fractional coloring @xmath233 can be written as a linear combination of the extreme ones .
hence , there exist @xmath234 $ ] such that @xmath235 and @xmath236 let @xmath237 .
we have @xmath238 applying lemma [ glue ] , there exists a fractional coloring @xmath239 .
therefore , we have @xmath228 .
@xmath166 [ cor1 ] let @xmath0 be a graph .
assume that @xmath203 and @xmath204 are two subgraphs such that @xmath205 and @xmath240 for some positive integer @xmath8 .
we have @xmath241 * proof : * without loss of generality , we assume @xmath242 . let @xmath243 .
we have @xmath244 as @xmath203 contains @xmath245 . since @xmath246
, we have @xmath247 . since @xmath169 is a complete graph
, @xmath226 contains only one fractional coloring ; namely , color all vertices of @xmath169 using distinct colors .
it is trivial that any extreme fractional coloring in @xmath226 is extensible in @xmath248 .
applying theorem [ extension ] , we have @xmath228 .
the other direction is trivial .
@xmath166 let @xmath249 be a non - edge of a graph @xmath204 .
we denote @xmath250 to be the supergraph of @xmath204 by adding the edge @xmath249 and denote @xmath251 be the quotient graph by identifying the vertex @xmath252 and the vertex @xmath78 .
[ l : cut2 ] let @xmath0 be a graph .
assume that @xmath203 and @xmath204 are two subgraphs such that @xmath205 and @xmath253 . 1 .
if @xmath249 is an edge of @xmath0 , then we have @xmath254 2 .
if @xmath249 is not an edge of @xmath0 , then we have @xmath255 * proof : * part 1 is a simple application of corollary [ cor1 ] . for the proof of part 2 , let @xmath256 .
note @xmath257 .
let @xmath258 be the empty graph on the set of two vertices @xmath252 and @xmath78 .
all fractional colorings of @xmath259 can be represented by the following weighted venn diagram , see figure [ ven ] .
the parameter @xmath90 measure the fraction of common colors shared by @xmath252 and @xmath78 .
there are two extreme points in the convex hull : @xmath260 and @xmath261 .
the extremal fractional coloring corresponding to @xmath260 is extensible in @xmath227 since @xmath262 .
the extremal fractional coloring corresponding to @xmath261 is extensible in @xmath227 since @xmath263 . applying theorem [ extension ] and lemma [ fully ] , we have @xmath228 .
part 2 is proved .
@xmath166 in this subsection , we will apply our machinery to triangle - free graphs with maximum degree at most @xmath5 .
recall that a graph @xmath0 is _
( for a positive integer @xmath30 ) if @xmath264 and @xmath265 for any proper subgraph @xmath169 of @xmath0 .
for any rational number @xmath257 , a graph @xmath0 is _ @xmath85-fractionally - critical _ if @xmath266 and @xmath267 for any proper subgraph @xmath169 of @xmath0 . for simplicity ,
we say @xmath0 is _ fractionally - critical _ if @xmath0 is @xmath12-fractionally - critical .
we will study the properties of fractionally - critical graphs .
the following lemma is a consequence of corollary [ cor1 ] .
assume that @xmath0 is a fractionally - critical graph with @xmath268 .
we have @xmath0 is @xmath269-connected . moreover , if @xmath0 has a vertex - cut @xmath270 , then @xmath249 is not an edge of @xmath0 . for any vertex @xmath252 of a graph @xmath0 and a positive integer @xmath124
, we define + @xmath271 .
[ c5 ] assume that @xmath0 is a fractionally - critical triangle - free graph satisfying @xmath272 and @xmath273 .
for any vertex @xmath274 and any @xmath275-cycle @xmath276 of @xmath0 , we have either @xmath277 or @xmath278 . *
proof : * let @xmath279 .
we have @xmath280 .
we will prove the statement by contradiction .
suppose that there is a vertex @xmath281 and a @xmath275-cycle @xmath276 satisfying @xmath282 and @xmath283 . combined with the fact @xmath0 being triangle - free , we have the following two cases .
* case 1 : * @xmath284 .
since @xmath46 and @xmath0 is triangle - free , it is easy to check that @xmath0 contains the following subgraph @xmath285 as shown in figure [ fig : c5g9 ] .
since @xmath0 is @xmath269-connected , @xmath285 is the entire graph ( in @xcite ) .
thus @xmath286 .
contradiction ! * case 2 : * @xmath287 and there exists one vertex of @xmath276 having distance of @xmath5 to @xmath281 . hatami and
zhu @xcite showed that @xmath0 contains one of the five graphs in figure [ fig : c5s ] as a subgraph .
( note that some of the marked vertices @xmath252 , @xmath78 , and @xmath288 may be missing or overlapped ; these degenerated cases result in a smaller vertex - cut , and can be covered in a similar but easier way ; we will discuss them at the end of this proof . )
\(i ) ( ii ) ( iii ) \(iv ) ( v ) if @xmath0 contains a subgraph of type ( i ) , ( ii ) , or ( iii ) , then @xmath0 has a vertex - cut @xmath289 .
let @xmath203 and @xmath204 be the two connected subgraphs of @xmath0 such that @xmath290 , @xmath291 , and @xmath292 . in all three cases ,
@xmath250 and @xmath251 are @xmath293-colorable .
please see figure [ fig : c5s2 ] .
@xmath251 in ( i ) @xmath251 in ( ii ) @xmath251 in ( iii ) @xmath250 in ( i ) @xmath250 in ( ii ) @xmath250 in ( iii ) applying lemma [ l : cut2 ] , we have @xmath294 since @xmath295 , we must have @xmath296 , which is a contradiction to the assumption that @xmath0 is fractionally - critical .
if @xmath0 contains one of the subgraphs ( iv ) and ( v ) , then @xmath0 has a vertex - cut set @xmath297 as shown in figure [ fig : c5s ] .
let @xmath203 and @xmath204 be the two connected subgraphs of @xmath0 such that @xmath290 , @xmath298 , and @xmath292 .
we shall show @xmath299 .
suppose not .
we can assume @xmath300 , where @xmath301 . by theorem [ tphi ] ,
every fractional coloring in @xmath302 can be represented by a rational point in a convex polytope @xmath303 . from now on
, we will not distinguish the rational point in the convex polytope and the fractional coloring .
note the convex polytope for @xmath302 can be parametrized as @xmath304 see the weighted venn diagram in figure [ fig : venn1 ] .
the extreme fractional colorings of @xmath302 are represented ( under @xmath120 , see theorem [ tphi ] ) by : ( a ) : : @xmath305 and @xmath261 .
( b ) : : @xmath306 and @xmath307 .
( c ) : : @xmath308 and @xmath309 .
( d ) : : @xmath310 and @xmath311 .
( e ) : : @xmath312 and @xmath307 .
( f ) : : @xmath313 and @xmath309 . (
g ) : : @xmath314 and @xmath311 .
we will show that all 7 extreme fractional colorings are extensible in @xmath315 .
( a ) : : let @xmath316 be the quotient graph by identifying @xmath252 , @xmath78 , and @xmath288 as one vertex .
the fractional coloring @xmath317 is extensible in @xmath315 if and only if @xmath318 , which is verified by figure [ fig : uvw ] .
+ + @xmath316 in ( iv ) @xmath316 in ( v ) ( b ) : : let @xmath319 be the graph obtained by identifying @xmath78 and @xmath288 as one vertex @xmath320 followed by adding an edge @xmath321 .
the fractional coloring @xmath322 is extensible in @xmath315 if and only if @xmath323 , which is verified by figure [ fig : u - vw ] .
+ + @xmath319 in ( iv ) @xmath319 in ( v ) ( c ) : : let @xmath324 be the graph obtained by identifying @xmath252 and @xmath288 as one vertex @xmath325 followed by adding an edge @xmath326 .
the fractional coloring @xmath327 is extensible in @xmath315 if and only if @xmath328 , which is verified by figure [ fig : v - uw ] .
+ + @xmath324 in ( iv ) @xmath324 in ( iv ) ( d ) : : let @xmath329 be the graph obtained by identifying @xmath252 and @xmath78 as one vertex @xmath249 followed by adding an edge @xmath330 .
the fractional coloring @xmath331 is extensible in @xmath315 if and only if @xmath332 , which is verified by figure [ fig : w - uv ] .
+ + @xmath329 in ( iv ) @xmath329 in ( v ) ( e ) : : choose @xmath333 .
since @xmath334 , we have @xmath335 . note that @xmath336 observe that @xmath337 being fully extensible in @xmath338 implies that @xmath337 is extensible in @xmath315 . to show @xmath337 is fully extensible in @xmath338 , it suffices to show both @xmath339 and @xmath340 are fully extensible in @xmath338 by lemma [ fully ] ( see figure [ fig : f0 ] and [ fig : f1 ] ) .
+ + \(iv ) ( v ) + + \(iv ) ( v ) ( f ) : : similarly , to show @xmath341 is extensible in @xmath315 , it suffices to show @xmath339 and @xmath342 are fully extensible in @xmath338 ( see figure [ fig : f0 ] and [ fig : f2 ] ) .
+ + \(iv ) ( v ) ( g ) : : similarly , to show @xmath343 is extensible in @xmath315 , it suffices to show @xmath339 and @xmath344 are fully extensible in @xmath338 ( see figure [ fig : f0 ] and [ fig : f3 ] ) .
+ + \(iv ) ( v ) applying theorem [ extension ] , we have @xmath345 , which is a contradiction and so @xmath299 .
however , @xmath0 is fractionally - critical .
contradiction !
now we consider the degenerated cases . for graphs ( i ) , ( ii ) , and ( iii ) , @xmath270 is degenerated into a set of size one .
since @xmath0 is @xmath269-connected , @xmath0 is a subgraph of one of graphs listed in figure [ fig : c5s2 ] .
thus @xmath0 is 8:3-colorable . contradiction ! for graph ( iv ) and
( iv ) , @xmath346 is degenerated into a set @xmath169 of size at most @xmath269 . if @xmath347 , then @xmath0 is a subgraph of one of graphs in figure [ fig : uvw ] . if @xmath348 , then @xmath349 and @xmath350 are subgraphs of graphs from figure [ fig : uvw ] to figure [ fig : w - uv ] .
applying lemma [ l : cut2 ] , we get @xmath351 contradiction ! hence , lemma [ c5 ] follows .
@xmath166 [ c7 ] assume that @xmath0 is a fractionally - critical triangle - free graph satisfying @xmath272 and @xmath352 .
for any vertex @xmath353 and any @xmath354-cycle @xmath276 of @xmath0 , we have @xmath355 .
* proof : * we prove the statement by contradiction .
suppose that there is a vertex @xmath281 and a @xmath354-cycle @xmath276 satisfying @xmath356 .
recall that @xmath357 .
we have @xmath358 .
combined with the fact that @xmath0 is triangle - free and 2-connected , @xmath0 must be one of the following graphs in figure [ fig : c7 ] .
all of them are @xmath293-colorable , see figure [ fig : c7 ] .
! @xmath166
the following approach is similar to the one used in @xcite .
a significant difference is a new concept `` admissible set '' .
basically , it replaces the independent set @xmath36 ( of @xmath40 in @xcite ) by three independent sets @xmath359 .
recall that @xmath360 if there exists a @xmath249-path of length @xmath124 in @xmath0 .
a set @xmath361 is called _ admissible _ if @xmath36 can be partitioned into three sets @xmath42 , @xmath43 , @xmath44 satisfying 1 . if @xmath362 for some @xmath363 , then @xmath364 .
2 . if @xmath365 and @xmath366 for some @xmath124 and @xmath367 satisfying @xmath368 , then @xmath369 .
the following key theorem connects the @xmath12 to a partition of @xmath0 into admissible sets .
[ t2 ] assume that @xmath0 is a fractionally - critical triangle - free graph satisfying @xmath272 and @xmath370 . if @xmath371 can be partitioned into @xmath30 admissible sets , then @xmath372 let @xmath373 be an admissible set . inspired by the method used in @xcite , we define an auxiliary graph @xmath374 as follows .
we use the notation @xmath375 to denote the neighborhood of @xmath36 in @xmath0 . for each @xmath363 , let @xmath376 . by admissible conditions , @xmath377 , @xmath378 , and @xmath379 are all independent sets of @xmath0 .
let @xmath380 be a graph obtained from @xmath0 by deleting @xmath36 ; identifying each @xmath381 as a single vertex @xmath382 for @xmath383 ; and adding three edges @xmath384 , @xmath385 , @xmath386 .
we have the following lemma , which will be proved later . [ l:3colorable ]
assume that @xmath0 is fractionally - critical triangle - free graph satisfying @xmath272 and @xmath273 .
let @xmath36 be an admissible set of @xmath0 and @xmath45 be the graph defined as above .
we have @xmath45 is @xmath5-colorable .
* proof of theorem [ t2 ] : * assume @xmath0 can be partitioned into @xmath30 admissible sets , say @xmath387 , where @xmath388 . for each @xmath389 and @xmath390 ,
let @xmath391 . from the definition of an admissible set and @xmath0 being triangle - free , we have @xmath392 is an independent set for all @xmath389 and @xmath393 . by lemma [ l:3colorable ] ,
@xmath394 is @xmath5-colorable for all @xmath395 .
let @xmath396 be a 3-coloring of @xmath394 with the color set @xmath397 . here
all the colors @xmath398 s are pairwise distinct .
we use @xmath399 to denote the set of all subsets of @xmath400 .
we define @xmath401 satisfying @xmath402 note that for a fixed @xmath403 , @xmath404 denotes the vertex of @xmath394 obtained from contracting @xmath392 for @xmath393 .
observe that each vertex in @xmath405 receives two colors from @xmath406 and every other vertex receives one color .
it is clear that any two adjacent vertices receive disjoint colors .
let @xmath407 be a mapping defined as @xmath408 .
now @xmath409 is a @xmath410-fold coloring of @xmath0 such that each color is drawn from a palette of @xmath411 colors .
thus we have @xmath412 .
we completed the proof of theorem [ t2 ] .
@xmath166 before we prove lemma [ l:3colorable ] , we first prove a lemma on coloring the graph obtained by splitting the hub of an odd wheel .
let @xmath413 be the set of vertices of an odd cycle @xmath414 in a circular order .
let @xmath415 .
we construct a graph @xmath169 as follows : 1 .
2 . @xmath417 .
each @xmath418 is adjacent to exactly one element of @xmath419 .
4 . @xmath420 .
5 . @xmath169 can have at most one vertex ( of @xmath421 , @xmath422 , and @xmath423 ) with degree @xmath269 .
the graph @xmath169 can be viewed as splitting the hub of the odd wheel into @xmath5 new hubs where each spoke has to choose one new hub to connect , then connect all new hubs .
one special case it that one of the new hubs has no neighbor in @xmath276 .
[ lemma1 ] a graph @xmath169 constructed as described above is @xmath5-colorable . * proof : * without loss of generality , we assume @xmath424 and @xmath425 . we construct a proper @xmath5-coloring @xmath426 of @xmath169 as follows .
first , let @xmath427 , @xmath428 , and @xmath429 .
again , without loss of generality , we assume @xmath430 .
the neighbors of @xmath421 divide @xmath431 into several intervals .
the vertices in each interval are either connecting to @xmath422 or @xmath423 but not connected to @xmath421 . as @xmath78 goes through each interval counter - clockwisely , we list the neighbor of @xmath78 ( in @xmath419 ) and get a sequence consisting of @xmath422 @xmath423 .
then we delete the repetitions of @xmath432 and @xmath433 in the sequence .
there are @xmath434 types of intervals based on the result seqeunce : type i : @xmath435 .
type ii : @xmath436 .
type iii : @xmath437 .
type iv : @xmath438 .
if @xmath423 is a vertex of degree two in @xmath169 , then the interval @xmath439 has only one type , which is degenerated into @xmath422 .
given an interval @xmath439 , let @xmath440 ( or @xmath441 ) be the common neighbor of @xmath421 and the left ( or right ) end of @xmath439 respectively .
we color @xmath440 and @xmath441 first and then try to extend it as a proper coloring of @xmath439 .
sometimes we succeed while sometimes we fail .
we ask a question whether we can always get a proper coloring .
the answer depends only on the type of @xmath439 and the coloring combination of @xmath440 and @xmath441 . in table
[ tab:1 ] , the column is classified by the coloring combination of @xmath440 and @xmath441 , while the row is classified by the types of @xmath439 . here `` yes '' means the coloring process always succeeds , while `` no '' means it sometimes fails .
.can a coloring be extended to @xmath439 properly ? [ cols="^,^,^,^,^",options="header " , ] an ending vertex @xmath288 of an interval @xmath439 is called a _ free _
end if @xmath288 s two neighbors outside @xmath439 receiving the same color .
observe that in all yes entries , there exist at least one free end .
note that each vertex on @xmath439 has degree @xmath5 .
we can always color the vertices of @xmath439 greedily starting from the end not equaling @xmath288 . since @xmath288 is a free end
, there is no difficulty to color @xmath288 at the end .
now we put them together .
we color the neighbors of @xmath421 one by one counter - clockwise starting from @xmath442 according to the following rules : 1 .
when we meet an interval @xmath439 of type ii or iii , we keep the colors of @xmath440 and @xmath441 the same .
2 . when we meet an interval @xmath439 of type i or iv , we keep the colors of @xmath440 and @xmath441 different .
there are two possibilities .
if the last interval obeys the rules , then by table [ tab:1 ] , we can extend the partial coloring into a proper 3-coloring of @xmath169 .
if the last interval does not obey the rules , then we swap the colors @xmath269 and @xmath5 of the neighbors of @xmath421 . by table
[ tab:1 ] , the new partial coloring can be extended into a proper 3-coloring of @xmath169 .
we completed the proof .
@xmath166 a maximal @xmath269-connected subgraph @xmath443 of a graph is called a _ block _ of @xmath0 . a _ gallai tree _ is a connected graph in which all blocks are either complete graphs or odd cycles .
gallai forest _ is a graph all of whose components are gallai trees .
a @xmath30-_gallai tree ( forest ) _ is a gallai tree ( forest ) such that the degree of all vertices are at most @xmath444 .
@xmath30-critical graph _ is a graph @xmath0 whose chromatic number is @xmath30 and the chromatic number of any proper subgraph is strictly less than @xmath30 .
gallai showed the following lemma .
[ lemma3 ] if @xmath0 is a @xmath30-critical graph , then the subgraph of @xmath0 induced on the vertices of degree @xmath444 is a k - gallai forest .
now , we are ready to prove lemma [ l:3colorable ] . *
proof of lemma [ l:3colorable ] : * write @xmath374 for short .
note that the only possible vertices of degree greater than @xmath5 in @xmath380 are @xmath421 , @xmath422 , and @xmath423 .
we can color @xmath421 , @xmath422 , and @xmath423 by @xmath445 , @xmath269 , and @xmath5 , respectively .
since the remaining vertices have degree at most @xmath5 , we can color @xmath380 properly with @xmath434 colors greedily , i.e. , @xmath446 .
suppose that @xmath380 is not @xmath5-colorable .
let @xmath169 be a 4-critical subgraph of @xmath380 .
we have @xmath447 for all @xmath448 except for possible @xmath421 , @xmath422 , and @xmath423 .
let @xmath449 be the subgraph induced by all @xmath448 such that @xmath447 ; then @xmath449 is not empty since @xmath450 . by lemma [ lemma3 ]
the subgraph of @xmath169 induced by @xmath449 is a @xmath434-gallai forest .
we known @xmath449 may contain one or more vertices in @xmath451 .
let @xmath452 .
observe that any induced subgraph of a @xmath434-gallai forest is still a @xmath434-gallai forest and so the subgraph of @xmath169 induced by @xmath453 is also a @xmath434-gallai forest .
recall the definition of an admissible set . if @xmath365 and @xmath454 for some @xmath124 and @xmath367 satisfying @xmath368 , then @xmath455 , which implies that any vertex @xmath281 in @xmath453 can have at most one neighbor in @xmath456 .
note @xmath457 .
we have @xmath458 .
let @xmath443 be a leaf block in the gallai - forest @xmath453 ; then @xmath443 is a complete graph or an odd cycle by the definition of the gallai - forest .
observe @xmath443 can not be a single vertex or @xmath55 since every vertex in @xmath443 has at least two neighbors in @xmath453 . as @xmath0 is triangle - free , then @xmath443 must be an odd cycle @xmath459 with @xmath460 .
case ( a ) : : : @xmath461 .
since @xmath169 is @xmath434-critical , @xmath462 is @xmath5-colorable .
let @xmath426 be a proper @xmath5-coloring of @xmath462 .
since @xmath462 contains a triangle @xmath463 , we have @xmath421 , @xmath422 , and @xmath423 receive different colors .
we have @xmath464 by lemma [ lemma1 ] , we can extend the coloring @xmath426 to all vertices on @xmath443 as well .
thus @xmath169 is @xmath5-colorable .
contradiction . case ( b ) : : : @xmath465 .
since @xmath443 is a leaf block , there is at most one vertex , say @xmath466 , which is connected to another block in @xmath453 .
list the vertices of @xmath443 in a circular order as @xmath467 .
all @xmath468 connect to one @xmath382 , say @xmath421 , which implies that for @xmath469 , there exists a vertex @xmath470 and a vertex @xmath471 so that @xmath472-@xmath473-@xmath418 form a path of length @xmath269 . since @xmath0 is triangle - free
, we have @xmath474 for all @xmath475 .
note that @xmath418-@xmath473-@xmath472-@xmath476-@xmath477-@xmath478 forms a path of length @xmath275 unless @xmath479 .
+ recall the admissible conditions : if @xmath362 for some @xmath363 , then @xmath480 .
we must have @xmath481 .
denote this common vertex by @xmath281 .
now have @xmath482 note @xmath483 .
we have @xmath484 .
the possible values for @xmath8 are @xmath269 and @xmath5 .
if @xmath485 , then @xmath443 is a @xmath275-cycle .
since @xmath443 is in @xmath453 , we have @xmath486 ; this is a contradiction to lemma [ c5 ] .
if @xmath487 , then @xmath443 is a @xmath354-cycle ; this is a contradiction to lemma [ c7 ] .
the proof of lemma [ l:3colorable ] is finished .
[ t5 ] let @xmath0 be a triangle - free graph with maximum degree at most @xmath5 .
if @xmath0 is @xmath269-connected and @xmath488 , then @xmath0 can be partitioned into at most @xmath60 admissible sets .
* proof of theorem [ t5 ] : * we will define a proper coloring @xmath489 such that for @xmath490 , the @xmath124-th admissible set is @xmath491 .
we refer to @xmath492 as a color block for all @xmath493 .
since @xmath494 , there is a cycle @xmath276 of length @xmath434 , @xmath275 , or @xmath495 .
let @xmath496 and @xmath497 be a pair of adjacent vertices of @xmath276 .
we assume that @xmath498 is connected .
( if not , we start the greedy algorithm below from each of the components of @xmath499 separately . )
we can find a vertex @xmath500 other than @xmath496 and @xmath501 such that @xmath502 is connected .
inductively , for each @xmath503 , we can find a vertex @xmath504 other than @xmath496 and @xmath501 such that @xmath505 is connected .
therefore , we get an order of vertices @xmath506 such that for @xmath507 , the induced graph on @xmath508 is connected . 1 . for each @xmath512 , we have @xmath511 is not in the same block of @xmath513 .
2 . for each @xmath514
, we have @xmath515 .
3 . for each @xmath516
, we have @xmath511 could equal @xmath513 but not equal the other two colors in the color block of @xmath513 . for @xmath517 , there is at least one vertex in @xmath518 and one vertex in @xmath519 still uncolored .
thus @xmath520 , + @xmath521 , @xmath522 , @xmath523 , and @xmath524 . since @xmath525 it is always possible to color the vertex @xmath510 properly .
case @xmath526 : : : for any vertex @xmath78 on @xmath526 , there are two vertices in @xmath527 .
we have @xmath528 and @xmath529 .
thus the number of colors forbidden to be assigned to @xmath78 is at most @xmath530 case @xmath54 : : : for any vertex @xmath78 on @xmath54 , there are two vertices in @xmath531 .
we also have @xmath532 .
the number of colors forbidden to be assigned to @xmath78 is at most @xmath533 case @xmath534 : : : for any vertex @xmath78 on @xmath534 , there are two vertices in @xmath535 and two vertices in @xmath536 .
we have @xmath537 .
thus the number of colors forbidden to be assigned to @xmath78 is at most @xmath538 the 42 admissible sets can be obtained from the coloring @xmath426 as follows . for @xmath539 ,
the @xmath367-th admissible set has the following partition @xmath540 those are admissible sets by the construction of the coloring @xmath426 . @xmath166 * proof of theorem [ t1 ] : * suppose that there exists a graph @xmath0 which is triangle - free , @xmath541 , and @xmath542 .
without loss of generality , we can assume @xmath0 has the smallest number of edges among all such graphs .
thus @xmath0 is @xmath269-connected and fractionally - critical .
if @xmath543 , hatami and zhu @xcite showed @xmath544 . contradiction !
99 o. borodin and a. kostochka , on an upper bound on a graph s chromatic number , depending on the graph s degree and density , _ j. comb .
b , _ * 23 * ( 1977 ) 247 - 250 .
k. edwards and a. d. king , bounding the fractional chromatic numrber of @xmath547-free graphs , http://arxiv.org/abs/1206.2384 .
s. fajtlowicz , on the size of independent sets in graphs , in _ proceedings of the ninth southeastern conference on combinatorics , graph theory and computing_(florida atlantic univ , boca raton , fl,1978 ) , congress , numer , xxi , utilitas math , winnipeg , mb , 1978 , 269 - 274 , t. gallai , kritische graphen i , _ magyar tud .
akad . mat .
kutat int .
kzl _ , 8 ( 1963 ) , pp .
165 - 192 , j. griggs and o. murphy , edge density and independence ratio in triangle - free graphs with maximum degree three , _ discrete math .
_ , * 152 * ( 1996 ) , 157 - 170 , h. hatami and x. zhu , the fractional chromatic number of graphs of maximum degree at most three , _ siam j. discrete math . ,
_ * 24 * ( 2009 ) , 1762 - 1775 . c. c. heckman and r. thomas , a new proof of the independence ratio of triangle - free cubic graphs ,
_ discrete math .
_ , * 233 * ( 2001 ) , 233 - 237 , a. johansson , asymptotic choice number for triangle free graphs , _ dimacs technical report , _ * 91 - 5*. a. d. king , l. lu , and x. peng , a fractional analogue of brooks theorem , _ siam . j. discrete math . , _ * 26 * -2 ( 2012 ) , 452 - 471 .
m. molloy and b. reed , _ graph colouring and the probabilistic method _ , volume 23 of algorithms and combinatorics .
springer - verlag , berlin , 2002 .
b. reed , @xmath548 and @xmath549 , _ j. of graph theory _ ,
vol * 27 * -4 ( 1998 ) , 177 - 227 .
e. r. scheinerman and d. h. ullman , _ fractional graph theory_. a rational approach to the theory of graphs , wiley - intersci .
ser . discrete math .
optim , john wiley & sons , inc , new york , 1997 , w. staton , some ramsey - type numbers and the independence ration , _ trans .
_ , * 256 * ( 1979 ) , 353 - 370 , | let @xmath0 be a triangle - free graph with maximum degree at most 3 .
staton proved that the independence number of @xmath0 is at least @xmath1 .
heckman and thomas conjectured that staton s result can be strengthened into a bound on the fractional chromatic number of @xmath0 , namely @xmath2 . recently ,
hatami and zhu proved that @xmath3 . in this paper , we prove @xmath4 . |
all stars with initial masses @xmath5 are expected to end their lives as core - collapse supernovae ( sne ) or , possibly , disappearance due to direct infall onto black holes during core - collapse @xcite .
recent efforts to detect sn progenitors have been successful for type iip sne , confirming that the progenitors of these particular sne to be red supergiants ( rsgs ) with initial masses @xmath6 @xcite .
the progenitors of the h - deficient type ibc sne have , however , proved difficult to detect .
stellar evolution models predict that the progenitors for these types of sne are either very massive single stars ( @xmath7 ) , which undergo extreme levels of mass loss during phases as luminous blue variables and wolf - rayet ( wr ) stars ( removing their outer hydrogen layers ) , and lower mass stars in binaries , which are stripped of their outer hydrogen layers by mass transfer onto a binary companion . the progenitors for these types of sne are expected to be blue , compact stars with high effective temperatures .
such stars lie on the blue - side of the hertzsprung - russell ( hr ) diagram , whereas the rsg progenitors of type iip sne occupy the low effective temperature , red side of the hr diagram .
the review of pre - explosion observations of 12 type ibc sne presented by @xcite concluded that , statistically , it was unlikely that all could have arisen from single massive wr stars ; suggesting a significant contribution to the type ibc sn rate from the binary progenitor channel . for the study by @xcite the case of sn
2002ap was particularly important , with the deep pre - explosion images effectively ruling out all massive single stars @xcite .
similar conclusions , based on the ratio of type ibc sn to h - rich sne and the slope of the initial mass function ( imf ) , suggested that type ibc sne were too common to just arise from single massive stars @xcite .
more recently , the progenitor of the type ibc sn iptf 13bvn has been authoritatively detected in pre - explosion observations .
while some analyses of the properties of the pre - explosion progenitor candidate suggest a possible single massive progenitor star , other studies of the pre - explosion source and models of the properties of the sn light curve are all consistent with a lower mass progenitor that was stripped through interaction with a binary companion .
there have been some suggestions of bright , massive progenitor stars detected in pre - explosion observations of interacting type iin sne , such as 2005gl @xcite , 2010jl @xcite and the possible sn 2009ip @xcite .
limitations in the nature of the pre - explosion imaging for these sne , however , make it difficult to draw firm conclusions about the nature of their progenitors .
such type iin sne are , however , rare and the fate of the majority of massive stars with @xmath8 remains uncertain . here
we report the results of late - time observations ( @xmath9 years post - explosion ) of the site of the type ic sn 2007gr in the galaxy ngc 1058 in an effort to gather further information about the nature of the progenitor star .
pre - explosion observations of the site of sn 2007gr , in two hubble space telescope ( hst ) wide field planetary camera 2 ( wfpc2 ) bands , were reported by @xcite .
@xcite did not find a source coincident with the sn position , but instead identified a nearby bright source as a possible host cluster and determined a mass estimate for the progenitor based on the age of such cluster in comparison with the lifetimes of possible progenitors .
the properties of this candidate host cluster were further analysed with additional hst wfpc2 observations by @xcite .
we adopt a distance modulus to ngc 1058 of @xmath10 @xcite , which was also used in the previous analyses of @xcite and @xcite , and an explosion epoch of 2007 aug 13 ( jd @xmath11 ; ) . applying the o3n2 metallicity measure @xcite to the measured strengths of emission lines of h ii regions in ngc 1058 @xcite
, we estimate the oxygen abundance at the deprojected radius of sn 2007gr of @xmath12 to be approximately solar .
a log of the pre - explosion , post - explosion and late - time hst observations of the site of sn 2007gr is presented in table [ tab : obs : log ] .
the hst wfpc2 observations were retrieved from the stsci archive , having been processed through the on - the - fly recalibration pipeline .
pairs of exposures for each filter , at each epoch , were combined , for display purposes , using the pyraf task _
astrodrizzle_. photometry of the wfpc2 frames was conducted using the dolphot package @xcite with the specific wfpc2 module .
the iraf task _ crrej _ was used to create a cosmic ray mask , which enabled the removal of cosmic rays from the pairs of exposures for each filter prior to the application of dolphot . for the post - explosion ( @xmath13d post - explosion ) and late - time ( @xmath14d post - explosion ) wfpc2 observations , the sn was imaged on the planetary camera ( pc ) chip , with pixel scale 0.046 arcsecs , while for the pre - explosion observations ( @xmath15d pre - explosion ) the sn position fell on the wide field 2 ( wf2 ) chip with pixel scale 0.0996 arcsec .
later wfc3/uvis observations ( @xmath16d post - explosion ) of the site of sn 2007gr were acquired using a 4-point box dither pattern for each of the three filters used . the individual distorted images were retrieved from the stsci archive and combined and corrected for geometric distortion using the _ astrodrizzle _ package . due to the nature of the dither pattern , it was possible to resample the image to a finer pixel scale of 0.02 arcsecs ( compared to the original wfc3/uvis pixel scale of 0.04 arcsecs ) .
photometry of the wfc3 frames was also conducted using dolphot , with the wfc3 specific module , using the drizzled , distortion corrected images as references . in both the late - time wfpc2 and wfc3 frames , detection limits for those stars not detected at certain wavelengths
were established through artificial star tests .
artificial stars were inserted at the positions determined for the stars at wavelengths in which the stars were detected .
the completeness of the recovery of the artificial stars ( at @xmath17 , within 1 pixel of the inserted position ) as a function of input magnitude was modelled by a cumulative normal distribution , parameterised by the magnitude at which 50% completeness was achieved . given the complexity of the region hosting sn 2007gr
, the derived detection thresholds were found to vary significantly across the field due to varying background levels and stellar crowding ; therefore , detection thresholds were calculated independently for each star .
artificial star tests at random locations around the position of sn 2007gr were also used to calculate corrections to the uncertainty on the photometry due to crowding effects .
sn 2007gr was still bright in the post - explosion wfpc2 observations , and the position of the sn in the f450w image , relative to surrounding , nearby stars , was used to determine the sn position in the pre - explosion wfpc2 and late - time wfpc2 and wfc3 observations using the iraf task _
geomap_. the post - explosion wfpc2 f450w image was used to determine the sn position in the pre - explosion wfpc2 f450w observation with an accuracy of 37 milliarcsecs .
geometric transformations were determined between the post - explosion wfpc2 f450w image and the late - time wfpc2 and wfc3 f555w observations .
the sn position was identified on the late - time wfpc2 and wfc3 images to within 15 and 11 milliarcsecs , respectively .
small shifts between the late - time wfpc2 and wfc3 observations taken with other filters were calculated using the pyraf task _ crosscor _ and were generally found to be small ( @xmath18pix ) , with associated uncertainties that were significantly smaller . lccccl date & instrument & filter & exposure & pixel + & & & time ( s ) & scale ( @xmath19 ) + + 2001 jul 3.72@xmath20 & wfpc2/wf2 & f450w & 460 & 0.097 + 2001 jul 3.74@xmath20 & wfpc2/wf2 & f814w & 460 & 0.097 + + + 2007 oct 29.78@xmath21 & wfpc2/pc1 & f450w & 700 & 0.046 + 2007 oct 29.80@xmath21 & wfpc2/pc1 & f814w & 320 & 0.046 + + + 2008 nov 20.66@xmath22 & wfpc2/pc1 & f555w & 460 & 0.046 + 2008 nov 20.67@xmath22 & wfpc2/pc1 & f814w & 700 & 0.046 + + 2008 nov 25.59@xmath22 & wfpc2/pc1 & f450w & 800 & 0.046 + 2008 nov 25.60@xmath22 & wfpc2/pc1 & f675w & 360 & 0.046 + + 2010 jan 10.81@xmath23 & wfc3/uvis & f336w & 1060 & 0.02 + 2010 jan 10.83@xmath23 & wfc3/uvis & f555w & 640 & 0.02 + 2010 jan 10.38@xmath23 & wfc3/uvis & f625w & 640 & 0.02 + + @xmath20 snap-9042 ( pi : s. smartt ) + @xmath21 go-11119 ( pi : s. van dyk ) + @xmath22 snap-10877 ( pi : w. li ) + @xmath23 go-11675 ( pi : j.r .
the site of sn 2007gr , in the late - time observations , is presented in figures [ fig : res : image ] and [ fig : res : latepre ] .
the sn resides in a bright , young stellar population dominated by blue stars covering a region of approximately 6 arcsecs in diameter .
our analysis of the pre - explosion wfpc2 observations confirms the results of @xcite and @xcite , who similarly found there was no source at the sn position in both the pre - explosion f450w and f814w images .
using dolphot , we estimate pre - explosion detection limits at the sn position of @xmath24 and @xmath25 mags . in the late - time wfpc2 observations
we recover the fading sn , with brightness @xmath26 , @xmath27 , @xmath28 and @xmath29 mags .
the sn is clearly not detected in the late - time wfc3 f336w image , however dolphot suggests a source might have been recovered at the sn position in the corresponding f555w and f625w images .
we note , however , that these detections are not confirmed in each of the four constituent dithered exposures and are complicated by the location of the sn position relative to the wings of the point spread function ( psf ) of a nearby star ( see section [ sec : res : srca ] ) , that is particularly bright at these wavelengths , and complexities in the wfc3/uvis psf in these filters .
we tested these possible detections by inserting faint artificial stars at the sn position , with a view to simulating different noise conditions at that location .
we found that the reported detection was highly sensitive to even the slightest modifications of the pixel counts at and around the sn position .
this suggests that the reported detections , if real , would be at the detection limit at the sn position and any photometry would be extremely uncertain , being highly sensitive to readout noise and noise associated with flux originating from the bright nearby star .
as such , we can not confidently report the detection of sn 2007gr or any other object at the sn position in the late - time wfc3 observations . for the wfc3 observations , we determined detection limits of @xmath30 , @xmath31 and @xmath32 mags . in both the late - time wfpc2 and wfc3 observations
we recover a bright source in close proximity ( offset by 0.12 arcsecs east ) to the position of sn 2007gr , previously identified by @xcite , and labelled source a ( see fig .
[ fig : res : latepre ] ) . in the late - time
wfc3/uvis f336w image we also recover another , fainter source at @xmath33 mags , which is offset from the sn position by 0.06 arcsecs , approximately south east ( which we label source b ) .
source b was not detected in any of the other wfc3 bands , suggesting that this is a blue source that , if it is star , has a spectral type of b2 or earlier .
+ photometry of source a in the late - time wfc3/uvis and wfpc2 images is presented in table [ tab : res : srcaphot ] .
we note that the late - time brightness of source a is consistent with the brightness of the source measured in the pre - explosion wfpc2 f450w and f814w observations , suggesting that source a is not variable . previously , @xcite and @xcite presented interpretations of the spectral energy distribution ( sed ) of source a in the context of it being a compact cluster , that might be related to the progenitor ( despite the large offset corresponding to a projected distance of @xmath34 ) .
the analyses of @xcite and @xcite were based on subsampled wfpc2 observations from 2001 and 2008 , respectively .
the late - time wfc3 observations were purposefully acquired to resample the psf to probe the spatial extent of source a and determine if it was a star or a compact cluster . + the dolphot photometry package returns measures of the shape and apparent physical extent of sources , through the @xmath35 and sharpness parameters .
for all the late - time wfc3 and wfpc2 images , dolphot identified source a as a point - like source .
we also utilised the ishape package to fit the observed spatial profile of source a with underlying cluster profiles ( here we used elliptical moffat profiles of order 1.5 ) convolved with the corresponding psf for each resampled wfc3 image .
two iterations of the ishape analysis were conducted with empirical psfs for each of the drizzled , resampled wfc3 images , derived directly from the images using the daophot package running under iraf , and psfs computed using the tinytim psf simulator .
the results from these analyses were found to be identical . to assess the suitability of an extended , cluster - like profile to explain the observed spatial properties of source a , we compared the quality of the fits using the moffat function with fits derived using a simple delta function convolved with the appropriate psf ( corresponding to a simple stellar - like , point source ) . following the analysis of
, we adopt the ratio of the @xmath35-values for fits using the two underlying functions to assess the relative merit of the extended cluster - like profile .
the results of the fits applied to source a , as observed in the late - time wfc3 f336w , f555w and f625w images , are presented in table [ tab : res : ishape ] .
we find that the extended moffat profile does not provide a significantly better fit to the spatial profile of source a than a simple delta function convolved with the psf .
we also inspected the output images produced by ishape , in which the model fit to source a was subtracted , and find that the only appreciable difference between the moffat and delta function fits was a slight modification to the background count levels .
we converted the corresponding values of the full width at half maximum ( fwhm ) for the moffat profiles , modified for the ellipticity of the profile , to values of effective radius ( @xmath36 ; ) .
given the scales down to which ishape is sensitive to truly extended profiles , we find that the values of the effective radius returned by ishape are below this threshold ( corresponding to 0.2 pc at the distance of ngc 1058 ) for source a as observed in the f555w and f625w images .
the inferred fwhm in the f336w image is only slightly larger than this threshold , however we note that the s / n level for source a is significantly lower than for the other filters and that the quality of the fit is also affected by the proximity of the nearby source b. the values of the effective radius of source a , if it is extended , are also significantly smaller than the limit ( @xmath37 ) for the cluster sample , for example , observed by in m51 .
we find no evidence , therefore , for source a being an extended source .
+ we also examined the observed sed of source a , in the three wfc3 uvis and four wfpc2 bands using our own bayesian sed fitting algorithm , in comparison with atlas9 @xcite stellar and starburst99 @xcite cluster seds . for each fit
we assumed a standard @xcite @xmath38 reddening law and varied the effective temperature and age for the stellar and cluster seds , respectively .
the results of these fits are presented in figure [ fig : res : srcased ] .
comparing the bayesian evidence computed for each of the fits , we calculate a bayes factor of @xmath39 which overwhelmingly supports the hypothesis that the underlying sed that describes the photometry of source a is a stellar sed .
this result is primarily driven by the difference in the predicted f336w magnitude and the size of the balmer jump between the stellar and cluster models .
+ from the considerations of both the observed sed and the spatial extent of source a , we conclude that this object is in a fact a star and not a cluster , as previously assumed .
the best fit parameters for the stellar fit to the sed of source a correspond to @xmath40 and @xmath41 mags , or an a1 - 3 supergiant following the calibration of @xcite and its application to the atlas9 models presented in the synphot manual .
the inferred reddening is higher than the value of @xmath42 estimated by @xcite and @xcite and the similar value derived by @xcite from measurements of the interstellar na i line strength for sn 2007gr .
the position of source a on the hr diagram is shown on figure [ fig : res : srcahrd ] , and the inferred luminosity of source a is found to be consistent with a @xmath43 star undergoing a short phase of evolution as a yellow supergiant @xcite .
cccccccc & & + + @xmath44 & @xmath45 & @xmath46 & & @xmath47 & @xmath45 & @xmath48 & @xmath49 + 22.057 & 21.333 & 21.152 & & 21.512 & 21.259 & 21.121 & 20.857 + ( 0.025 ) & ( 0.007 ) & ( 0.008 ) & & ( 0.017 ) & ( 0.016 ) & ( 0.022 ) & ( 0.017 ) + ccc + & @xmath50 & @xmath51 ( pc ) + + f336w & 0.98 & 0.42 + f555w & 0.99 & 0.10 + f625w & 0.99 & 0.10 + as noted previously , sn 2007gr is located at the centre of a young , massive star complex of approximate size 300 pc in diameter ( see fig . [
fig : res : image ] ) and in close proximity to a @xmath43 star .
the strong association with very massive stars , including those that might undergo a wolf - rayet phase , may have implications for the progenitor of sn 2007gr , if the progenitor star belonged to this population . to measure the properties of the surrounding stellar population we adopt a bayesian approach , based on the method and other similar methods , such as that presented by @xcite .
+ for a general star @xmath52 , with observed photometric data @xmath53 , we are primarily interested in determining the posterior probability for the age ( @xmath54 or @xmath55 ) and extinction ( @xmath56 ) through comparison with predicted brightnesses and colours derived from stellar evolution models and presented as isochrones . for this study
we use the padova isochrones . for the comparison of the observed data with isochrones ,
however , we also depend on other parameters such as the assumed distance modulus ( @xmath57 ) and the mass of the star ( @xmath58 ) .
in addition , one might also consider the effect of metallicity , however we will assume a single value corresponding to solar metallicity ( see section [ sec : intro ] ) .
+ following bayes theorem , the posterior probability distribution can be expressed in terms of the likelihood and the prior probability of the vector of parameters that underly the isochrone model : @xmath59 @xmath60 where @xmath61 constitutes the prior probabilities on the parameters , which we assume are independent .
following , we take the prior probability for the initial mass of the star to follow the form of the initial mass function ( imf ) : @xmath62 where the denominator normalises the prior over the range of masses present in a given isochrone and for a @xcite imf @xmath63 .
we note that @xmath64 only if @xmath65 , where @xmath66 and @xmath67 are the minimum and maximum masses for an isochrone with age @xmath54 .
the distance modulus @xmath57 is a nuisance parameter , which serves to shift the isochrones relative to the observed data .
given a reported distance modulus to the host galaxy of @xmath68 , the prior probability for @xmath57 is @xmath69 .
the initial mass @xmath58 serves as a coordinate along the length of the isochrone and for a given isochrone , the observed data may suggest a range of masses .
the mass is treated as a nuisance parameter and is also marginalised over .
+ the likelihood function is defined with the observed photometry in each filter compared directly to the predicted magnitudes from the isochrones .
rather than comparing the observed photometry with isochrones in the colour - magnitude plane , our approach specifically avoids complications associated with correlated uncertainties that arise in using colours , but necessitates the explicit use of the distance modulus as a nuisance parameter @xcite .
the likelihood takes the form : @xmath70 where @xmath71 is the @xmath72th _ apparent _ magnitude for star @xmath52 and @xmath73 is the _ absolute _ magnitude predicted for a star with mass @xmath58 lying on an ischrone with age @xmath54 and extinction @xmath56 .
in addition , if the star is not detected in a given filter the corresponding detection limit ( @xmath74 , @xmath75 ; see section [ sec : obs ] ) can be substituted into the likelihood expression as : @xmath76 this enables us to consider photometry covering a wide wavelength range that may be partially incomplete .
following we enforce the normalisation requirement for the likelihood such that : @xmath77 for flat priors on @xmath54 and @xmath56 , where : @xmath78 + the padova isochrones are only computed to reproduce the properties of single stars .
this is likely to provide an incomplete representation of the observed properties of a given stellar population , if a proportion of the observed sources are unresolved binaries with colours that deviate from single star isochrones due to the composite nature of their seds . in order to consider the role of binarity on the observed stellar population
, we adopt the approximation that the observed brightness and colours of a binary system arise from the simple addition of flux from two single stars that lie on the same isochrone .
this approximation is , however , only valid for non - interacting binaries , for which the evolution of the two stars proceeds independently , which in some cases may have significant flaws @xcite . we assign the mass @xmath58 to describe the mass of the primary star , and introduce an additional nuisance parameter @xmath79 ( where @xmath80 ) to define the binary mass ratio ( for which we assume a flat prior ) . for each point along the isochrone we consider the magnitudes resulting from the sum of fluxes from a star with mass @xmath58 at that point and a secondary with mass equal or less than @xmath58 . in the equivalent form of equation [ eqn : post ] , for the binary star scenario
the parameter @xmath79 is also marginilised over .
while a single star isochrone defines an infinitely narrow track across the magnitude - magnitude plane , the inclusion of binaries makes the isochrone `` fuzzier '' , with brighter stars also accommodated by fainter single star isochrones .
+ the benefit of a bayesian approach to this problem is that we do not need to explicitly determine if a star is a single star or a binary .
bayesian mixture models enable us to embrace the ambiguity between the two cases , by considering the final probability as a sum of the single and binary scenarios : @xmath81 where @xmath82 is the probability of a source being a binary ( or binary fraction ) , for which we adopt @xmath83 .
+ due to the non - monotonic nature of the predicted properties of isochrones and degeneracies in the underlying parameters , it is unlikely that there will be a unique solution for a given star in the @xmath84 plane .
in addition , if one considers a population of stars , there is no reason that this population may itself be described by a single , unique solution .
this issue becomes more problematic if the constituent stars arise from a spread of population characteristics . in the bayesian context
it is convenient to consider this as an hierarchical problem , with the aim to identify the parameters of the underlying distributions from which the observed population of stars is drawn .
this approach uses each star as a separate probe of the underlying properties of the population , and uses the entire ensemble of stellar properties to resolve the degeneracies in the isochrone fitting problem . similarly to @xcite , we consider underlying age distributions of stars to take the form @xmath85 , which have normal distributions in log(years ) and log - normal distributions in years .
a given star may have age @xmath54 , which is drawn from distributions parameterised by @xmath86 and @xmath87 .
using bayes theorem , the hierarchical form becomes : @xmath88 as shown by @xcite , we may also consider the observed population as arising from a mixture of underlying distributions ( or epochs of star formation ) , in analogy to the mixture of single and binary stars discussed above .
the mixture of underlying distributions can be described by the vector of parameters @xmath89 , where @xmath90 is a vector containing the relative weights of each component distribution in the mixture , such that @xmath91 .
the likelihood of the observed data for the @xmath92 star arising from the @xmath93 distribution of ages is @xmath94 , such that for the entire set of observations @xmath95 for all @xmath96 stars : @xmath97 where the total number of components in the mixture is @xmath98 and it is assumed , in this case , that all populations share the same extinction @xmath56 .
@xcite used a reversal jump markov chain monte carlo technique to determine @xmath98 as part of the fitting process .
we adopt a different approach by using the nested sampling algorithm @xcite , with importance sampling @xcite , to calculate the bayesian evidence or marginal likelihood ( @xmath99 ; the constant of proportionality in equation [ eqn : hierarchical ] ) for different @xmath98 and using bayesian model selection to determine the optimal value of @xmath98 to describe the observed population . in using nested sampling
there are two important considerations : 1 .
the labels for each underlying distribution are interchangeable and , as such , the nested sampling algorithm simultaneously finds all labelling permutations for @xmath100 .
this leads to the evidence @xmath99 being overestimated by a factor of @xmath101 . due to the trivial nature of the correction that needs to be applied to the bayesian evidence , it is convenient to find all modes , including mirror modes , and only sort the equally weighted posterior samples once the algorithm has converged , to find individual solutions for each component 2 .
the addition of mixture components in excess of the optimal number of components will not result in an increase in @xmath99 ( following the interpretation of @xcite ) .
the minimum value of @xmath98 that maximises @xmath102 represents the most parsimonious solution , that adequately describes the observed population using the minimum number of parameters .
we applied this technique to the 284 and 99 detected stars in the late - time wfc3 and wfpc2 observations , respectively , within 3 arcsecs of the position of sn 2007gr .
due to the differences in depth and spatial resolution between the wfpc2 and wfc3 observations , we kept the two datasets separate and conducted the analyses in parallel .
the positions of these stars in colour magnitude diagrams are shown on figures [ fig : cmd : wfc3 ] and [ fig : cmd : wfpc2 ] .
the colour - magnitude diagrams for both sets of observations are dominated by a `` blue plume '' of bright stars and a fainter collection of stars consistent with the main sequence .
we utilised padova isochrones assuming a solar metallicity .
we used isochrones with ages in the range @xmath103 ( at @xmath104 intervals ) for extinctions in the range @xmath105 ( with a @xcite reddening law ) .
the resulting values for the bayesian evidence for @xmath106 mixture components are presented in table [ tab : evid ] and on figure [ fig : res : evidence ] , from which it is clear that the optimal value of mixture components to describe both datasets is @xmath107 .
the addition of further components does not significantly improve the quality of the fits to either the wfpc2 or wfc3 data .
the best fit parameters for the two populations are presented in table [ tab : res : sfh ] and the corresponding `` effective '' star formation histories are presented on fig .
[ fig : res : sfh ] .
the position of the corresponding isochrones for the best fit solutions ( and also the @xmath108 solutions ) are shown on figures [ fig : cmd : wfc3 ] and [ fig : cmd : wfpc2 ] .
we find that that the wfpc2 observations suggest a slightly younger , less reddened population ; however , this is likely due to the limited number of stars observed in the shallower wfpc2 observations .
for both sets of observations , we find evidence for an older stellar population solution that is composed of @xmath109 ( by number ) of the stars observed in the field .
+ under the assumption that the progenitor of sn 2007gr arose from the observed stellar populations , given the corresponding dependence of a star s lifetime on its initial mass we can derive the probability for the initial mass of the progenitor . on figure [ fig : res : mass ] , we show that wfpc2 observations , which imply younger , narrower epochs of star formation , support a higher mass progenitor than the wfc3 observations .
for the wfpc2 and wfc3 observations , the mass at which @xmath110 occurs for @xmath111 and @xmath112 , respectively .
both sets of observations , therefore , support a progenitor significantly more massive than those found for type iip or iib sne .
ccccc & & + + @xmath98 & @xmath113 & wfc3/uvis & & wfpc2 + 1 & 3 & @xmath114 & & @xmath115 + 2 & 7 & @xmath116 & & @xmath117 + 3 & 10 & @xmath118 & & @xmath119 + 4 & 13 & @xmath120 & & @xmath121 + + @xmath122 normalised with respect to @xmath123 .
cccccccccc & & & & & + + & @xmath56 & & @xmath124 & @xmath125 & @xmath126 & & @xmath127 & @xmath128 & @xmath129 + wfc3/uvis & @xmath130 & & @xmath131 & @xmath132 & @xmath133 & & @xmath134 & @xmath135 & @xmath136 + wfpc2 & @xmath137 & & @xmath138 & @xmath139 & @xmath140 & & @xmath141 & @xmath142 & @xmath143 +
despite the lack of a detection of a progenitor in pre - explosion images , the nature of the progenitor of sn 2007gr is particularly intriguing given the classification of the sn as a type ic sn and the location of the sn at almost the centre of a dense young , massive star association and in close proximity to a star as massive as @xmath43 . from the wfc3 and wfpc2 observations there is immediate support for the hypothesis that sn 2007gr resulted from a wolf - rayet progenitor that was an initially massive star .
sn 2007gr is an important type ic sn , principally because of the proximity of its host galaxy which resulted in a detailed photometric and spectroscopic monitoring campaign .
studies by @xcite and @xcite reported that sn 2007gr also hosted a relativistic outflow , that might imply sn 2007gr had a connection with the gamma ray burst phenomenon .
using further x - ray and radio observations of sn 2007gr , @xcite robustly disagree with the previous conclusions and find sn 2007gr to only be expanding at trans - relativistic ( @xmath144c ) velocities .
sn 2007gr is extraordinary for its very prompt radio peak in luminosity which , in comparison with the compilation of observations of type ii ( including iip , iin and iil ) and type ib / c sne compiled by @xcite , implies high blast wave speeds that are only compatible with compact progenitors such as wr stars .
based on previous hst observations of the site of sn 2007gr , @xcite and @xcite have presented mass estimates based on ages determined from the sed of source a under the assumption that it was a cluster .
based on the high - resolution hst images presented here , we find source a is consistent with a point stellar - like source , and not a cluster ; hence , the previously derived masses for the progenitor are based on an assumption that we propose to be flawed . from a census of the ejecta , conducted using nebular phase spectroscopy of sn 2007gr , @xcite calculated that the supernova resulted from the explosion of a relatively low mass star with @xmath145 .
this is at the lower limit for the masses we estimate for the progenitor from the age of the surrounding stellar population . given the radically different nature of the two approaches , the source of the discrepancy between our mass estimate for the progenitor and the estimate presented by @xcite is unclear .
@xcite used integral field spectroscopy of knots of @xmath146 emission surrounding the site of sn 2007gr to estimate an age for the progenitor in the range of @xmath147 using the equivalent width of @xmath146 as a proxy for the age .
the reported age range is consistent with the age for the stellar population found here .
@xcite note , however , that given the limitations of the seeing conditions of the observations , individual sources in the vicinity of the sn were not resolved .
previously @xcite discussed a continuum subtracted @xmath146 image of the site of sn 2007gr acquired on 2005 january 13 with the wide field camera ( wfc ) of the isaac newton telescope ( int ) . on figure
[ fig : disc : halpha ] , we show contours corresponding to @xmath146 emission over the late - time wfc3/uvis f336w observation . despite the significant difference in spatial resolution between the two images ,
it is clear that the site of sn 2007gr and the centre of the massive star association are not associated with significant @xmath146 emission .
this has important consequences for the use of the @xmath146 equivalent width as an age proxy , since the sn position may itself not be associated with any @xmath146 emission .
the use of pixel statistics to assess the proximity of sn positions with tracers of star formation ( e.g. @xmath146 emission ; for a review see @xcite ) may also be undermined by the use of low resolution ground based imaging which , as in the case of sn 2007gr , might correlate @xmath146 emission with a sn , where there is actually no association ; further reinforcing the conclusions of @xcite .
this highlights the fact that weak association with @xmath146 emission does not immediately imply a low mass progenitor .
+ here , we have used observations of the surrounding stellar population covering a large wavelength range from the ultraviolet to the near - infrared .
the observations , as presented in fig .
[ fig : cmd : wfc3 ] and [ fig : cmd : wfpc2 ] , probe a large section of the main sequence of massive stars , as well as a small number of stars that have evolved off the main sequence .
the only requirement of this analysis is that the unseen progenitor is drawn from the same underlying distributions as the observed population of stars .
previously , studies of the ages of resolved stellar populations associated with the sites of ccsne have concentrated on the nearby population within a radius of @xmath148 , under the assumption that stars in such proximity will be effectively co - eval with the progenitor , while at the same time limiting contamination from the background population @xcite .
we have not placed a strict radius constraint on the size of the region associated with sn 2007gr , except by eye constraining the size of the massive star association to @xmath149 .
the bayesian analysis presented here permits us to include as many additional populations as required to fit the observed data .
in addition , a background stellar population could also be treated as an additional mixture component , described by stars away from the central region of interest .
we have approximately estimated the spatial extent of the massive stellar population with which sn 2007gr is associated to have a diameter of @xmath150 .
the ages we have derived from the wfpc2 and wfc3 observations are significantly shorter than the expected sound crossing for a single cloud of this size ( @xmath151 * ? ? ?
there are , however , massive star complexes that cover large spatial scales in the galaxy , that have small age spreads that are also incompatible with the expected sound crossing times . by comparison (
* and references therein ) studied the ob association u sco , which has an approximate spatial extent of @xmath152 , and found the stellar population could be described by a single age of 5 myr , without any significant age spread , implying a single burst of star formation .
@xcite suggested that star formation was triggered by a sn shock wave .
although u sco has smaller dimensions than the massive star association in ngc 1058 , it does illustrate that such large structures with small age spreads can exist .
conversely , if the constituent stars of the association were formed in a single location , velocity dispersions of only @xmath153 would be required to reproduce the observed spatial distribution of stars in the lifetime of the young age solution .
it is also worth contrasting the dense environment in which the , apparently , massive progenitor of sn 2007gr resides , compared to the relatively sparse neighbourhood of sn 2002ap , for which deep pre - explosion images ruled out all single massive progenitor scenarios @xcite . an important factor not included in our bayesian treatment is the possibility of differential extinction .
we have assumed that the stars that make up the two observed populations share the same , single valued reddening .
we note that , similarly to the differences in age inferred from the wfc3 and wfpc2 observations , the difference in wavelength coverage may be behind the difference in extinction ; with observations in the uv more heavily affected by reddening , than the corresponding redder wfpc2 observations . if the extinction for each star arises from a distribution of extinctions , rather than a single value , this could also be implemented using an hierarchical scheme .
care , however , would need to be taken to avoid any biases introduced due to the requirement that @xmath154 .
it is expected that the inclusion of differential reddening would make the age distributions narrower ; as a single isochrone could accommodate a larger range of colours , than under the assumption of a single value for the extinction
. such analysis of the host environments of sne , as presented here , has been previously employed by @xcite , @xcite , and @xcite for estimating the progenitor masses for ccsne .
@xcite found that for the type iip sne in their sample they were able to achieve similar estimates for the median mass as those derived directly for the progenitor in pre - explosion images ; as was also achieved by @xcite from their analysis of the population surrounding the type iib sn 2011dh .
although , as evidenced by figure [ fig : res : mass ] , there are large uncertainties associated with this approach , the case of sn 2007gr does highlight that the lack of a detection of a progenitor in pre - explosion images does not imply a low mass .
@xcite argues that the lack of detection of progenitors of high mass stars ( @xmath155 ) suggests that these stars may not be exploding at all as sne , but rather collapsing to form black holes .
caution is required in drawing such a conclusion as there are a number of major issues concerning the analysis of the pre - explosion detections of the progenitors of ccsne that have yet to be resolved .
a fundamental problem in the consideration of the ensemble of progenitor detections and non - detections in a statistical context is that it is unclear how the , generally poor , quality of available pre - explosion observations effectively censors the underlying distribution of sn progenitors . in the case of the pre - explosion observations of sn 2007gr , consisting of shallow wfpc2 f450w and f814w images
, it was not surprising that a possible wolf - rayet progenitor might not be detected @xcite .
we used the seds ofwrh / powr / powrgrid1.html ] ( over a temperature range of @xmath156 and transformed radius of @xmath157 ) to probe the sensitivity of the pre - explosion observations ( see section [ sec : res : location ] ) to a wc progenitor . using the extinction estimate for the host stellar population determined here ,
the minimum luminosity that might yield a detectable progenitor ( @xmath158 ) in the pre - explosion observations is @xmath159 .
this limit is significantly higher than the luminosities predicted for stars that will end their lives as wr stars and the luminosities derived by for a sample of galactic wc stars .
based on stellar populations associated with sn remnants in m31 and m33 , @xcite conclude that the maximum mass for a star to explode may be as high as @xmath160 . in the absence of pre - explosion data of sufficient quality to detect the progenitors of certain types of sne , in particular those from higher mass stars , this analysis could play an important role in establishing the fate of those stars that do not lead to progenitors as easy to detect as rsgs .
it is clear from analyses such as this that contextual information about the locations of sn explosions can be just as important as direct detection of progenitors .
indeed , if a progenitor is detected a significant amount of information might be missed by ignoring the stars nearest to it . in studying population of stars associated with a sn
, we may potentially directly probe the physical conditions in which the progenitor formed and in which its entire evolution took place ; to which some observational proxies ( such as @xmath146 emission equivalent width ) might be insensitive in individual cases , only providing insight for a large statistical sample .
the research of jrm is supported through royal society university research fellowship .
the authors thank daniel mortlock for reviewing a draft of the paper and to the referee nancy elias - rosa .
the authors are grateful to jason kalirai , paul crowther and simon goodwin for their useful discussions during the preparation of this manuscript . , y. , kasliwal , m. m. , arcavi , i. , horesh , a. , hancock , p. , valenti , s. , cenko , s. b. , kulkarni , s. r. , gal - yam , a. , gorbikov , e. , ofek , e. o. , sand , d. , yaron , o. , graham , m. , silverman , j. m. , wheeler , j. c. , marion , g. h. , walker , e. s. , mazzali , p. , howell , d. a. , li , k. l. , kong , a. k. h. , bloom , j. s. , nugent , p. e. , surace , j. , masci , f. , carpenter , j. , degenaar , n. , & gelino , c. r. 2013 , , 775 , l7 , d. j. , valenti , s. , kotak , r. , meikle , w. p. s. , taubenberger , s. , pastorello , a. , benetti , s. , stanishev , v. , smartt , s. j. , trundle , c. , arkharov , a. a. , bufano , f. , cappellaro , e. , di carlo , e. , dolci , m. , elias - rosa , n. , frandsen , s. , fynbo , j. u. , hopp , u. , larionov , v. m. , laursen , p. , mazzali , p. , navasardyan , h. , ries , c. , riffeser , a. , rizzi , l. , tsvetkov , d. y. , turatto , m. , & wilke , s. 2009 , , 508 , 371 , z. , taylor , g. b.and kouveliotou , c. , granot , j. , ramirez - ruiz , e. , bietenholz , m. , van der horst , a. j. , pidopryhora , y. , van langevelde , h. j. , garrett , m. a. , szomoru , a. , argo , m. k. , bourke , s. , & paczyski , b. 2010 , , 463 , 516 , s. , elias - rosa , n. , taubenberger , s. , stanishev , v. , agnoletto , i. , sauer , d. , cappellaro , e. , pastorello , a. , benetti , s. , riffeser , a. , hopp , u. , navasardyan , h. , tsvetkov , d. , lorenzi , v. , patat , f. , turatto , m. , barbon , r. , ciroi , s. , di mille , f. , frandsen , s. , fynbo , j. p. u. , laursen , p. , & mazzali , p. a. 2008 , , 673 , l155 | we present an analysis of late - time hubble space telescope wide field camera 3 and wide field planetary camera 2 observations of the site of the type ic sn 2007gr in ngc 1058 .
the sn is barely recovered in the late - time wfpc2 observations , while a possible detection in the later wfc3 data is debatable .
these observations were used to conduct a multiwavelength study of the surrounding stellar population .
we fit spatial profiles to a nearby bright source that was previously proposed to be a host cluster .
we find that , rather than being an extended cluster , it is consistent with a single point - like object .
fitting stellar models to the observed spectral energy distribution of this source , we conclude it is a1-a3 yellow supergiant , possibly corresponding to a star with @xmath0 .
sn 2007gr is situated in a massive star association , with diameter of @xmath1 .
we present a bayesian scheme to determine the properties of the surrounding massive star population , in conjunction with the padova isochrones .
we find that the stellar population , as observed in either the wfc3 and wfpc2 observations , can be well fit by two age distributions with mean ages : @xmath2myr and @xmath3myr .
the stellar population is clearly dominated by the younger age solution ( by factors of 3.5 and 5.7 from the wfpc2 and wfc3 observations , respectively ) , which corresponds to the lifetime of a star with @xmath4 .
this is strong evidence in favour of the hypothesis that sn 2007gr arose from a massive progenitor star , possibly capable of becoming a wolf - rayet star . stellar evolution : general supernovae : general supernovae : individual:2007gr galaxies : individual : ngc1058 methods : statistical |
it is well known that a pure @xmath1 gauge theory at high temperature undergoes a phase transition .
this phase transition is of special interest because of many its aspects can be characterized precisely @xcite . in particular , the order parameter is given by the polyakov loop @xmath2 \,,\ ] ] where the trace is over the fundamental representation , @xmath3 is a periodic variable of period @xmath4 , with @xmath5 the temperature , @xmath6 is a gauge coupling constant , and @xmath7 is a vector potential in the time direction .
the usual interpretation of is as a phase factor associated to the propagation of an infinitely heavy test quark in the fundamental representation of the gauge group . until recently ,
the lattice formulation , still struggling with limitations and system errors , and effective field theories were the main computational tools to deal with non - weakly coupled gauge theories .
the polyakov loop was also intensively studied ( see , for example , @xcite and references therein ) .
the situation changed drastically with the invention of the ads / cft correspondence @xcite that resumed interest in another tool , string theory . in this note
we continue a series of recent studies @xcite devoted to a search for an effective string description of pure gauge theories . in @xcite ,
the model was presented for computing the heavy quark and multi - quark potentials at zero temperature .
subsequent comparison @xcite with the available lattice data has made it clear that the model should be taken seriously .
later , this model was extended to finite temperature .
the results obtained for the spatial string tension @xcite and the thermodynamics @xcite are remarkably consistent with the lattice , too . as is known , qcd is a very rich theory supposed to describe the whole spectrum of strong interaction phenomena .
the question naturally arises : how well does the model describe other aspects of quenched qcd ? here , we attempt to _ analytically _ evaluate the polyakov loop as an important step toward answering this question @xcite . in addition
, a good motivation for this test is lattice data revealed recently by @xcite . before proceeding to the detailed analysis ,
let us set the basic framework .
as in @xcite , we take the following ansatz for the five - dimensional background geometry @xmath8 where @xmath9 .
@xmath10 is a deformation parameter whose value can be fixed from the critical temperature @xcite .
we take a constant dilaton and discard other background fields . in discussing the wilson and polyakov loops within the gauge / string duality @xcite , one first chooses a contour @xmath11 on a four - manifold which is the boundary of a five - dimensional manifold .
next , one has to study fundamental strings on this manifold such that the string world - sheet has @xmath11 as its boundary . in the case of interest
, @xmath11 is an interval between @xmath12 and @xmath4 on the @xmath3-axis .
the expectation value of the polyakov loop is schematically given by the world - sheet path integral @xmath13 where @xmath14 denotes a set of world - sheet fields .
@xmath15 is a world - sheet action . in principle , the integral can be evaluated approximately in terms of minimal surfaces that obey the boundary conditions .
the result is written as @xmath16 $ ] , where @xmath17 means a renormalized minimal area whose weight is @xmath18 .
given the background metric , we can attempt to calculate the expectation value of the polyakov loop by using the nambu - goto action for @xmath15 in @xmath19 here @xmath20 is the background metric . in the case of interest , this action describes a fundamental string stretched between the test quark on @xmath11 ( at @xmath21 ) and the horizon at @xmath22 .
since we are interested in static configurations , we choose @xmath23 , @xmath24 .
this yields @xmath25 where @xmath26 . a prime stands for a derivative with respect to @xmath27 .
now it is easy to find the equation of motion for @xmath28 @xmath29'=0 \,.\ ] ] it is obvious that eq . has a special solution @xmath30 that represents a straight string stretched between the boundary and the horizon .
since this solution makes the dominant contribution , as seen from the integrand in , we wo nt dwell on other solutions here .
having found the solution , we can now compute the corresponding minimal area . since the integral is divergent at @xmath21 due to the factor @xmath31 in the metric
, we regularize it by imposing a cutoff @xmath32 @xmath33 subtracting the @xmath34 term ( quark mass ) and letting @xmath35 , we get a renormalized area @xmath36 where @xmath37 is a normalization constant which is scheme - dependent .
next , we can perform the integral over @xmath27 .
the result is @xmath38 in this formula @xmath39 is given by @xmath40 @xcite . combining the weight factor with the normalization constant as @xmath41 , we find @xmath42 \,,\ ] ] with @xmath43 the imaginary error function
this is our main result .
it is of great interest to compare the temperature dependence of with other results for the high temperature phase of @xmath1 gauge theory . in doing so , we start with lattice
qcd . clearly , @xmath44 is of primary importance . in fig.1
a comparison is shown with the recent data of @xcite .
we see that our model is in a quite good agreement with the lattice for a broad temperature range @xmath45 .
the maximum discrepancy occurred at @xmath46 is of order 15% .
it rapidly decreases with temperature reaching 2% at @xmath47 and becoming almost negligible up to @xmath48 .
then , it starts to grow back again . for completeness , we can fit the value of @xmath49 to be @xmath50 that significantly improves accuracy .
for example , at @xmath46 it becomes of order 6% .
one possible explanation for the better fit is that we have evaluated classically ( in terms of strings ) .
if we take into account semi - classical corrections , then the value of @xmath49 gets renormalized . for practical purposes , the expression looks somewhat awkward .
following @xcite , we expand @xmath51 and @xmath52 in powers of @xmath53 . if we ignore all higher terms , then a final result can be written in two simple forms : @xmath54 \,,\ ] ] or @xmath55 in fig.2 we have plotted the results . as can be seen , above @xmath56
the discrepancy between the expression and approximations - is negligible . at lower t the approximation ( exponential law )
it shows a significant deviation from the lattice .
in particular , the discrepancy occurred at @xmath46 is of order 27% . on the other hand ,
the agreement between the approximation ( power law ) and the lattice is spectacular .
for the temperature range @xmath45 the power law provides a reliable approximation to lattice qcd with accuracy better than 5% !
moreover , one can use it to describe all available lattice data of @xcite at lower @xmath5 .
then , the maximum discrepancy occurred at the lowest available value @xmath57 is of order 7% .
it is worth noting that the exponential law has been suggested in @xcite based on a dimension - two condensate @xmath58 @xcite .
such a condensate as well as its possible links to the uv renormalon and @xmath59 corrections got intensively discussed in the qcd literature @xcite . as was first shown in @xcite ,
the deformation parameter @xmath10 of the background geometry is tied into the appearance of the quadratic corrections .
it is not , therefore , surprising that we have recovered in our calculations .
interestingly , the power law is very similar to that observed for the pressure in @xcite .
indeed , for @xmath60 the pressure is simply @xmath61 .
in this note we have evaluated the polyakov loop using the now standard ideas motivated by gauge / string duality .
a key point is the use of the background metric which is singled out by the earlier works @xcite .
( note that there is no need for any free parameters except a scheme - dependent normalization constant @xmath62 . )
the overall conclusion is that the same background metric results in a very satisfactory description of the polyakov loop as well .
of course , we still have a lot more to learn before answering the question posed at the beginning of this note .
we would like to thank r.d .
pisarski and p. weisz for useful discussions , and s. hofmann for reading the manuscript .
this work is supported in part by dfg " excellence cluster and the alexander von humboldt foundation under grant no .
phys0167 .
99 a.m. polyakov , phys.lett.b * 72 * , 477 ( 1978 ) ; g.t hooft , nucl.phys.b * 138 * , 1 ( 1978 ) ; l. susskind , phys.rev.d * 20 * , 2610 ( 1979 ) ; b. svetitsky and l.g .
yaffe , nucl.phys.b * 210 * , 423 ( 1982 ) .
pisarski , `` qcd phase diagram , lectures presented at the ' ' 47 internationale universit " atswochen f " ur theoretische physik
, schladming , austria , march 2009 ; + see also http://physik.uni-graz.at/itp/iutp/iutp@xmath6309/ + lecturenotes/pisarski/[email protected] .
maldacena , adv.theor.math.phys .
* 2 * , 231 ( 1998 ) ; + s.s .
gubser , i.r .
klebanov , and a.m. polyakov , phys . + lett .
b * 428 * , 105 ( 1998 ) ; e. witten , adv.theor.math .
* 2 * , 253 ( 1998 ) .
o. andreev and v.i .
zakharov , phys.rev.d * 74 * , 025023 ( 2006 ) ; o. andreev , phys.rev.d * 78 * , 065007 ( 2008 ) .
o. andreev and v.i .
zakharov , phys.lett.b * 645 * , 437 ( 2007 ) ; o. andreev , phys.lett.b * 659 * , 416 ( 2008 ) .
o. andreev , phys.rev.d * 76 * , 087702 ( 2007 ) .
white , phys.lett.b * 652 * , 79 ( 2007 ) .
see also the earlier work which deals with numerical estimates , o. andreev and v.i .
zakharov , jhep * 04 * , ( 2007 ) 100 .
s. gupta , k. huebner , and o. kaczmarek , phys.rev .
d*77 * , 034503 ( 2008 ) .
alternatively , it may be fixed from the heavy quark potentials @xcite .
while a significant literature on the wilson loops has grown , there has been relatively little investigation of the polyakov loop . for some developments , see however e. witten , adv.theor.math.phys . * 2 * , 505 ( 1998 ) ; a. hartnoll and s. prem kumar , phys.rev.d*74 * , 026001 ( 2006 ) ; m. headrick , phys.rev.d*77 * , 105017 ( 2008 ) .
e. megias , e. ruiz arriola , and l.l .
salcedo , jhep * 0601 * , 073 ( 2006 ) .
celenza and c.m .
shakin , phys.rev.d * 34 * , 1591 ( 1986 ) . for a review ,
zakharov , nucl.phys.proc.suppl .
* 74 * , 392 ( 1999 ) and references therein .
o. andreev , phys.rev.d * 73 * , 107901 ( 2006 ) .
pisarski , prog.theor.phys.suppl.*168 * , 276 ( 2007 ) . | we use gauge / string duality to _ analytically _ evaluate the renormalized polyakov loop in pure yang - mills theories . for @xmath0 , the result is in a quite good agreement with lattice simulations for a broad temperature range . |
numerical investigations of the dynamics of dust particles in the neighborhood of a gap - opening planet reveal that they tend to concentrate at gap edges , for planet masses between @xmath1 and @xmath2 , with @xmath3 the mass of jupiter @xcite .
these accumulations are of great interest not only because they could facilitate planetesimal formation , but also because they could supply solid material to the planetary atmosphere through deposition of small grains , and hence contribute to the atmospheric opacity , although perhaps modestly @xcite .
in particular , a jupiter - mass planet accretes predominantly particles smaller than 10 @xmath0 m @xcite , which places a tight constraint on its solid enrichment . one aspect of the evolution of dust in the vicinity of a gap - opening planet that has not been considered is the internal structure of solid aggregates .
the porosity of dust particles affects their aerodynamic coupling to the disk gas .
their structure can also affect the ionization state of the protoplanetary disk : the increased surface area associated with internal voids within the dust aggregates allows for fast electron recombination , which can lower the ionization levels of the gas phase , and thus render the disk gas stable to the magnetorotational instability , or mri @xcite , a likely candidate to produce turbulent viscosity in disks [ although other mechanisms have been proposed to transfer a protoplanetary disk s angular momentum , such as centrifugally - driven winds and jets @xcite and magnetic breaking @xcite ] .
dust in the vicinity of a gap - opening planet is subject to a dynamic and energetic environment .
for example , inside the gap , shocks are generated when streams of gas flow through the l1 and l2 points towards the interior of the planet s hill sphere @xcite .
these shocks can modify the crystalline structure of silicates through thermal annealing .
knowledge of the porous structure of dust aggregates is essential to determine their heat conductivity during these events .
previous studies of dust coagulation and dust charging in protoplanetary disks show that electric charges lead to larger , more massive and more porous aggregates than in the case of neutral coagulation @xcite . in turn , the evolution of the aggregate porosity may define a region in weakly ionized disks where the growth of sub - micron - sized particles becomes stalled due to electrostatic repulsion between negatively - charged aggregates @xcite
. the dusty ingredient in the mri recipe could also be crucial for the dynamics of giant - planet circumplanetary disks .
@xcite showed that , under some circumstances , the circumjovian disk has sufficient electrical conductivity for magnetic forces to drive accretion stresses .
but if the disk contains enough dust , mri turbulence is unlikely to occur , even in the presence of ionizing x - ray radiation .
the circumjovian disk can develop a substantial magnetically inactive dead zone " , where regular satellites could form .
it is thus evident that the size and internal structure of dust grains are key to understanding the birth environment of planets that are capable of opening gaps . to better characterize the spatial distribution of these dust properties around one such planet , in this study
we combine a numerical model of mri turbulence with a dust - coagulation algorithm .
our results will provide a clearer picture of the dust aggregate structure around a young , jupiter - mass planet . in the interest of specificity , we use our models to invoke a scenario in which jupiter underwent migration within the primitive solar nebula , the so - called grand tack hypothesis @xcite . according to this hypothesis ,
jupiter migrated from inside its current position ( but beyond the main asteroid belt ) to a distance of about 1.5 au from the sun , where it encountered an orbital resonance with the trailing saturn . both planets then reversed their orbital motion outwards .
the grand tack helps reproduce the formation of mars analogs with the correct mass from a disk of planetary embryos . in section [ sec :
method ] we describe our numerical models .
section [ sec : results ] presents our results . in section [ sec : discussion ]
we discuss these results , and we make concluding remarks in section [ sec : conclusions ] .
we use a magnetohydrodynamic ( mhd ) protoplanetary disk model in the local shearing box approximation @xcite , in which the mhd equations are solved in a rectangular coordinate system that corotates with the disk at a fiducial orbital radius @xmath4 , with angular frequency @xmath5 . in this system
, the @xmath6 axis is oriented along the radial direction , the @xmath7 axis along the azimuthal direction , and the @xmath8 axis is parallel to the disk s angular momentum vector . our solver is the athena code @xcite , a grid - based algorithm that has been extensively tested and employed for various studies of protoplanetary disks .
our numerical setup is similar to that of @xcite : the local , 3d disk model is isothermal , and does not include vertical stratification of the gas density .
the box dimensions are @xmath9 , where @xmath10 is the disk scale height .
we use a numerical resolution twice as large as that in @xcite , namely 64 grid cells per @xmath10 . in the code s system of units , the gas sound speed @xmath11 ,
the initial gas density @xmath12 , and the angular frequency @xmath13 are all equal to 1 .
the initial magnetic field strength is given by the plasma beta , @xmath14 , and the initial field configuration corresponds to a non - zero net vertical magnetic flux .
to model the gravitational effect of a planet on the surrounding disk gas , we place a cylindrical potential at the center of the box , with axis coincident with the box s vertical axis . as in @xcite ,
the planet potential is given by @xmath15 where @xmath16 is the gravitational constant , @xmath17 is the planet mass , @xmath18 is the distance to the @xmath8 axis , and @xmath19 is a smoothing length to avoid small time steps close to the source . the potential is also smoothed out to take into account discontinuities at the box boundaries : @xmath20^{2 } + g^{2}m^{2}_{\rm p}/r^{2}_{\rm co } } , \\
\hfill \text{if $ r < r_f$}\\ \phi_{\rm p}\left(r_f\right)-gm_{\rm p}/r_{\rm co } , \hfill \text{if $ r > r_f$ } \end{cases}\ ] ] where @xmath21 is the cutoff distance to the @xmath8 axis beyond which the potential flattens out , and @xmath22 is a smoothing length at the cutoff radius . following @xcite , we set @xmath23 and @xmath24 . in athena , we express the planet mass in terms of the so - called thermal mass @xmath25 , the mass at which the hill radius and the bondi radius of the planet are comparable to the disk scale height @xcite . for the minimum - mass solar nebula model [ mmsn ; @xcite ] , the thermal mass is @xcite @xmath26 where @xmath27 is the planet s semi - major axis , and @xmath28 is the mass of earth . within the grand tack scenario ,
jupiter begins its inward migration at @xmath29 au , reaches 1.5 au , and reverses its motion towards its current orbital position at 5.2 au @xcite .
since we are not modeling the actual migration , but effectively only a snapshot " of that process , we choose @xmath30 au , because it is an interesting region of the solar nebula in terms of the dynamical evolution of chondritic parent bodies @xcite . in that case , from relation ( [ eq : mth ] ) a value of @xmath31 gives our planet a mass @xmath32 .
we run this high - resolution setup for 51 orbits ( 265 yr at 3 au in the mmsn ) , and use the resulting data from the last 12 orbits , after the gas flow reaches an approximate steady state , as input to the coagulation code described next .
our treatment of dust coagulation follows the implementation of @xcite , which in turn is based on the monte carlo model by @xcite .
a population of @xmath33 particles ( we refer to monomers and aggregates jointly as particles ) is assumed to be uniformly distributed inside an abstract volume @xmath34 .
the collision rate between particles @xmath35 and @xmath36 is calculated as @xmath37 where @xmath38 is the collision cross - section , @xmath39 and @xmath40 are the respective particle radii , and @xmath41 is the relative speed between the two particles ( described below ) . in this coagulation algorithm , we track the radius of any particle @xmath42 , @xmath43 , through the enlargement factor @xmath44 , where @xmath45 is the extended volume of the particle ( i.e. , the volume corresponding to the geometric cross - section of the aggregate ) , and @xmath46 is the volume of a monomer of radius @xmath47 .
we then have @xmath48 .
we also track the evolution of the particle mass @xmath49 .
the particle pair @xmath50 that will be involved in each collision is chosen using random numbers and partial sums of the collision rate @xmath51 .
we treat two possible outcomes of a binary collision : particles either merely stick without further restructuring , or are compacted at the expense of internal voids .
these possibilities occur when the kinetic energy of collision ( @xmath52 ) is less than the critical energy to initiate compaction ( @xmath53 ) , or exceeds it , respectively @xcite . to conserve the total number @xmath33 of particles in the volume @xmath34 after a collision ,
one of the remaining @xmath54 particles is randomly chosen and duplicated .
the duplication procedure requires that the volume @xmath34 be increased to preserve the spatial density of dust , @xmath55 . in order to include the effect of the evolving porosity of aggregates formed in collisions , the enlargement factor @xmath56 of an aggregate is calculated according to eqs .
( 15 ) or ( 17 ) of @xcite , depending on whether @xmath57 or @xmath58 , respectively .
the relative speed @xmath41 in eq .
( [ eq : cij ] ) has contributions from brownian motion and from turbulence .
we use an expression for the turbulent relative speed based on eq .
( 10 ) of @xcite .
we insert into that expression values of the gas turbulent velocity , turbulent viscosity ( given by magnetic and hydrodynamic stresses ) and gas density , all obtained from the mhd simulation described above .
these values are taken from four different regions in the vicinity of the planet , as shown in fig .
[ fig : xymap ] , which portrays the system at the end of our mhd simulation .
the contours represent the gas surface density , obtained by averaging the gas density over the vertical extent of the box .
the gap opened by the planet is clearly defined .
the regions r1 through r4 are each divided into 128 subregions of @xmath59 grid cells each .
we call these subregions macrocells .
the gas variables are averaged inside each macrocell , and are then fed to the monte carlo coagulation code . note that although the volume @xmath34 where each particle population resides is effectively _ associated _ to each macrocell , it is _ not _ the volume of the macrocell .
the particles do not have position coordinates associated with them .
we set the number of particles @xmath33 in the volume @xmath34 to @xmath60 , and we use two different initial conditions for the particle sizes : @xmath47=1 @xmath0 m , and @xmath47 distributed according to the mathis - rumpl - nordsieck ( mrn ) distribution of dust grains in the interstellar medium , @xmath61 @xcite . in the latter case ,
the range of monomer radii is 0.5 @xmath62 m . in both cases ,
the monomer bulk density is @xmath63=3 g @xmath64 .
figure [ fig : radprofs ] shows radial profiles of various quantities associated with the mhd flow .
these quantities have been averaged over time ( the last 12 orbits of the simulation ) and over the @xmath7 and @xmath8 directions ( regions for which @xmath65 are not included ) .
the vertical gray bars mark the radial position of the reference regions . in fig .
[ fig : radprofs]a , the gas density , normalized by the initial density , exhibits a drop of a factor of @xmath66 inside the gap opened by the planet with respect to the surrounding gas .
panel c shows a corresponding factor - of-8 drop in the maxwell stress , which is the main contributor to the turbulent viscosity in accretion disks @xcite .
figure [ fig : radprofs]b shows the turbulent gas velocity in units of the gas sound speed .
overall , turbulent gas speeds are relatively high , reflecting the turbulent strength for our case of a non - vanishing initial magnetic flux . finally , fig .
2d reveals that the mean vertical magnetic field has two maxima at either side of the planet s radial position , close to the gap edges .
this is in contrast to the centrally - peaked profile of @xmath67 measured by @xcite , for a lower planet mass of @xmath68 ( @xmath69 at 3 au in the mmsn ) , modeled with half the numerical resolution . the coagulation evolution of the particle populations with an initial monomer radius of 1 @xmath0 m is shown in fig .
[ fig : aggradmon ] .
each curve color refers to one of the four regions of fig .
[ fig : xymap ] . figure [ fig : aggradmon]a shows the mean radius of the aggregates , normalized by the initial monomer radius @xmath70 @xmath0 m .
sticking without restructuring occurs more effectively than compaction for grains subject to the flow conditions in region r4 ( green curve ) , where gas velocities are relatively low , as can be seen in fig .
[ fig : radprofs]b . on the other hand ,
growth is stalled for particles in regions r1 , r2 and r3 ( black , red and blue curves , respectively ) , which have higher turbulent gas velocities . the evolution of porosity , defined by the mass - weighted average enlargement factor @xmath71 as @xmath72 , is shown in fig .
[ fig : aggradmon]b . the compaction of aggregates in regions r2 and r3 is evident , and best understood by looking at the relative impact speeds of the aggregate pairs involved in each collision , as shown in fig . [
fig : aggradmon]c .
the highest relative speeds are generated in region r3 , as expected from the magnitude of the gas velocity there ( fig . [ fig : radprofs]b ) .
the case of an initial mrn - type size distribution is qualitatively similar to the monodisperse case , but with less overall growth .
figure [ fig : aggradmrn]a , which also displays the mean particle radius , shows that aggregates in all four regions have their growth reversed during the span of the calculation , with the largest growth achieved only by particles in region r4 ; they grow by @xmath73% from the initial mean radius of 0.83 @xmath0 m .
the resulting dusty structures are also more compact than in the monodisperse case , with final porosities in region r3 of 30% , compared to 48% in the same region when @xmath74 m .
our results indicate that growth by sticking occurs most effectively outside the gap ( region r4 ) , where the gas density is higher and the gas turbulent velocity is lower ( figs .
[ fig : radprofs]a , b ) . in this region
, we also found that the mean vertical component of the magnetic field , @xmath75 , is relatively low ( fig .
[ fig : radprofs]d ) .
region r2 , located in one of the two planetary wakes , is characterized by lower densities than r4 , but higher gas velocities .
it is interesting that r2 and r3 ( the latter one located in the middle of the low - density gap ) are two regions where @xmath75 has a local maximum and a local minimum , respectively , yet the level of aggregate compaction is similar .
the evolution of the aggregate radius exhibits different trends in the high - gas - density region r4 , for the two dust populations with different initial size distributions . in the case of the initially monodisperse size distribution , the mean aggregate radius in r4 increases with time ( although somewhat more slowly in the last third of the coagulation calculation ) , whereas in the case of the mrn - type initial distribution the mean radius decreases in the last four orbits .
it is possible , then , that at least in some regions in the vicinity of the planet , different initial dust size distributions could lead to significant differences in aggregate growth rates .
this point needs to be addressed in future investigations , using , for example , semi - analytical distributions obtained for coagulation / fragmentation equilibrium @xcite . perhaps the most salient omissions in our mhd calculation are non - ideal effects caused by the varying ionization fraction of the disk .
magnetic field lines can be easily drawn into the planet - induced gap , and the hall effect dominates the onset of the mri depending on the relative orientation between the vertical component of the field and the rotation axis of the disk @xcite .
nevertheless , certain combinations of column density , magnetic field strength , and dust content can render the gap susceptible to the mri in the ideal - mhd regime , such as we assume here .
our use of an orbital radius of 3 au for a jupiter - mass planet is consistent with the grand tack hypothesis .
our mhd gap model can be assumed to represent a snapshot of the grand tack scenario . in this setting , it is instructive to look at our calculations of aggregate porosities within the context of the formation of meteorite components . in the cv chondrite allende , chondrule rims , composed of sub - micron grains ,
may have accreted with high porosities of 7080% @xcite .
if jupiter s migration did indeed occur and was contemporaneous with the formation of meteorite parent bodies , such high porosities may have occurred outside the gap , where turbulent gas velocities were lower . in any case
, further exploration of our simulation parameters is needed to determine the effect of varying magnetic field strengths and geometries , a different equation of state , and electrical charging of dust grains .
the latter effect could delay the growth of rim - forming grains due to electrostatic repulsion @xcite .
our monte carlo calculations of dust growth in and around a gap opened by a jupiter - mass planet , in the presence of turbulence generated by the magnetorotational instability , indicate that aggregate compaction is effective inside the gap , near the gap edge , and in the planetary wake .
this is due to the high turbulent relative speeds between aggregates ( in the range @xmath76 20 600 cm / s ) . in these regions ,
the relative kinetic energies frequently exceed the minimum energy required for restructuring .
the lowest porosities occur inside the gap , with a value of @xmath77 if the initial size distribution of the population is monodisperse ( with the radius of all initial monomers equal to 1 @xmath0 m ) , and @xmath78 if the initial size distribution follows a power law with exponent -3.5 , as in the mrn distribution for interstellar dust .
the most porous aggregates occur in a high - gas - density region outside the gap , where the turbulent relative speeds are lower . in that region , porosities reach extremely high values of @xmath79 for the initially monodisperse population . in the case of the mrn - type population , the porosity reaches @xmath80 .
the outcome of our mhd model of a planet - induced gap , with the planet s mass equal to the mass of jupiter , differs noticeably from some of the results obtained by @xcite .
in particular , the radial profile of the vertical component of the magnetic field is doubly peaked in our simulation , with one peak at either side of the planet s radial position @xmath81 , whereas in the @xcite study the profile exhibits only one peak at @xmath81 , for a planet 6.7 times less massive .
future investigations need to determine the effect of the planetary mass on this radial profile , as the vertical component of the magnetic field plays a key role in the susceptibility of a gap to the mri @xcite .
balbus , s. a. , & hawley , j. f. 1991 , apj , 376 , 214 birnsteil , t. , ormel , c. w. , & dullemond , c. p. 2011 , a & a , 525 , a11 bland , p. a. , howard , l. e. , prior , d. j. , wheeler , j. , hough , r. m. , & dyl , k. a. 2011 , nat .
geosci . , 4 , 244 blandford , r. d. , & payne , d. g. 1982 , mnras , 199 , 883 cuzzi , j. n. , dobrovolskis , a. r. , & champney , j. m. 1993 , icarus , 106 , 102 dong , r. , rafikov , r. r. , & stone , j. m. 2011 , apj , 741 , 57 .
fouchet , l. , maddison , s. t. , gonzalez , j .- f . , & murray , j. r. 2007 , a&a , 474 , 1037 .
gillespie , d. t. 1975 , j. atmos .
, 32 , 1977 .
gonzalez , j .-
f , laibe , g. , maddison , s. t. , pinte , c. , & mnard , f. 2015b , planetary & space science , preprint ( arxiv:1506.00430 ) .
hawley , j. f. , gammie , c. f. , & balbus , s. a. 1995 , apj , 440 , 742 hayashi , c. 1981 , prog .
suppl . , 70 , 35 .
keith , s. l. , & wardle , m. 2015 , mnras , 451 , 1104 lubow , s. h. , seibert , m. , & artymowicz , p. 1999 , apj , 526 , 1001 mathis , j. s. , rumpl , w. , & nordsieck , k. h. 1977 , apj , 217 , 425 matsumoto , t. , & tomisaka , k. 2004 , apj , 616 , 266 matthews , l. s. , land , v. , & hyde , t. w. 2012 , apj , 744 , 8 okuzumi , s. , tanaka , h. , takeuchi , t. , & sakagami , m .- a .
2011 , apj , 731 , 96 .
ormel , c. w. , spaans , m. , & tielens , a. g. g. m. 2007 , a&a , 461 , 215 .
ormel , c. w. , cuzzi , j. n. , & tielens , a. g. g. m. 2008 , apj , 679 , 1588 .
ormel , c. w. 2014 , apj , 789 , l18 .
owen , j. 2014 , apj , 789 , 59 .
paardekooper , s .- j . 2007 , a&a , 462 , 355 .
picogna , g. , & kley , w. 2015 , a&a , preprint ( arxiv:1510.01498 ) .
rafikov , r. r. 2006 , apj , 648 , 666 .
stone , j. m. , gardiner , t. a. , teuben , p. , hawley , j. f. , & simon , j. b. 2008 , apjs , 178 , 137 .
turner , n. j. , lee , m. h. , & sano , t. 2014 , apj , 783 , 14 .
walsh , k. j. , morbidelli , a. , raymond , s. n. , obrien , d. p. , & mandell , a. 2011 , nature , 475 , 206 .
zhu , z. , stone , j. m. , & rafikov , r. r. 2013 , apj , 768 , 143 .
zhu , z. , stone , j. m. , rafikov , r. r. , & bai , x .-
2014 , apj , 785 , 122 . | we analyze the coagulation of dust in and around a gap opened by a jupiter - mass planet . to this end , we carry out a high - resolution magnetohydrodynamic ( mhd ) simulation of the gap environment , which is turbulent due to the magnetorotational instability . from the mhd simulation ,
we obtain values of the gas velocities , densities and turbulent stresses a ) close to the gap edge , b ) in one of the two gas streams that accrete onto the planet , c ) inside the low - density gap , and d ) outside the gap .
the mhd values are then supplied to a monte carlo dust coagulation algorithm , which models grain sticking and compaction .
we consider two dust populations for each region : one whose initial size distribution is monodisperse , with monomer radius equal to 1 @xmath0 m , and another one whose initial size distribution follows the mathis - rumpl - nordsieck distribution for interstellar dust grains , with an initial range of monomer radii between 0.5 and 10 @xmath0 m .
our monte carlo calculations show initial growth of dust aggregates followed by compaction in all cases but one , that of aggregates belonging to the initially monodisperse population subject to gas conditions outside the gap . in this latter case ,
the mass - weighted ( mw ) average porosity of the population reaches extremely high final values of 98% .
the final mw porosities in all other cases range between 30% and 82% .
the efficiency of compaction is due to high turbulent relative speeds between dust particles .
future studies will need to explore the effect of different planet masses and electric charge on grains . |
the objective of the na61/shine ion program is to search for the critical point of strongly interacting matter and to study the properties of the onset of deconfinement @xcite .
this goal will be pursued by an experimental investigation of nucleus - nucleus , proton - proton and proton - lead integrations at at the cern sps .
na61/shine intends to carry out for the first time in the history of nucleus - nucleus collisions a comprehensive scan in two dimensional parameter space : the size of the colliding nuclei versus the interaction energy @xcite . the project has the potential for an important discovery the experimental observation of the critical point of strongly interacting matter . on top of
that it guarantees precision measurements crucial for the understanding of the properties of the onset of deconfinement .
the synergy of the na61/shine physics program @xcite on physics of strongly interacting matter with cosmic - ray and neutrino programs , as well as the use of the existing accelerator chain and detectors offers the unique opportunity to reach the physics goals in an efficient and cost effective way .
na61/shine plans to perform the necessary measurements using the upgraded na49 apparatus @xcite .
the most essential upgrades are the increase of data taking and event rate by a factor of 10 and the construction of a projectile spectator detector which will improve the accuracy of determination of the number of projectile spectators by a factor of about 20 . the na49 apparatus at the cern sps served during the last 10 years as a very reliable , large acceptance hadron spectrometer and delivered experimental data of high precision over the full range of sps beams ( from proton to lead ) @xcite and energies ( from 20@xmath35 gev to 158@xmath35 gev ) @xcite .
among the most important results from this study is the observation @xcite of narrow structures in the energy dependence of hadron production in central pb+pb collisions .
these structures are found at the low cern sps energies ( 30@xmath3580@xmath35 gev ) and they are consistent with the predictions @xcite for the onset of the deconfinement phase transition .
the questions raised by this observation serve as an important motivation for new measurements with nuclear beams proposed in the sps energy range at the cern sps and also envisaged at bnl rhic @xcite , fair sis-300 @xcite as well as nica mpd @xcite .
these programs are to a large extent complementary .
however , specific questions addressed in the na61/shine proposal can be best studied by na61/shine at the cern sps .
[ na61-setup ] the na61/shine experiment uses a large acceptance hadron spectrometer at the cern - sps for the study of the hadronic final states produced in interactions of various beam particles ( @xmath36 , p , c , s and in ) with a variety of fixed targets at the sps energies .
the layout of the na61/shine set - up is shown in fig .
[ na61-setup ] .
the main components of the current detector were constructed and used by the na49 experiment @xcite .
the main tracking devices are four large volume time projection chambers ( tpcs in fig . [ na61-setup ] ) , which are capable of detecting up to 70% of all charged particles created in the studied reactions . two of them , the vertex tpcs ( vtpc-1 and vtpc-2 ) , are located in the magnetic field of two super - conducting dipole magnets ( maximum bending power of 9 tm ) and two others ( mtpc - l and mtpc - r ) are positioned downstream of the magnets symmetrically with respect to the beam line .
one additional small tpc , the so - called gap tpc ( gtpc ) , is installed on the beam axis between the vertex tpcs .
the setup is supplemented by time of flight detector arrays two of which ( tof - l / r ) were inherited from na49 and can provide a time measurement resolution of @xmath37 ps . in a 2007 pilot run
data for the t2k neutrino experiment on p+c interactions at 31 gev / c were recorded @xcite . for this run a new forward time of flight detector ( tof - f )
was constructed in order to extend in these low - multiplicity events the acceptance for pion and kaon identification to low momenta ( @xmath38 gev / c ) .
the tof - f wall is installed downstream of tpc - l and mtpc - r ( see fig . [ na61-setup ] ) , closing the gap between the tof - r and tof - l walls .
furthermore , one super - module of the projectile spectator detector ( psd ) was installed downstream of tof - f and tested @xcite .
the required energy resolution @xmath39 was reached .
a major step forward in the detector performance , i.e. the tpc readout and daq upgrade , was achieved for the 2008 run . during the run
the upgrade was tested .
it results in an increase of the data rate by a factor of about 10 compared to the na49 rate .
the 2008 run was cut short due to the lhc incident .
compatibility with the i - lhc schedule requires the use of _ secondary _ ion beams for the na61/shine ion program .
feasibility studies are in progress and first results indicate that the secondary ion beam properties are sufficient to reach the physics goals of na61/shine .
the na61/shine data taking plan is presented in table [ beam2 ] .
the data taking program starts with @xmath40 and @xmath41 beams for the t2k and cosmic ray experiments .
three energy scans with the secondary ion beams are planned with nuclei of mass numbers of @xmath42 10 , 30 and 100 . in addition , energy scans of p+p and p+pb collisions will be performed .
they will include high statistics runs at 158 gev dedicated for a study of hadron production at large transverse momenta . * * l r r r r r beam+target & energy & year & days & physics & status + & ( @xmath35 gev ) & & & & + * p+c(t2k ) & * 30 & * 2007 & * 30 & * t2k , c - r & _ performed + * p+c(t2k ) & * 30 & * 2009 & * 21 & * t2k , c - r & _ recommended + * @xmath41+c & * 158 , 350 & * 2009 & * 14 & * c - r & _ recommended + * p+p & * 10 , 20 , 30 , 40 , 80 , 158 & * 2009 & * 49 & * cp&od & _ recommended + * p+p & * 158 & * 2010 & * 77 & * high p@xmath43 & _ recommended + * @xmath44(30 + 30 ) & * 10 , 20 , 30 , 40 , 80 , 158 & * 2011 & * 42 & * cp&od & _ recommended + * p+pb & * 158 & * 2011 & * 42 & * high p@xmath43 & _ recommended + * @xmath44(10 + 10 ) & * 10 , 20 , 30 , 40 , 80 , 158 & * 2012 & * 42 & * cp&od & _ to be discussed + * p+p & * 10 , 20 , 30 , 40 , 80 , 158 & * 2012 & * 42 & * cp&od & _ recommended + * @xmath44(100 + 100 ) & * 10 , 20 , 30 , 40 , 80 , 158 & * 2013 & * 42 & * cp&od & _ to be discussed + _ * * * * * _ * * * * * _ * * * * * _ * * * * * _ * * * * * _ * * * * * _ * * * * * _ * * * * * _ * * * * * _ * * * * * the data sets planned to be recoded by na61/shine for the ion program and those recorded by na49 are compared in fig . [ runs ] .
[ comp ] the exciting and rich physics which can be studied in nucleus - nucleus collisions at the cern sps energies motivates groups from bnl , jinr and fair to propose experimental studies which will complement the cern sps program .
two fixed target programs ( cern sps @xcite and fair sis-300 @xcite ) and two programs with ion colliders ( bnl rhic @xcite and jinr nica @xcite ) are foreseen .
the basic parameters of the future programs are summarized in fig .
the sps and rhic energy range covers energies from below up to well above the energy of the onset of deconfinement ( @xmath4430@xmath35 gev in the fixed target mode ) .
thus these machines are well suited for the study of the properties of the onset of deconfinement and the search for the critical point .
the top energies of nica and sis-300 are just above the energy of the onset of deconfinement .
the physics at these machines will thus focus on the study of the properties of dense confined matter close to the transition to the qgp .
this is illustrated in fig .
[ phase_acc ] which shows the coverage of the new programs in the baryon chemical potential together with the existing data and physics benchmarks in the @xmath45 plane .
* measurements of identified hadron spectra in a broad rapidity range , which in particular , allow to obtain the mean hadron multiplicities in full phase - space , * measurements of the total number of the projectile spectator nucleons including free nucleons and nucleons in nuclear fragments , * high event rate in the full sps energy range including the lowest energies , * high flexibility in selecting the nuclear mass number ( thanks to the secondary ion beam option ) and energy ( thanks to the sps features ) of the projectile ions .
the importance of the physics questions addressed by the sps and rhic programs requires two independent , partly complementary measurements .
thus , both planned experimental programs should be performed .
* acknowledgments : * this work was supported by the virtual institute vi-146 of helmholtz gemeinschaft , germany , korea research foundation ( krf-2008 - 313-c00200 ) , the hungarian scientific research fund ( otka 68506 ) , the polish ministry of science and higher education ( n n202 3956 33 ) , the federal agency of education of the ministry of education and science of the russian federation ( grant rnp 2.2.2.2.1547 ) and the russian foundation for basic research ( grant 08 - 02 - 00018 ) , the ministry of education , culture , sports , science and technology , japan , grant - in - aid for scientific research ( 18071005 , 19034011 , 19740162 ) , swiss nationalfonds foundation 200020 - 117913/1 and eth reseach grant th-01 07 - 3 .
n. antoniou _ et al . _
[ na61/shine collaboration ] , cern - spsc-2006 - 034 .
m. gazdzicki _ et al .
_ [ na61/shine collaboration ] , pos c * pod2006 * , 016 ( 2006 ) . s. afanasev _ et al . _
[ na49 collaboration ] , nucl .
instrum .
a * 430 * , 210 ( 1999 ) . c. alt _ et al . _
[ na49 collaboration ] , phys .
lett . *
94 * , 052301 ( 2005 ) .
p. dinkelaker [ na49 collaboration ] , j. phys .
g * 31 * , s1131 ( 2005 ) . c. alt _ et al . _
[ na49 collaboration ] , eur .
j. c * 45 * , 343 ( 2006 ) .
s. v. afanasiev _ et al . _
[ the na49 collaboration ] , phys .
c * 66 * , 054902 ( 2002 ) .
m. gazdzicki _ et al . _
[ na49 collaboration ] , j. phys .
g * 30 * , s701 ( 2004 ) .
m. gazdzicki and m. i. gorenstein , acta phys .
b * 30 * , 2705 ( 1999 ) .
g. s. f. stephans , _ critrhic : the rhic low energy program _ ,
arxiv : nucl - ex/0607030 .
p. senger , t. galatyuk , d. kresan , a. kiseleva and e. kryshen , pos c * pod2006 * ( 2006 ) 018 .
a. n. sissakian , a. s. sorin and v. d. toneev , arxiv : nucl - th/0608032 .
n. abgrall _ et al . _
[ na61/shine collaboration ] , cern - spsc-2008 - 018 , cern - spsc - sr-033 , jul 2008 .
m. golubeva _ et al .
_ , nucl . instrum .
a * 589 * , 268 ( 2009 ) . | the na61/shine experiment at the cern sps aims to discover the critical point of strongly interacting matter and study properties of the onset of deconfinement .
these goals will be reached by measurements of hadron production properties in nucleus - nucleus , proton - proton and proton - lead interactions as a function of collision energy and size of the colliding nuclei .
furthermore , na61/shine will perform numerous precision measurements needed for neutrino ( t2k ) and cosmic - ray ( pierre auger observatory and kascade ) experiments .
this paper summarizes physics arguments for the na61/shine ion program and presents the status and plans of the experiment .
ion program of na61/shine at the cern sps marek gazdzicki for the na61/shine collaboration + _ institut fr kernphysik , university of frankfurt , frankfurt , germany + _ _ jan kochanowski university , kielce , poland + _ the na61/shine collaboration n. abgrall@xmath0 , a. aduszkiewicz@xmath1 , b. andrieu@xmath2 , t. anticic@xmath3 , n. antoniou@xmath4 , a. g. asryan@xmath5 , b. baatar@xmath6 , a. blondel@xmath0 , j. blumer@xmath7 , l. boldizsar@xmath8 , a. bravar@xmath0 , j. brzychczyk@xmath9 , s. a. bunyatov@xmath6 , k .- u .
choi@xmath10 , p. christakoglou@xmath4 , p. chung@xmath11 , j. cleymans@xmath12 , d. a. derkach@xmath5 , f. diakonos@xmath4 , w. dominik@xmath1 , j. dumarchez@xmath2 , r. engel@xmath7 , a. ereditato@xmath13 , g. a. feofilov@xmath5 , z. fodor@xmath8 , m. gadzicki@xmath14 , m. golubeva@xmath15 , k. grebieszkow@xmath16 , f. guber@xmath15 , t. hasegawa@xmath17 , a. haungs@xmath7 , m. hess@xmath13 , s. igolkin@xmath5 , a. s. ivanov@xmath5 , a. ivashkin@xmath15 , k. kadija@xmath3 , n. katrynska@xmath9 , d. kielczewska@xmath1 , d. kikola@xmath16 , j .- h .
kim@xmath10 , t. kobayashi@xmath17 , v. i. kolesnikov@xmath6 , d. kolev@xmath18 , r. s. kolevatov@xmath5 , v. p. kondratiev@xmath5 , a. kurepin@xmath15 , r. lacey@xmath11 , a. laszlo@xmath8 , s. lehmann@xmath13 , b. lungwitz@xmath19 , v. v. lyubushkin@xmath6 , a. maevskaya@xmath15 , z. majka@xmath9 , a. i. malakhov@xmath6 , a. marchionni@xmath20 , a. marcinek@xmath9 , m. di marco@xmath0 , i. maris@xmath7 , v. matveev@xmath15 , g. l. melkumov@xmath6 , a. meregaglia@xmath20 , m. messina@xmath13 , c. meurer@xmath7 , p. mijakowski@xmath21 , m. mitrovski@xmath19 , t. montaruli@xmath22 , st .
mrwczyski@xmath23 , s. murphy@xmath0 , t. nakadaira@xmath17 , p. a. naumenko@xmath5 , v. nikolic@xmath3 , k. nishikawa@xmath17 , t. palczewski@xmath21 , g. palla@xmath8 , a. d. panagiotou@xmath4 , w. peryt@xmath16 , a. petridis@xmath4 , r. planeta@xmath9 , j. pluta@xmath16 , b. a. popov@xmath6 , m. posiadala@xmath1 , p. przewlocki@xmath21 , w. rauch@xmath24 , m. ravonel@xmath0 , r. renfordt@xmath19 , d. rhrich@xmath25 , e. rondio@xmath21 , b. rossi@xmath13 , m. roth@xmath7 , a. rubbia@xmath20 , m. rybczynski@xmath23 , a. sadovsky@xmath15 , k. sakashita@xmath17 , t. schuster@xmath19 , t. sekiguchi@xmath17 , p. seyboth@xmath23 , k. shileev@xmath15 , a. n. sissakian@xmath6 , e. skrzypczak@xmath1 , m. slodkowski@xmath16 , a. s. sorin@xmath6 , p. staszel@xmath9 , g. stefanek@xmath23 , j. stepaniak@xmath21 , c. strabel@xmath20 , h. stroebele@xmath19 , t. susa@xmath3 , i. szentpetery@xmath8 , m. szuba@xmath16 , a. taranenko@xmath11 , r. tsenov@xmath18 , r. ulrich@xmath7 , m. unger@xmath7 , m. vassiliou@xmath4 , v. v. vechernin@xmath5 , g. vesztergombi@xmath8 , z. wlodarczyk@xmath23 , a. wojtaszek@xmath23 , j .- g .
yi@xmath10 , i .- k . yoo@xmath10 @xmath26cape town university , cape town , south africa + @xmath27eth , zurich , switzerland + @xmath28fachhochschule frankfurt , frankfurt , germany + @xmath29faculty of physics , university of sofia , sofia , bulgaria + @xmath30forschungszentrum karlsruhe , karlsruhe , germany + @xmath31institute for nuclear research , moscow , russia + @xmath32institute for particle and nuclear studies , kek , tsukuba , japan + @xmath33jagiellonian university , cracow , poland + @xmath34joint institute for nuclear research , dubna , russia + @xmath8kfki research institute for particle and nuclear physics , budapest , hungary + @xmath2lpnhe , university of paris vi and vii , paris , france + @xmath10pusan national university , pusan , republic of korea + @xmath3rudjer boskovic institute , zagreb , croatia + @xmath21soltan institute for nuclear studies , warsaw , poland + @xmath5st .
petersburg state university , st .
petersburg , russia + @xmath11state university of new york , stony brook , usa + @xmath23jan kochanowski university in kielce , poland + @xmath4university of athens , athens , greece + @xmath25university of bergen , bergen , norway + @xmath13university of bern , bern , switzerland + @xmath19university of frankfurt , frankfurt , germany + @xmath0university of geneva , geneva , switzerland + @xmath1university of warsaw , warsaw , poland + @xmath16warsaw university of technology , warsaw , poland + |
the cosmic microwave background ( cmb ) carries detailed information about the high energy physical processes occurred in the early universe .
most probably , the microphysics still hidden to our knowledge left traces that have been stretched out to large and observable scales by a period of accelerated expansion ; at decoupling between matter and radiation , they imprinted anisotropies in the cmb .
this is the reason of the contemporary theoretical and experimental efforts to understand the cmb physics .
the theory of the cmb anisotropies has been deeply explored in the past ( see @xcite and references therein ) and , recently , it has been casted in a complete and organic form @xcite . at the same time , many experiments are at work to explore the cmb anisotropies toward smaller and smaller angular scales ( see @xcite for reviews ) ; this experimental enterprise will culminate with the planck mission of the next decade , that will provide the whole sky temperature and polarization anisotropy map down to a minimum detectable perturbation of one part over 1 million and with an angular resolution of about @xmath4 @xcite .
according to the inflationary phenomenology , a scalar field ( the inflaton ) slowly rolls toward the minimum of its potential , giving the non - zero vacuum energy responsible for the expansion itself .
the quantum fluctuations are thought to arise from the vacuum in a curved background ; they are stretched out to large scales by the inflationary expansion itself , and set up the seeds of the cosmological perturbations we observe today ( see @xcite for reviews ) . however , even adopting this inflationary scenario , things are still unclear for what concerns the release of the energy stored in the inflaton into ordinary matter and radiation , the so called reheating ( or _ pre_heating ) era @xcite .
the oscillations of the inflaton around its minimum , combined with the coupling to other fields , can restore high energy symmetries that have to be broken to reach our low energy minimum ; consequently , a post - inflationary generation of topological defects may arise , and this occurrence is at the present under investigation @xcite .
also , during inflation itself many fundamental fields may act on stage and the effective potential may have several minima separated by potential barriers . if this is the case , tunneling phenomena occur , and the nucleated bubbles are stretched out to large scales as the ordinary quantum fluctuations ( see @xcite for reviews ) ; at reheating the energy stored in the shells is converted into matter and radiation and bubbly traces may be left in the density distribution ( this possibility , with different points of view , has been considered in the last decade @xcite ) .
suppose that one of these relics from very high energy physics is plunged from some very early time into cosmic matter and radiation , no matter of its composition , that could be scalar field or cosmic fluid or other .
it generates perturbations around itself , in particular in the photon - baryon fluid .
if also it intersects the last scattering surface ( lss , the place of origin of the cmb ) , these perturbations become anisotropies that we could observe today .
these are expected to be well recognizable , since in most cases such seed is a spatially limited structure , very different from the diffuse fluctuations of the pure slow - roll inflation ; technically speaking , such signal would be strongly non - gaussian and non - scale - invariant .
also , such structures are expected to possess ( approximate ) symmetries , like a bubble or a monopole ( spherical ) and a string ( cylindrical ) .
their detection in the cmb anisotropies would be the first observational evidence of the existence of high energy symmetries , and this hope is precisely the motive of this work .
i develop here some useful formulas for the cmb perturbations and anisotropies from symmetric structures ; the results are _ independent _ from the particular seed , the only characterization being its symmetry , that i take here spherical or cylindrical .
i perform some numerical integrations using these formulas and adopting toy symmetric seeds , in order to investigate the geometrical and dynamical properties of their own cmb perturbations and anisotropies . in forthcoming works
i will compute the cmb anisotropies from realistic relics left from high energy physics in the early universe ; a pretty example , valid simply for large bubbles in the density distribution , may be found in @xcite .
as already mentioned , the treatment of the cmb inhomogeneities has been casted recently in a complete and organic form , the total angular momentum method @xcite . in turn , it is based on the general treatment of the linear cosmological perturbations @xcite ; i perform the calculations in this frame , respecting the notations as much as possible .
the cmb perturbations involve temperature ( @xmath5 in the following ) and polarization , that is expressed via the stokes parameters @xmath6 and @xmath7 describing linear polarization . for a given fourier mode specified by the @xmath8 vector , it is convenient to express the relevant quantities in a frame in which the @xmath9 direction is the polar axis ( the @xmath9-frame in the following ) .
the reason is that , in the new frame , the scalar , vector and tensor components of the perturbed metric quantities are coupled respectively to the @xmath10 indexes of the spherical harmonics @xcite . of course
, transforming back to the real space , the @xmath9-frame quantities must be expressed in the fixed laboratory frame ( the @xmath11-frame in the following ) . for a given fourier mode @xmath8 ,
@xmath6 is the difference in temperature fluctuations polarized in the @xmath12 and @xmath13 directions ( @xmath14 and @xmath15 being the usual angles in spherical coordinates ) ; @xmath7 is the same difference where the axes have been rotated by 45@xmath16 around the photon propagation direction .
equivalently , @xmath6 and @xmath7 may be seen as the expansion coefficients of the polarization tensor into the pauli matrices @xmath17 and @xmath18 , defined on the basis vectors @xmath12 and @xmath13 in the @xmath9-frame .
the background friedmann robertson walker ( frw ) metric is @xmath19 where @xmath20 is the conformal time and @xmath21 the spatial curvature ; i will assume a flat @xmath22 background in this work .
the perturbed metric tensor is @xmath23 where @xmath24 represents the background .
since @xmath25 , a gauge freedom reduces the number of physically significant quantities in the perturbation metric tensor ; in this work i adopt the generalized newtonian gauge in which the two scalar perturbed metric component are @xmath26 and @xmath27 @xcite .
the cmb perturbations depend on the spacetime point and on the photon propagation direction @xmath28 , so an appropriate normal mode expansion is needed : @xmath29 @xmath30 @xmath31\ , \ ] ] where @xmath32 are convenient basis matrices for the polarization tensor . @xmath33 and
@xmath34 include both spatial and angular functions ; the spatial ones are the eigenmodes of the laplacian in the metric ( [ frw ] ) : @xmath35 the angular functions are instead spherical harmonics . in the case of flatness ( @xmath22 )
the laplace equation ( [ lap ] ) gives plane waves , and the expression of the normal modes becomes @xmath36 @xmath37 as a difference with respect to @xcite , the notation @xmath38 , @xmath39 has been used to underline that , for each @xmath8 mode , all the quantities in ( [ glms],[glmt ] ) , as well as the expansion coefficients in ( [ dttdeveloped],[qudeveloped ] ) , are expressed in the @xmath9-frame ; as customary , the expansion coefficients of the stokes parameters @xmath40 have been decomposed into real and imaginary parts . throughout this work , in order to characterize the polarization within symmetric seeds , i make use of the useful definition of _ polarization direction _
@xcite , given entirely in terms of @xmath6 and @xmath7 as follows .
it is easy to see that , due to the rotation properties of the pauli matrices , the angle @xmath41 goes into @xmath42 under a rotation by @xmath15 around @xmath28 ; thus it defines a fixed axis on the plane orthogonal @xmath28 , that is the polarization direction .
the underlying cosmological inhomogeneities move the cmb perturbations and are encoded in the expansion coefficients in ( [ dttdeveloped],[qudeveloped ] ) . before going to the content of this work , it is useful to point out the following important distinction .
the fourier transform of any perturbation quantity @xmath43 may be written as @xmath44 it is gaussian if the phases in @xmath45 are random ; specifically in these hypothesis , the statistics is completely described by the power spectrum , @xmath46 .
also it is scale - invariant if the modulus depends only on the scale ( @xmath47 ) in such a way that the power associated to each one is the same at the horizon reenter . on the contrary ,
cmb anisotropies from sources like the ones considered here are non - gaussian and non - scale - invariant ; their symmetries , encoded in precise properties of both modulus and phases in ( [ ft ] ) , are their unique sign in the cmb .
moreover , i do not require that they are dominant for structure formation .
an high resolution cmb map could contain the unambiguous imprint of one single symmetric seed existing at decoupling plunged in a global gaussian signal ; even if the power spectrum does not contain its sign at all , that would be enormously interesting ! the work is organized as follows .
sections ii and iii contain the analysis of the cmb perturbations in spherical and cylindrical symmetry respectively .
section iv contains the method for the computation of the cmb polarization and temperature anisotropies as they would appear on the sky . in section
v the results from numerical integrations are shown .
finally , section vi contains the conclusions .
it is easy to see that spherical structures may be scalar only , and thus are described by the @xmath48 modes of the linear expansion ; there is no way to comb the hair of a sphere in such a way to obtain a spherical distribution , and this prevents spherical structures to be made of genuinely vector ( or tensor ) components .
thus i drop the @xmath49 index in the following , and consider flat space geometry , @xmath22 .
the problem to solve is the following : at a conformal time @xmath50 , a perfect cmb detector is placed in a point @xmath51 nearby a primordial spherical structure ; what s the cmb perturbations carried by photons scattered on a direction @xmath28 ? the center of the coordinate frame is placed at the center of the spherical seed .
its fourier transform depends only on the wavevector modulus @xmath52 and it is therefore the same for any axes orientation : @xmath53 first , let us find the consequence of ( [ sphere ] ) on the pure temperature perturbation @xmath54 .
the expansion coefficients @xmath55 in ( [ dttdeveloped ] ) are proportional to the fourier transformed perturbation ( see section v ) and do not depend at all on the orientation of the perturbation in the @xmath9-frame , simply because it is spherical : they depend on @xmath52 only . consequently ,
posing @xmath56 in ( [ dttdeveloped ] ) the @xmath55 coefficients may be extracted from the angular integral .
thus let us face the pure geometric quantity @xmath57 where the argument of the spherical harmonics in ( [ glms ] ) has been shown ( see appendix a ) .
the integral ( [ int1 ] ) is easily computed expanding the plane wave into bessel and legendre functions @xmath58 and employing the useful relation ( [ legendre ] ) with @xmath59 , @xmath60 and @xmath61 .
the result is : @xmath62 this expression gives the cmb temperature perturbation at any time for the most general spherical perturbation , encoded in the fourier integral .
the dependence on @xmath28 and @xmath63 has been factored out , and enters only in the legendre polynomials argument @xmath64 .
this is an expected feature of this spherical case : for example , focus on the @xmath65 term , better known as the doppler effect ( @xmath66 is essentially the velocity of baryons @xcite ) ; the motion of each particle in this spherical case is radial of course ; then , since this legendre polynomial is just @xmath64 , photons propagating on the direction @xmath28 pick up the usual doppler cosine contribution at the scattering point .
let us face now the polarization for a spherical seed .
a first simplification is that the scalar perturbations excite the @xmath67 modes only @xcite , so we can drop the @xmath68 terms in the following .
then , as before , the @xmath67 coefficients depend on @xmath52 only , so they can be extracted from the angular integral . as a difference from the temperature case , the _ tensor _ spherical harmonics describe now the angular dependence in ( [ glmt ] ) ; fortunately they admit , for @xmath48 , a simple expression in terms of the elementary legendre polynomials , as it is demonstrated in appendix a : @xmath69 where the @xmath70 index has been suppressed since it makes no difference in the @xmath48 case .
focus now on the @xmath39 matrices .
they have to be expressed in terms of the fixed @xmath11-frame matrices @xmath71 .
this is obtained performing a rotation around the @xmath28 axis in order to make the @xmath12 and the @xmath13 vectors in the @xmath9-frame coincident with the laboratory ones : the rotation angle is essentially the angular coordinate of the projection of @xmath9 into the plane orthogonal to @xmath28 . for simplicity , but without any loss of generality , let s orient the @xmath11-frame so that @xmath28 is the polar axis ; then , it is easy to see that the rotation angle is simply @xmath72 , where @xmath73 is just the @xmath15 coordinate of @xmath8 in the @xmath74frame ; thus , from elementary rotation properties of the pauli matrices , the expression of @xmath39 as seen in the @xmath74frame is @xmath75 the integral in ( [ qudeveloped ] ) has now the following form : @xmath76 moreover , it is useful to employ the expansion ( [ wavebessel ] ) together with the addition relation ( [ addition ] ) with @xmath77 and @xmath78 : @xmath79 the integral ( [ intpol1 ] ) in @xmath80 can now be calculated : the @xmath81 phases in ( [ kpolarlab ] ) select the @xmath82 terms respectively above ; once this is done , the integral on @xmath83 is simple using the spherical harmonics orthogonality ( [ orthoplm ] ) ; the final result is @xmath84 @xmath85 where @xmath86 is the angular coordinate of the projection of the @xmath51 vector on the plane orthogonal to @xmath28 .
let s check out the meanings of ( [ intpol ] ) .
again the dependence on @xmath28 and @xmath63 has been completely extracted from the fourier integral ; really the matrices @xmath71 , basis for the polarization tensor , are outside the sum on @xmath87 and multiply appropriate phases : this makes easy the following geometric consideration . if we choose the @xmath11-axes so that @xmath88 , the matrix in ( [ intpol ] ) is simply @xmath89 and the polarization quantities results in a pure @xmath6 term ; thus ( [ intpol ] ) gives the difference in the polarization amplitudes relative to the axes displayed in the upper panel of figure 1 , one lying on the plane formed by the @xmath28 and @xmath63 directions and the other orthogonal to the same plane . with this axes orientation ,
the angle @xmath90 in ( [ pd ] ) is zero : this means that the polarization direction within a spherical seed lies on the plane formed by @xmath28 and @xmath63 , as sketched in figure 1 . as a related important point ,
note the second order legendre polynomial @xmath91 ( the temperature case had @xmath92 ) ; it is meaningful since it guarantees that light propagating radially is not polarized ( @xmath93 ) : the radial propagation in spherical symmetry is an axial symmetric problem , so that no preferred direction exists for the polarization , since it belongs on the plane orthogonal to the symmetry axis . these results ,
together with the temperature ones , completely characterize the cmb perturbation carried by photons moving in a spherical seed , independently from any other specification .
the next section contains the same analysis developed here , but based on cylindrical seeds .
scalars can be arranged cylindrically of course , but also vectors ( vorticity is a vectorial feature ) . consequently , the @xmath94 are allowed . in the vector case however , the generic @xmath8 mode of the fourier transform is a vector of course ; thus its orientation enters in the angular integrals of ( [ dttdeveloped ] ) and ( [ qudeveloped ] ) , that become strongly dependent on the particular seed considered . for this reason ,
again i restrict to the scalar case ( dropping the @xmath49 index ) , and employ flat frw , @xmath22 .
the fourier transform of a cylindrically symmetric quantity @xmath43 may be expressed as @xmath95 where @xmath96 are the component of @xmath8 on the symmetry axis and on the equatorial plane respectively , @xmath97 .
if the seed is invariant under traslations along the symmetry axis ( this case is mentioned as infinite cylindrical seed in the following ) , then the expression is @xmath98 where @xmath99 is the dirac delta .
consequently , the expansion coefficients of ( [ dttdeveloped ] ) and ( [ qudeveloped ] ) depend only on @xmath100 in the first case and on @xmath101 in the second .
first , consider the temperature perturbation .
in cylindrical coordinates @xmath102 and the @xmath55 coefficients come out of the integral in @xmath80 : @xmath103 again i make use of the addition relation ( [ addition ] ) , choosing the polar axis in the @xmath11-frame coincident with the symmetry axis ; i apply it twice , both on the @xmath104 above and on the plane wave , expanded as ( [ auxi ] ) .
paying the price to increase the number of sums , equation ( [ cylint ] ) becomes @xmath105 @xmath106 where of course @xmath107 and @xmath108 .
now , it is manifest that the integral in @xmath80 kills everything except for the @xmath109 terms ; thus , the phase of @xmath110 precisely fits together with the phase of @xmath111 , making them relative .
writing in full the spherical harmonics , the final result is : @xmath112 @xmath113 it is useful to note that writing @xmath114 and extracting the exponential regarding the last term from the integral in @xmath80 , the expression ( [ cylt ] ) may be simplified : @xmath115 @xmath116 note that the sum over @xmath117 is restricted to the even @xmath118 terms because of equation ( [ legendre0 m ] ) . as an alternative approach , one can give up the expansion of the plane wave in ( [ cylint ] ) , and express its argument as @xmath119 .
as above , the phase of @xmath120 is @xmath121 ; the second factor comes out of the integral , and again fits together with the corresponding phase of @xmath111 .
the advantage of this approach is that the integral in @xmath80 has a note form and the result is : @xmath122 where the function @xmath123 is a combination of the anger and weber functions , defined in appendix b. the expressions corresponding to ( [ cylt],[cyltsimple ] ) and ( [ cyltaw ] ) for an infinite cylindrical structure are simpler because of the effect of the dirac delta ; it eliminates the dependence on @xmath124 , reducing the argument of the exponential in ( [ cylint ] ) to @xmath125 . also the legendre polynomials into the integral in @xmath80 have now to be calculated for @xmath126 , their values being found using ( [ legendre0 ] ) ; from ( [ legendre0 m ] ) only the terms with even @xmath127 and @xmath118 ( and therefore @xmath128 ) survive .
thus , the expression for @xmath54 in this case is similar to ( [ cyltsimple ] ) , but simpler : @xmath129 @xmath130 in the second approach @xmath131 let us check the geometric meanings of the above expressions .
first note how the cylindrical symmetry caused complications , both in the geometric and integral quantities , with respect to the spherical case
. however , again the dependence on @xmath28 and @xmath63 has been separated and factored out .
the symmetry forces the phases of the harmonics with argument @xmath28 and @xmath51 to be relative : for @xmath132 , the perturbation depends , together with the angle between the symmetry axis and @xmath28 , on the direction of the projection of @xmath28 on the equatorial plane with respect to @xmath133 , as it is intuitive in a cylindrical problem ; the pure doppler contribution from the peculiar velocity of photons and baryons ( @xmath66 ) may be easily recognized in the @xmath134 terms . if @xmath51 lies on the symmetry
axis itself the @xmath123 function in ( [ cyltaw ] ) and ( [ cylitaw ] ) reduces simply to @xmath135 , as shown in appendix b. as a final intuitive feature , note how in the case @xmath136 , the cmb perturbation for an infinitely long seed possesses a parity symmetry , @xmath137 , since all the @xmath138 terms vanish , making @xmath87 and @xmath117 even .
let us face now the cmb polarization from cylindrical sources .
as in the previous section , the @xmath67 coefficients come out of the integral in @xmath80 and the tensor harmonics are expressed as in ( [ tensorlegendre ] ) .
then the polarization matrices in the @xmath9-frame have to be expressed in terms of the corresponding ones in the @xmath11-frame : @xmath139 ; as before @xmath140 is the angular coordinate of the projection of the @xmath9 versor on the plane orthogonal to @xmath28 .
it is indicated differently from ( [ kpolarlab ] ) because of the following reason . in the previous section we were dealing with spherical perturbations ; no matter of how the @xmath11-frame axes were oriented .
this freedom allowed us to orient the polar axis as @xmath28 , so that @xmath140 was simply related to the @xmath15 coordinate of @xmath9 .
now things are different : the perturbation source has a preferred axis , and the equatorial plane is therefore different from the polarization plane ( orthogonal to @xmath28 ) ; consequently , @xmath140 depends on @xmath73 in a less simple way , as i write below , and this complicates the computations of course .
highlighting again the integral in @xmath80 , the quantities in ( [ qudeveloped ] ) take the form @xmath141 in spite of its innocent appearance , the integral in @xmath80 is not so available for extracting the dependence on @xmath28 as in the previous cases .
this is due to the expression of @xmath140 ; according to the definition above , and taking as reference axis the intersection between the planes orthogonal to @xmath28 and @xmath142 , its expression is latexmath:[\[\label{cumber } \cos\alpha_{\hat{k}}={\hat{n}\times\hat{z}\cdot\hat{k}\over @xmath140 is related to @xmath73 by the following relation , that may be easily verified : @xmath144 unfortunately , ( in my knowledge ) there is no simple treatment of the angular integral in ( [ cylqudeveloped ] ) with @xmath140 given by ( [ akpk ] ) .
however , there are some interesting and useful particular cases in which computations are simpler .
first , suppose that the photon propagation direction @xmath28 is parallel to the symmetry axis .
thus @xmath145 , ( [ kpolarlab ] ) holds , and the legendre polynomials can be extracted from the integral in @xmath80 , since now @xmath146 . in the first approach
all the task consists in expanding the exponential in ( [ cylqudeveloped ] ) , while the second is straightforward .
the integral precisely kills everything except for the @xmath82 terms : @xmath147 @xmath148 @xmath149 @xmath150 @xmath151 @xmath152 this corresponds to the case sketched in the upper panel of figure 2 , where the propagation direction is parallel to the symmetry axis . as in the spherical case , orienting the @xmath11-axis as in the figure ( so that @xmath88 ) yields an equal contribution from the @xmath70 terms ; the polarization is given by a pure @xmath6 term , giving the difference between the temperature fluctuations of the light polarized in the directions shown in the upper panel of figure 2 ; also , @xmath153 in ( [ pd ] ) , meaning that the polarization direction lies on the plane formed by @xmath28 and @xmath142 ( and it is orthogonal to @xmath28 of course ) . the same quantities for an infinite cylindrical seed are easily gained using the dirac delta ( the sum is restricted to even @xmath87 and @xmath117 from ( [ legendre0 m ] ) ) : @xmath154 @xmath155 @xmath151 @xmath156 just like the spherical case , photons propagating exactly _ on _ the symmetry axis have to be not polarized , since no preferred axis exists on the polarization plane .
let s check that the above results are consistent with this geometric expectation . in equations ( [ cylqu11 ] ) and ( [ cyliqu11 ] ) this is manifest because the only bessel function that would survive on the axis ( @xmath157 ) would be @xmath158 , but it s not present , since @xmath159 . for what concerns equations ( [ cylqu12 ] ) and ( [ cyliqu12 ] ) ,
the @xmath123 function in @xmath157 is trivially @xmath160 as is evident from ( [ useful ] ) .
there is another case of interest for an infinite cylindrical structure : precisely when the @xmath28 direction is orthogonal to the axis . in this case
the polarization plane and the equatorial plane are orthogonal ; than it s easy to see that @xmath140 is @xmath161 or @xmath162 where the constant @xmath161 is simply the angular coordinate of the projection of the equatorial plane into the polarization one : this is simply because , for the effect of the dirac delta , the integration is confined into the equatorial plane @xmath163 .
a necessary step here is to use a note expansion of the second order legendre polynomial in term of the elementary ones @xmath164 where the coefficients @xmath165 are defined in appendix b , equation ( [ numerical ] ) .
the arguments widely applied in this section lead to the following expressions of this interesting case : @xmath166 @xmath167 @xmath168 @xmath169 the restriction to the sum in ( [ cyliquo ] ) comes from the properties of the expansion coefficients @xmath165 in ( [ numerical ] ) and again from ( [ legendre0 m ] ) . despite of the large number of sums , equation ( [ cyliquo ] ) is workable because all the legendre polynomials are calculated in the equatorial plane , @xmath170 ; it will be used for the numerical integrations in section v. both the expressions explicitly show the symmetry of the seed ; choosing the axes on the polarization plane parallel and orthogonal to the symmetry axis ( so that @xmath171 ) implies that ( [ cyliquo ] ) and ( [ cyliquawo ] ) give no distinction between the @xmath70 modes , giving again a pure @xmath6 term ; thus the polarization direction lies in the equatorial plane , as displayed in the upper panel of figure 2 . as the very final observation , note that , in contrast to the case @xmath146 , now photons propagating away from the symmetry axis at @xmath157 _ can _ be polarized ; a numerical demonstration of this occurrence will be given in section v. physically this is because there is a preferred axis on the polarization plane , the symmetry axis itself ; formally , now the @xmath172 term is admitted , so that @xmath158 at @xmath157 in ( [ cyliquo ] ) and @xmath173 in ( [ cyliquawo ] ) survive ; it is straightforward to write down the polarization tensor in this particular case : @xmath174^{2}\cdot \int{k_{+}dk_{+}\over 2\pi}e_{l}(\eta , k_{+ } ) \ \ ( { \rm even}\ l , j)\ ; \ ] ] it depends on nothing more , except for the nature of the infinite seed , encoded in the @xmath67 coefficients .
the equations i have developed here and in the previous section describe cmb perturbations , both for temperature and polarization , around symmetric seeds at a given time specified by @xmath50 . in the next section
i show how to get their appearance on the cmb sky .
the expressions in the previous sections describe the cmb polarization and temperature _ perturbations _ around a symmetric structure , as a function of the conformal time @xmath50 and the geometry of the seed itself . at any time , if a perfect cmb detector is placed around one of the seed analyzed , the measure of the cmb perturbation carried by a photon propagating on a direction @xmath28 would give the appropriate result from the above formulas .
now let s face the computation of the cmb _ anisotropy _ from a symmetric seed .
this requires the convolution of the cmb perturbation with the decoupling history of the universe . according to the current scenario , cmb photons
were last scattered far from us in spacetime , when the scale factor was approximatively one thousandth than now .
such process is described by the last scattering probability between @xmath50 and @xmath175 , function of several cosmological parameters and of the time of course ; its expression in terms of the differential optical depth @xmath176 ( see section v ) is very simple : @xmath177 with the appropriate numbers , the last scattering probability peaks on a spherical corona around us moving away with the light speed of course ; it has present radius and thickness of about @xmath178 and @xmath179 comoving mpc respectively : for its thinness this zone is called last scattering surface ( lss ) .
since it is useful here , i recall that using the conformal time as temporal coordinate is also convenient since a photon last scattered at @xmath50 has to travel a comoving distance @xmath180 to reach our spacetime position , indicated in the following with the subscript @xmath181 .
as mentioned in the introduction , the most known class of primordial perturbations is gaussian and ( nearly ) scale - invariant ; a simplification allowed by this statistics is that the cmb anisotropies have the same spectrum regardless of the position of the observer ( cosmic variance subtracted of course @xcite ) .
the seeds analyzed here represent a radically different cmb anisotropy source ; technically speaking they are non - gaussian and non - scale - invariant . as a consequence of this ,
the _ position _ of the source along the photons path becomes here a physical degree of freedom , and the classification of the various possibilities is essential to predict how the cmb signal from a symmetric seed could appear .
let us start from the spherical symmetry .
as sketched in the lower panel of figure 1 , the cmb anisotropies are completely specified by the comoving distance @xmath182 between the seed center and the lss peak : the latter is defined as the point from which we receive cmb photons with highest probability ( peak of @xmath183 ) on the direction @xmath184 corresponding to the center of the spherical seed ; the observer is far on the right and receives on a direction @xmath28 the cmb photons last scattered inside the spherical perturbation , with probability sketched as a gaussian in the figure .
the whole signal is symmetric with respect to rotations around @xmath184 .
also it is convenient to define the useful angle @xmath14 by @xmath185 it is simply the angle between the photon propagation direction @xmath28 and the direction corresponding to photons coming from the center of the spherical seed in the sky .
a photon last scattered at @xmath50 with direction @xmath28 carries a cmb perturbation computable with the formulas developed in the previous section , that require its radial coordinate @xmath186 ; the latter is completely fixed by @xmath50 , @xmath182 and @xmath14 : @xmath187^{1/2}\ , \ ] ] where @xmath188 and @xmath189 mean lss peak and present conformal times respectively ; in fact , since @xmath180 is just the comoving causal distance covered by a photon last scattered at @xmath50 and reaching us today , gaining equation ( [ los ] ) is matter of simple trigonometry , see figure 1 .
this completes the spherically symmetric case .
once we have specified @xmath182 , gaining the cmb polarization and temperature anisotropies from a spherical seed means performing line of sight integrations for each direction specified by @xmath14 , as it is exposed below .
of course , the appearance on the sky of the cmb temperature and polarization anisotropies from one spherical seed is circular ; more interesting , while nothing forbids photons coming on the @xmath184 direction to carry a temperature perturbation , the geometric constraint treated in section ii forces them to be not polarized .
a nice example of this occurrence can be found in @xcite .
let s face now the case of cmb anisotropies coming from cylindrically symmetric seeds .
first of all , let s define the plane @xmath190 containing the seed symmetry axis and our observation point : the signal is of course symmetric with respect to reflections on this plane .
also let s define a @xmath190 orthogonal versor , @xmath191 , and one along the symmetry axis , @xmath142 , regardless of their direction .
take now a representative point @xmath192 on the symmetry axis ; in the spherical case it was the sphere s center , but here , in principle , it could be any point along the axis : inside the seeds itself , or the axis intersection with the lss peak , or ultimately the point of minimal distance from the observation point .
let s define @xmath193 as the direction of photons coming from @xmath192 and @xmath194 its comoving distance from the lss peak ( of course @xmath195 ) ; these simple geometric quantities are displayed in figure 2 , bottom panel .
now take a photon last scattered at @xmath50 on a direction @xmath28 , described with the usual angles @xmath14 and @xmath15 in the frame defined by @xmath196 , @xmath197 and @xmath198 ( only @xmath199 would be necessary , since the signal does not change for reflections on @xmath190 , look at figure 2 ) .
let s define for a moment @xmath200 and @xmath201 to be the photon scattering point and the observation point as seen by the frame centered in @xmath192 : @xmath202 @xmath203 in order to employ the equations developed in the previous section we need to know @xmath204 and @xmath124 .
this is easily done by expressing @xmath200 as seen in a system with the same axes orientation but centered in @xmath201 : @xmath205 this fixes the quantities needed : @xmath206-\ ] ] @xmath207^{1/2}\ , \ ] ] @xmath208 as expected , the cylindrical symmetry has introduced an angular variable more than the spherical case .
the quantities @xmath204 and @xmath124 defined above allow to employ the formulas developed in the previous section to compute the cmb anisotropy carried by the photon last scattered at @xmath50 on the direction @xmath28 ; of course , for an infinite cylindrical seed only the @xmath204 coordinate is necessary . while anisotropies in the spherical case are characterized by a circular imprint , here their shape may vary with the orientation of the symmetry axis .
if it coincides with @xmath193 , thus including the observation point , the imprint is circular around it , and again polarization anisotropies are absent on the direction corresponding to the symmetry axis itself . in any other case
, both polarization and temperature anisotropies would appear symmetric around a _ line _ in the sky , projection of the seed symmetry axis on the celestial sphere .
finally , the cmb anisotropies both for the spherical and cylindrical cases are obtained through a line of sight integration along the photon s path , convolved with the last scattering probability ( see @xcite ) : @xmath209d\eta\ , \ ] ] @xmath210 at each @xmath50 , ( [ los ] ) and ( [ cylrlos],[cylzlos ] ) give the necessary arguments ( @xmath211 ) to compute the cmb perturbations .
@xmath212 accounts for the sachs - wolfe effect , due to the work spent by the photon climbing out of the potential well ( or hill ) in which it was last scattered ; the time derivatives account for the integrated sachs - wolfe effect , due to the work spent by the same photon crossing the density perturbations on the way toward us .
the following consideration introduces to the next section .
as i have already mentioned , a symmetric seed could be a spatially limited structure , say a monopole or a bubble for the spherical case , or a string for the cylindrical case .
thus also the cmb anisotropy is spatially limited , since the evolution equations may transport the cmb perturbation at most at a sound horizon distance from the source .
therefore , if the perturbed zone does not intersects the lss , meaning that it occupies a spacetime region where @xmath183 is negligibly small , the terms @xmath54 , @xmath6 and @xmath7 above do not give contributions ; in this situation , the seed ca nt signal its presence , except for the integrated sachs - wolfe effect if it lies within our hubble sphere ( a distinctive and fascinating signal in this case arise from cosmic strings @xcite ) .
thus , in order to detect the genuine cmb signal from a symmetric spatially limited seed , we should be lucky with its spacetime location : it should intersect the lss .
let us apply the formulas developed in the previous sections .
i plunge a toy symmetric source in the cosmic fluid at the initial time @xmath213 , computing its evolution by using the linear theory of the cosmological perturbations . at different times during the evolution , some pictures of the corresponding cmb polarization and temperature perturbations are taken .
finally , the computation of the line of sight integrals ( [ dttani ] ) and ( [ quani ] ) simulates the cmb signal as it would appear in an high resolution observation .
first , let us define the initial density perturbations .
for the spherical case , i take a potential energy condensation with a gaussian shape extending on a comoving radial distance @xmath214 : @xmath215\ .\ ] ] for the cylindrical case , i take an infinitely long seed , with a potential energy condensation on the equatorial plane characterized again by a gaussian shape and extending on a scale @xmath216 : @xmath217\ .\ ] ] the normalization constants will be fixed below .
the fourier transforms are easily performed in the frame with origin in the center of the sources : @xmath218\ , \ ] ] @xmath219\ .\ ] ] the evolution equations for fluid and cmb quantities may be obtained by the boltzmann and linearized einstein equations @xcite . a standard cdm scenario is assumed , including cold dark matter ( @xmath220 ) , baryons ( @xmath221 ) , photons ( @xmath222 ) and three families of massless neutrinos ( @xmath223 ) .
all the equations in the following are written in fourier space .
the equations for the matter species are : @xmath224 @xmath225 where @xmath226 and @xmath227 is the differential optical depth ; @xmath228 is the electron number density and @xmath229 is the ionization fraction ( see the last work in reference @xcite for useful fitting formulas ) .
photon equations involve each multipole in the expansions ( [ dttdeveloped ] ) and ( [ qudeveloped ] ) : @xmath230 @xmath231 and for @xmath232 : @xmath233- \dot{\tau}\theta_{l } \ , \
\dot{e}_{l } = k \left[{\sqrt{l^{2}-4}\over 2l-1}e_{l-1}- { \sqrt{(l+1)^{2}-4}\over 2l+3}e_{l+1}\right]- \dot{\tau}e_{l}\ .\ ] ] in newtonian gauge the lowest multipoles are linked to the photon fluid quantities by @xmath234 , @xmath235 and @xmath236 .
massless neutrinos can be treated as photons without the polarization and thomson scattering terms .
finally , the equations for the gravitational potentials are : @xmath237\ , \ ] ] @xmath238 as it is known @xcite , at early times the above system can be solved by using the tight coupling approximation between photons and baryons .
the multipole equations are expanded in powers of @xmath239 .
the only zero order terms are @xmath240 and @xmath66 from ( [ b],[theta01 ] ) , and obey the following equations : @xmath241= k\theta_{0}+k\left(1+{3\rho_{b}\over 4\rho_{\gamma}}\right)\psi\ \ , \ ] ] where @xmath66 is assumed to concide with @xmath242 to the lowest order . increasing the order in @xmath243 the higher multipoles
are given by @xmath244 i integrate in time the system ( [ thetazerothetaone ] , [ tc2],[tcl ] ) until @xmath245 occurs , thereafter integrating the complete equations ; of course , care is taken that the results do not depend at all on this choice .
i take adiabatic initial conditions : at early times @xmath246 ( all the velocity are initially zero ) and the second member in equation ( [ phi ] ) at @xmath213 is proportional for each fourier mode to the initial perturbation spectrum ( [ toyspherek ] ) or ( [ toycylinderk ] ) ; in order to make the following results more clear , the latter is normalized with the density contrast @xmath2 taken in the center of the seed at decoupling .
this choice is not dependent on the particular gauge chosen here , since equation ( [ phi ] ) is gauge invariant @xcite . in the cmb equations , everything is initially zero except for the lowest multipole of the temperature perturbation @xcite : @xmath247 the background evolution is driven by the einstein equation @xmath248 where the index @xmath249 runs over all the fluid species .
now the computation system is ready .
the background parameters describe a standard cdm model ( @xmath250 ) .
figure 3 shows the time evolution of the cmb temperature perturbation from the spherical seed .
the chosen comoving size is @xmath251 mpc , that is well below the effective horizon at decoupling , ( approximatively @xmath252 mpc ) .
the radial profile is shown , and the temperature perturbation has been computed from equation ( [ thetasphere ] ) for photons propagating perpendicularly to the radial direction , as indicated .
the sum converges very rapidly : the heavy line shows the result from the first ten multipoles , while the light one , almost indistinguishable , indicates the result from the only @xmath253 multipole in ( [ thetasphere ] ) .
the figure points out the wave - like behavior of the cmb perturbations .
the initial condition ( panel @xmath249 ) remains unchanged until the horizon crossing , that occurs nearly at equivalence for the chosen size . at this time , in panel @xmath254 , baryons tend to fall into the potential well , and the perturbation amplitude grows .
after that , in panel @xmath255 , an opposite oscillation due to the pressure reaction takes place , pushing the perturbation away from the center .
finally , in panel @xmath182 the perturbation is shown just before decoupling : the oscillatory behavior caused a temperature perturbation wave that is propagating outward .
the wave crest is just at the position of the sound horizon at the time displayed .
this phenomenology is analogous in figure 4 , where the polarization amplitude , computed using equation ( [ intpol ] ) , is shown . at the initial time
no perturbation is visible , since all the fourier modes are outside the horizon . at the horizon
crossing the oscillations begin , producing a well visible polarization wave that travels outward with the cmb sound velocity .
note that , as an important distinction with respect to the temperature case , for the polarization there is no perturbation near the center , at small @xmath186 .
this is a practical realization of the geometric constraint exposed in section ii : photons propagating radially in a spherical density field must be not polarized , since no preferred axis exists on the polarization plane .
figures 5 and 6 show the same analysis on the cylindrical seed and remarkably the same undulatory phenomenology of the spherical case occurs .
the size is @xmath256 mpc and photons propagating perpendicularly to the symmetry and radial directions are considered , from equations ( [ cylit ] ) and ( [ cyliquo ] ) .
again the sum converges very rapidly : the light line in figure 5 corresponds to the @xmath253 terms in ( [ cylit ] ) . at the horizon
crossing , the competition between pressure and gravity generate cmb temperature and polarization waves propagating away from the symmetry axis . just before decoupling ,
panels @xmath182 , temperature and polarization waves are well visible a cmb sound horizon away from the axis of the cylindrical seed . as an interesting feature ,
note how in this case the polarization for photons scattered on the symmetry axis is non - vanishing : this is evident particularly in panel @xmath255 .
the central polarization amplitude is in any case smaller than the mean signal size , since for @xmath257 in ( [ cyliquo ] ) only the @xmath253 term survive .
figure 7 shows the results of the line of sight integrals ( [ dttani ] ) and ( [ quani ] ) , where for simplicity only the spherical case is shown ; i recall that @xmath14 is the angle between the line of sight and the one corresponding to the center of the seed .
the importance of the different positions of the seed with respect to the lss is evident : the solid line shows the signal if the spherical perturbation lies exactly on the last scattering surface , @xmath258 , while the dashed and dotted dashed lines corresponds to the cases @xmath259 mpc and @xmath260 mpc respectively .
the general features pointed of the time evolution have been preserved .
simply , the cmb temperature and polarization waves propagating outward from the spherical seed have been snapped by the decoupling photons .
the anisotropy waves extend on the scale of a cmb sound horizon at decoupling , that is roughly @xmath0 in the sky .
the temperature perturbation contains a central spot , that is absent in the polarization case .
it is important to point out the following considerations .
first , note that the mean amplitude of the signal follows the known expectations @xcite for a linear structure with size @xmath1 and density contrast @xmath2 at decoupling : @xmath3 , roughly ten times stronger than the polarization signal . from the point of view of the dark matter distribution ,
the seed lies in the very central part of the graph , say @xmath261 ( corresponding to less then @xmath179 mpc in figure 3 and 4 ) .
also the amplitude of the waves has the same mean magnitude of the signal coming from the location of the seed ; really , in the polarization from a spherical seed they are the very dominant component of the anisotropy .
thus they must be considered in any simulation aiming at the detection of this kind of signals .
also they could play some role in the structure formation around the seed , since they are physically made of photons and baryons .
besides , from an experimental point of view this undulatory occurrence could help the detection if structure like these ones should really exist .
indeed the cmb signal from a spatially limited seed is extended on the scale of a sound horizon at decoupling even if the size of the seed itself is smaller ; therefore it appears as a series of sub - degree rings centered on the position of the seed ; this can help to discriminate between the signals from point astrophysical sources from genuine cosmological seeds of primordial origin .
also , as it is evident from figure 7 , a marked correlation exists between the temperature and polarization signals .
of course , this would improve the signal to noise ratio for high resolution instruments like planck capable to detect both polarization and temperature anisotropy .
at the present time , very high energy physics is still rather unknown and only theoretically approached .
the breaking of high energy symmetries in the early universe may have left some traces of their occurrence , like topological defects or true vacuum bubbles .
these relics act as seeds for polarization and temperature anisotropies in the cosmic microwave background ( cmb ) , and this work aims at providing a general framework in order to predict their signal .
i have considered the cases of spherical and cylindrical symmetry of the perturbation source ; no other specification characterizes the seed .
i have obtained general formulas describing cmb polarization and temperature perturbations , as a function of time , generated by the most general structures characterized by the mentioned symmetries .
the analysis regards both the pure cmb perturbation nearby the seeds and their cmb anisotropy as observed in our sky .
such expressions explicitly show several nice features to their own cmb imprint .
in spherical symmetry , the polarization and temperature perturbations depend geometrically on the scalar product @xmath64 , where the first is the photon propagation direction and the second the radial versor in the point where cmb is being measured .
i give explicit expressions in which this dependence is factored out of the integral over the fourier perturbations modes .
in particular the polarization direction ( orthogonal to @xmath28 of course ) lies on the plane formed by @xmath28 and @xmath63 . as an important difference between polarization and temperature perturbations , the light propagating from the center of the seed is not polarized , since the radial propagation in spherical symmetry is an axial symmetric problem , so that no preferred axis exists for the polarization ; instead nothing forbids a temperature perturbation . in cylindrical symmetry
the polarization and temperature perturbations depend on the products @xmath262 and @xmath263 , where @xmath142 is the symmetry axis , as well as on the angular difference between the projections of @xmath28 and @xmath63 on the plane orthogonal to @xmath142 ; the @xmath263 dependence is lost if the seed is invariant for traslations along the symmetry axis ( mentioned as infinite in the following ) .
i give formal expressions showing these dependences , and extract them analytically from the fourier integral in the cases of propagation parallel and orthogonal to the symmetry axis . in the first case the polarization direction lies on the plane formed by @xmath28 and @xmath142 ; as for the spherical case , photons traveling exactly on the symmetry axis are not polarized . in the second case , and for an infinite seed ,
the polarization direction is orthogonal to the symmetry axis . for what concerns the cmb anisotropies
as observed in our sky , they are computed with an usual line of sight integration , but the seeds considered here introduce additional variables with respect to the ordinary gaussian case , that specify their position and orientation along the photons path toward us , characterizing their appearance on our cmb sky .
polarization and temperature anisotropies from a spherical seed are circular and specified by the distance @xmath182 between the seed center and the lss peak . as a consequence of the geometric constraints summarized above , cmb polarization anisotropy is absent for photons coming from the center of the seed ; on the other hand , nothing prevents them to possess a temperature perturbation .
anisotropies from a cylindrical seed are specified by the distance @xmath194 between a representative point on the symmetry axis and the lss peak , as well as on the angular orientation of the symmetry axis itself on the plane containing it and the observation point .
anisotropies may appear in different ways .
if the symmetry axis includes the observation point , what we would see is a circular imprint again ; as in the spherical case , cmb polarization anisotropy is absent for photons coming from the center . in any other case , anisotropies would appear symmetric around a line in the sky , projection of the axis on the celestial sphere , thus giving the genuine sign of a cylindrical seed .
i have performed some numerical work on the formulas developed here , adopting toy symmetric sources in order to see the pure cmb processes at work with this kind of seed .
the time evolution of the seed and of its corresponding cmb perturbation is performed from the initial time , and several pictures are taken before decoupling .
the integrations highlight the undulatory behavior of the cmb perturbations . just like a pebble in a pond , the oscillations occurring at the horizon crossing produce temperature and polarization perturbation waves that propagate outward with the cmb sound velocity .
consequently , the cmb anisotropies caused from structures like the ones analyzed here that intersect the last scattering surface extend at least on @xmath0 in the sky , that is the angular scale corresponding to the cmb sound horizon at decoupling ; the signals contain anisotropy waves , each one characterized by its own value of temperature and polarization perturbation .
this component of the signal possesses the same magnitude of the one coming directly from the seed interior .
the mean amplitude roughly follow the known expectations for a linear structure with size @xmath1 and density contrast @xmath2 at decoupling : @xmath3 , roughly ten times stronger than the polarization signal , where @xmath264 is the size of the hubble length at decoupling .
the anisotropy waves coming out of a symmetric spatially limited seed are a unique proof that the seed itself existed well before decoupling ; thus , these waves could allow to distinguish relics from high energy processes of the early universe from point - like astrophysical sources , because of the angular extension and amplitude . also , this phenomenology offers cross correlation possibilities for detectors like planck capable to explore both temperature and polarization cmb sky .
future works will deal with models of real symmetric structures , relics from high energy physics .
these works aim at predicting their appearance on the cmb map itself before than on the anisotropy power spectrum . their detection in the high resolution
cmb maps provided by the microwave anisotropy probe and planck missions in the next decade would be an invaluable insight into the hidden sector of high energy physics .
the first half of this work was performed at the nasa / fermilab astrophysics center .
it was supported by the doe and the nasa grant nag 5 - 7092 .
i warmly thank the hospitality of the theoretical astrophysics group .
also i wish to thank luca amendola and franco occhionero for constant encouragement .
finally , i am grateful to luigi danese and the astrophysics sector at sissa / isas , where this work was completed .
the spherical harmonics are expressed as usual as @xmath265 where the legendre polynomials are defined by @xmath266 where @xmath267 .
legendre polynomials and spherical harmonics obey the orthogonality relations @xmath268 and are eigenmodes of the parity operator : @xmath269 legendre polynomials satisfy the following note recurrence relations : @xmath270 @xmath271 @xmath272 they can be used to gain the value of any legendre polynomials in @xmath273 : @xmath274 @xmath275 also note that @xmath276 in this work i have often used the addition relation for spherical harmonics , given by @xmath277 and the following useful integral relation , that may be verified easily using the addition relation itself : @xmath278 second order legendre polynomials admit the following expansion @xcite : @xmath279 @xmath280 the tensor spherical harmonics are defined in terms of the ordinary ones by @xmath281y_{l}^{m}(\theta,\phi ) \ ; \ ] ] the normalization coefficient may vary in literature .
equation ( [ tensorlegendre ] ) is easily obtained in the following way . from ( [ known ] ) one can immediately sees that @xmath282 also the equality @xmath283 holds by using elementary derivation . using again ( [ known ] ) for @xmath48 and putting ( [ step ] ) and ( [ der ] ) together , the wanted equation is obtained : @xmath284
this appendix contains some useful integration relations .
focus on the integral @xmath285d\phi= \pi [ { \bf j}_{\nu}(\beta ) \pm i{\bf e}_{\nu}(\beta ) ] \ \ , \ \ ( { \rm re } \beta > 0)\ , \ ] ] where @xmath286 and @xmath287 are the anger and weber functions respectively ( see @xcite for useful recurrence relations ) : @xmath288 @xmath289 the two following equalities are easily gained using elementary integration algebra : @xmath290d\phi= \pi [ { \bf j}_{\nu}(\beta ) \pm i{\bf e}_{\nu}(\beta ) ] + \pi e^{\pm i\nu\pi } [ { \bf j}_{-\nu}(\beta ) \mp i{\bf
e}_{-\nu}(\beta ) ] \ , \ ] ] @xmath291d\phi= e^{\pm i\nu\pi /2}\int_{-\pi /2}^{3\pi /2 } \exp [ \mp i(-\nu\phi+\beta\cos\phi)]d\phi\ .\ ] ] if @xmath292 is integer , all the functions in the integrals are periodical on the @xmath293 interval , so as the integrals above do not depend on the starting point .
thus the following equality holds : @xmath294d\phi = \pi e^{\pm im\pi /2 } [ { \bf j}_{m}(\beta ) \mp
i{\bf e}_{m}(\beta ) ] + \pi e^{\mp i\nu\pi/2}[{\bf j}_{-m}(\beta ) \pm
i{\bf e}_{-m}(\beta ) ] = { \bf je}_{\pm}(m,\beta ) \ ; \ ] ] it s valid for @xmath295 integer and re@xmath296 ; the last equality is a pure definition .
note that in the particular case @xmath297 , the above expression reduces simply to @xmath135 .
baccigalupi c. 1998 , ap.j 496 615 , astro - ph/9711095 ; amendola l. , baccigalupi c. & occhionero f. 1998 ap.j .
492 , l5 ; baccigalupi c. 1999 , in preparation bardeen j.m .
1980 phys.rev.d 22 , 1882 ; kodama i. & sasaki m. 1984 progr .
of theor.phys.supp .
78 , 1 born m. & wolfe e. _ principles of optics _ , pergamon press 1980 kolb e.w .
1991 physica scripta t36 , 199 coulson d. , unpublished ph.d .
thesis , princeton university 1994 gradshteyn i.s . & ryzhik i.m . , _ tables of integrals , series , and products _ , academic press 1980 hu w. , seljak u. , white m. & zaldarriaga m. 1997 , phys.rev.d 56 , 596 , astro - ph/9709066 ; hu w. & white m. 1997 phys.rev.d 56 , 596 , astro - ph/9702170 ; hu w. & sugiyama n. 1995 , ap.j .
444 , 489 , astro - ph/9407093 kofman l. , linde a. & starobinsky a. 1997 , phys.rev.d 56 , 3258 la d. 1991 , phys . lett .
b 265 , 232 ; turner m.s . , weinberg e.j . & widrow l.m .
1992 , phys.rev.d 46 , 2384 ; occhionero f. & amendola l. 1994 , phys .
d 50 , 4846 kaiser n. & stebbins a. 1984 , nat .
310 , 391 ; bouchet f. , bennet d. & stebbins a. 1988 , nat .
355 , 410 mukhanov v.f .
, feldman h.a . &
brandenberger r.h .
1992 phys.rep .
215 , 203 lasenby a.n .
, jones a.w . & dabrowski y. , 1998 phil .
a. in press , astro - ph/9810196 ma & bertschinger 1995 , ap.j .
455 , 7 padmanabhan t. _ structure formation in the universe _ ,
cambridge university press 1993 planck surveyor : http://astro.estec.esa.nl/sa-general/projects/planck/ ; + microwave anisotropy probe : http://map.gsfc.nasa.gov/ tkachev i. , khlebnikov s. , kofman l. & linde a. 1998 , phys.lett .
b 440 , 262 , hep - ph/9805209 white m. , krauss l.m . & silk j. 1993 ap.j . 418 , 535 | perturbations in the cosmic microwave background ( cmb ) are generated by primordial inhomogeneities .
i consider the case of cmb anisotropies from one single ordered perturbation source , or seed , existing well before decoupling between matter and radiation .
such structures could have been left by high energy symmetries breaking in the early universe .
i focus on the cases of spherical and cylindrical symmetry of the seed .
i give general analytic expressions for the polarization and temperature linear perturbations , factoring out of the fourier integral the dependence on the photon propagation direction and on the geometric coordinates describing the seed .
i show how the cmb perturbations manifestly reflect the symmetries of their seeds .
in particular , polarization is uniquely linked to the shape of the source because of its tensorial nature .
cmb anisotropies are obtained with a line of sight integration .
they are function of the position and orientation of the seed along the photons path .
this treatment highlights the undulatory properties of the cmb .
i show with numerical examples how the polarization and temperature perturbations propagate beyond the size of their seeds , reaching the cmb sound horizon at the time considered . just like
the waves from a pebble thrown in a pond , cmb anisotropy from a seed intersecting the last scattering surface appears as a series of temperature and polarization waves surrounding the seed , extending on the scale of the cmb sound horizon at decoupling , roughly @xmath0 in the sky .
each wave is characterized by its own value of the cmb perturbation , with the same mean amplitude of the signal coming from the seed interior ; as expected for a linear structure with size @xmath1 and density contrast @xmath2 at decoupling , the temperature anisotropy is @xmath3 , roughly ten times stronger than the polarization .
these waves could allow to distinguish relics from high energy processes of the early universe from point - like astrophysical sources , because of their angular extension and amplitude . also , the marked analogy between polarization and temperature signals offers cross correlation possibilities for the future detection instruments
. it would be interesting to detect these signals in the next @xmath4 cmb map provided by the planck surveyor satellite experiment .
pacs : 98.70.vc 98.80.cq fermilab - pub-98/204-a |
in many real - life applications such as audio processing or medical image analysis , one encounters the situation when given observations ( most likely noisy ) have been generated by several sources @xmath1 that one wishes to reconstruct separately . in this case
, the reconstruction problem can be understood as an inverse problem of unmixing type , where the solution @xmath2 consists of several ( two or more ) components of different nature , which have to be identified and separated . in mathematical terms , an unmixing problem can be stated as the solution of an equation @xmath3 where @xmath4 and @xmath5 but @xmath6 for @xmath7 in the sense that @xmath8 for all @xmath9 and @xmath10 @xmath11 in general , we are interested to acquire the minimal amount of information on @xmath12 so that we can selectively reconstruct with the best accuracy one of the components @xmath13 , but not necessarily also the other components @xmath14 for @xmath15 . in this settings , we further assume that @xmath16 can not be specifically tuned to recover @xmath13 but should be suited to gain universal information to recover @xmath13 by a specifically tuned decoder . a concrete example of this setting is the _ noise folding phenomenon _ arising in compressed sensing , related to noise in the signal that is eventually amplified by the measurement procedure . in this setting , it is reasonable to consider a model problem of the type @xmath17 where @xmath18 is the random gaussian noise with variance @xmath19 on the original signal @xmath20 and @xmath21 is the linear measurement matrix .
several recent works ( see , for instance , @xcite and the references therein ) illustrate how the measurement process actually causes the noise folding phenomenon . to be more specific , one can show that ( [ model_problem ] ) is equivalent to solving @xmath22 where @xmath23 is composed by i.i.d .
gaussian entries with distribution @xmath24 , and the variance @xmath25 is related to the variance of the original signal by @xmath26 .
this implies that the variance of the noise on the original signal is amplified by a factor of @xmath27 . under the assumption that @xmath16 satisfies the so - called restricted isometry property , it is known from the work on the dantzig selector in @xcite that one can reconstruct @xmath28 from measurements @xmath29 as in ( [ model_problem_eq ] ) such that @xmath30 where @xmath31 denotes the number of nonzero elements of the solution @xmath32 the estimate ( [ recon_rate ] )
is considered ( folklore ) nearly - optimal in the sense that no other method can really improve the asymptotic error @xmath33 therefore , the noise folding phenomenon may in practice significantly reduce the potential advantages of compressed sensing in terms of the trade - off between robustness and efficient compression ( given by the factor @xmath27 here ) compared to other more traditional subsampling methods @xcite .
in @xcite the authors present a two - step numerical method which allows not only to recover the large entries of the original signal @xmath12 accurately , but also has enhanced properties in terms of support identification over simple @xmath0-minimization based algorithms . in particular , because of the lack of separation between noise and reconstructed signal components , the latter ones can easily fail to recover the support when the support is not given a priori . however , the computational cost of the second phase of the procedure presented in @xcite , being a non - smooth and non - convex optimization problem , is too demanding to be performed on problems with realistic dimensionalities .
it was also shown that other methods based on a different penalization of the signal and noise can lead to higher support detection rate .
the follow up work @xcite , which addresses the noise folding scenario by means of multi - penalty regularization , provides the first numerical evidence of the superior performance of multi - penalty regularization compared to its single parameter counterparts for problem ( [ model_problem ] ) .
in particular , the authors consider the functional @xmath34 here @xmath35 may all be considered as regularization parameters of the problem .
the parameter @xmath36 ensures the @xmath37coercivity of @xmath38 also with respect to the component @xmath18 . in the infinite dimensional setting the authors presented a numerical approach to the minimization of ( [ mp ] ) for @xmath39 , based on simple iterative thresholding steps , and analyzed its convergence .
the results presented in this paper are very much inspired not only by the above - mentioned works in the signal processing and compressed sensing fields , but also by theoretical developments in sparsity - based regularization ( see @xcite and references therein ) and multi - penalty regularization ( @xcite , just to mention a few ) .
while the latter two directions are considered separately in most of the literature , there have also been some efforts to understand regularization and convergence behavior for multiple parameters and functionals , especially for image analysis @xcite .
however , to the best of our knowledge , the present paper is the first one providing a theoretical analysis of the multi - penalty regularization with a non - smooth sparsity promoting regularization term , and an explicit comparison with the single - parameter counterpart .
in section 2 we concisely recall the pertinent features and concepts of multi - penalty and single - penalty regularization .
we further show that @xmath0-regularization can be considered as the limiting case of the multi - penalty one , and thus the theory of @xmath0-regularization can be applied to multi - penalty setting . in section 3
we recall and discuss conditions for exact support recovery in the single - parameter case .
the main contributions of the paper are presented in sections 4 and 5 , where we extend and generalize the results from the previous sections to the multi - penalty setting . in section 5
we also open the discussion on the set of admissible parameters for the exact support recovery for unmixing problem in single - parameter as well as multi - penalty cases .
in particular , we study the sensitivity of the multi - penalty scheme with respect to the parameter choice .
the theoretical findings and discussion are illustrated and supported by extensive numerical validation tests presented in section 6 .
finally , in section 7 we compare the performance of the multi - penalty regularization and its single - parameter counterpart for compressive sensing problems .
we first provide a short reminder and collect some definitions of the standard notation used in this paper .
the true solution @xmath28 of the unmixing problem is called @xmath31-sparse if it has at most @xmath31 non - zero entries , i.e. , @xmath40 , where @xmath41 denotes the support of @xmath28 .
we propose to solve the unmixing problem with @xmath31-sparse true solution using multi - penalty tikhonov regularization of the form @xmath42 the solution of which we will denote by @xmath43 .
we note that we can , formally , interpret standard @xmath0-regularization as the limiting case @xmath44 , setting @xmath45 obviously , the pair of minimizers of @xmath46 will always be equal to @xmath47 , where @xmath48 minimizes @xmath49 .
let @xmath50 be fixed .
we say that a set @xmath51 is a _ set of exact support recovery for the unmixing problem with operator @xmath16 _ , if there exists @xmath52 , such that @xmath53 whenever the given data @xmath29 has the form @xmath54 with @xmath55 . the parameters @xmath52 for which this property holds are called _
admissible for @xmath56_. specifically , we will study for @xmath57 the sets @xmath58 and the corresponding class @xmath59 the set @xmath60 is a set of exact support recovery , if there exists some regularization parameter @xmath52 , such that we can apply multi - penalty regularization with parameters @xmath61 and @xmath62 ( or single - parameter @xmath0-regularization with parameter @xmath61 in case @xmath44 ) to the unmixing problem and obtain a result with the correct support , provided that the @xmath63-norm of the noise is smaller than @xmath64 and the non - zero coefficients of @xmath28 are larger than @xmath65 .
typical examples of real - life signals that can be modeled by signals from the set @xmath60 can be found in asteroseismology , see for instance @xcite .
we note that this class of signals is very similar to the one studied in @xcite .
the main difference is that we focus on the case where the noise @xmath18 is bounded only componentwise ( that is , with respect to the @xmath63-norm ) , whereas @xcite deals with noise that has a bounded @xmath66-norm for some @xmath67 .
additionally , we allow the noise also to mix with the signal @xmath28 to be identified in the sense that the supports of @xmath18 and @xmath28 may have a non - empty intersection .
in contrast , the signal and the noise are assumed to be strictly separated in @xcite . throughout the paper
we will several times refer to the sign function @xmath68 , which we always interpret as being the set valued function @xmath69 if @xmath70 , @xmath71 if @xmath72 , and @xmath73 $ ] if @xmath74 , applied componentwise to the entries of the vector @xmath48 .
we use the notation @xmath75 to denote the restriction of the operator @xmath16 to the span of the support of @xmath28 .
additionally , we denote by @xmath76 the complement of @xmath77 , and by @xmath78 the restriction of @xmath16 to the span of @xmath79 .
we note that the adjoints @xmath80 and @xmath81 are simply the compositions of the adjoint @xmath82 of @xmath16 with the projections onto the spans of @xmath77 and @xmath79 , respectively . as a first result
, we show that the solution of the multi - penalty problem simultaneously solves a related single - penalty problem .
[ le : single ] the pair @xmath43 solves ( [ eq : multi ] ) if and only if @xmath83 and @xmath84 solves the optimization problem @xmath85 with @xmath86 and @xmath87 we can solve the optimization problem in in two steps , first with respect to @xmath18 and then with respect to @xmath12 . assuming that @xmath12 is fixed , the optimality condition for @xmath18 in reads @xmath88 that is , for fixed @xmath12 , the optimum in with respect to @xmath18 is obtained at @xmath89 inserting this into the tikhonov functional , we obtain the optimization problem @xmath90 using , we can write @xmath91 thus the optimization problem for @xmath12 simplifies to @xmath92 now note that @xmath93 inserting this equality in , we obtain the optimization problem @xmath94 which is the same as . as a consequence of lemma [ le : single ]
, we can apply the theory of @xmath0-regularization also to the multi - penalty setting we consider here .
in particular , this yields , for fixed @xmath95 , estimates of the form @xmath96 provided that @xmath28 satisfies a source condition of the form @xmath97 with @xmath98 for every @xmath99 , and the restriction of the mapping @xmath100 to the span of the support of @xmath28 is injective ( see @xcite ) .
additionally , it is easy to show that these conditions hold for @xmath100 provided that they hold for @xmath16 and @xmath62 is sufficiently large .
the main focus of this paper is the question whether multi - penalty regularization allows for the exact recovery of the support of the true solution @xmath28 and how it compares to single - penalty regularization . because , as we have seen in lemma [ le : single ] , multi - penalty regularization can be rewritten as single - parameter regularization for the regularized operator @xmath100 and right hand side @xmath101
, we will first discuss recovery conditions in the single - parameter setting . in order to find conditions for exact support recovery , we first recall the necessary and sufficient optimality condition for @xmath0-regularization : [ le : opt ] the vector @xmath48 minimizes @xmath102 if and only if @xmath103 using this result , we obtain a condition that guarantees exact support recovery for the single - penalty method : [ le : support ] we have @xmath104 , if and only if there exists @xmath105 such that @xmath106 this immediately follows from lemma [ le : opt ] by testing the optimality conditions on the vector @xmath48 given by @xmath107 for @xmath108 and @xmath109 else .
our main result concerning support recovery for single - parameter regularization is the following : [ pr : cond_single ] assume that @xmath75 is injective and that @xmath110 then the set @xmath60 defined in is a set of exact support recovery for the unmixing problem whenever @xmath111 moreover , every parameter @xmath52 satisfying @xmath112 is admissible on @xmath113 .
first we note that the injectivity of @xmath75 implies that the mapping @xmath114 is invertible .
thus the condition actually makes sense .
moreover , the inequality is necessary and sufficient for the existence of @xmath61 satisfying .
now let @xmath115 and assume that @xmath61 satisfies .
we denote @xmath116 and define @xmath117 because @xmath118 , it follows from the second inequality in that @xmath119 and therefore @xmath120 thus @xmath121 actually satisfies the equation @xmath122 and thus @xmath123 which is the first condition in lemma [ le : support ] .
it remains to show that @xmath124 however , @xmath125 and thus @xmath126 now the first inequality in implies that this term is smaller than @xmath61 .
thus @xmath121 satisfies the conditions of lemma [ le : support ] , and thus @xmath104 . in the case where @xmath127 is the identity operator ,
the conditions above reduce to the conditions that @xmath128 and @xmath129 .
since @xmath0-regularization in this setting reduces to soft thresholding , these conditions are very natural and are actually both sufficient and necessary : since the noise may componentwise reach the value of @xmath64 , it is necessary to choose a regularization parameter of at least @xmath64 in order to remove it .
however , on the support @xmath77 of the signal , the smallest values of the noisy signal value are at least of size @xmath130 .
thus they are retained as long as the regularization parameter does not exceed this value . for more complicated operators @xmath16 , the situation is similar , i.e. , a too small regularization parameter @xmath61 is not able to remove all the noise , while a too large one destroys part of the signal as well .
the exact bounds for the admissible regularization parameters , however , are much more complicated .
we now consider the setting of multi - penalty regularization for the solution of the unmixing problem . applying lemma [ le : single ]
, we can treat multi - penalty regularization with the same methods as single - penalty regularization . to that end
, we introduce the regularized operator @xmath131 in particular , we have with the notation of lemma [ le : support ] that @xmath132 and @xmath133 . as a first result
, we obtain the following analogon to lemma [ le : support ] : [ le : support_multi ] we have @xmath134 , if and only if there exists @xmath105 such that @xmath135 applying lemma [ le : opt ] to the single - penalty problem , we obtain the conditions @xmath136 now the claim follows from the equalities @xmath137 since the proof of proposition [ pr : cond_single ] only depends on the optimality conditions and the representation of the data as @xmath138 , we immediately obtain a generalization of proposition [ pr : cond_single ] to the multi - penalty setting .
[ pr : cond_multi ] assume that @xmath139 is such that @xmath140 then the set @xmath60 is a set of exact support recovery for the unmixing problem in the multi - penalty setting whenever @xmath141 moreover , all the pairs of parameter @xmath142 satisfying and @xmath143 are admissible on @xmath60 . the proof is analogous to the proof of proposition [ pr : cond_single ] .
we note that the condition @xmath144 implies the analogous inequality for @xmath145 provided that @xmath62 is sufficiently large .
similarly , if @xmath61 satisfies the conditions in proposition [ pr : cond_single ] that guarantee admissibility on @xmath113 , the pair @xmath142 will satisfy the conditions for admissibility in proposition [ pr : cond_multi ] provided that @xmath62 is sufficiently large . the converse , however , need not be true : if the pair @xmath142 is admissible for exact support recovery on @xmath146 with multi - penalty regularization , it need not be true that the single parameter @xmath61 is admissble for the single - penalty setting as well .
examples where this actually happens can be found in section [ se : valid ] ( see in particular table [ tb : cond ] ) .
as a consequence of propositions [ pr : cond_single ] and [ pr : cond_multi ] , we obtain that the condition @xmath147 is sufficient for @xmath148 to be a set of exact support recovery for the unmixing problem , provided that the ratio @xmath149 is sufficiently large ; the condition for the single - parameter case can be extracted from by setting @xmath44 , in which case @xmath145 reduces to @xmath16 .
now define the signal - to - noise ratio of a pair @xmath150 as @xmath151 that is , @xmath152 is the ratio of the smallest significant value of the signal @xmath12 , and the largest value of the noise @xmath18 .
denote moreover @xmath153 then the inequality implies that multi - penalty regularization with parameter @xmath62 allows for the recovery of the support of @xmath31-sparse vectors @xmath12 from data @xmath154 provided the signal - to - noise ratio of the pair @xmath150 satisfies @xmath155 whenever the signal - to - noise ratio is larger than @xmath156 , we can recover the support of the vector @xmath12 with _ some _ regularization parameter @xmath61 .
there are , however , upper and lower limits for the admissible parameters @xmath61 , given by inequality . in order to visualize them ,
we consider instead the ratio @xmath157 defining @xmath158 and @xmath159 we then obtain the condition @xmath160 for exact support recovery .
if the ratio @xmath161 is smaller than @xmath162 , then it can happen that some of the noise @xmath18 is not filtered out by the regularization method .
on the other hand , if @xmath161 is larger than @xmath163 , then some parts of the signal @xmath28 might actually be lost because of the regularization .
we note that the function @xmath164 is piecewise linear and concave , and @xmath165 .
thus the region of admissible parameters defined by is a convex and unbounded polyhedron .
moreover , we have that @xmath166 .
additionally , we note that the behaviour of the function @xmath167 near infinity is determined by the term @xmath168 if this value is small , then the slope of the function @xmath169 for large values of @xmath170 is large , and thus the set of admissible parameter grows fast with increasing signal - to - noise ratio .
if , on the other hand , @xmath171 is large , then the set of admissible parameters is relatively small even for large signal - to - noise ratio . thus @xmath171 can be reasonably interpreted as the sensitivity of multi - parameter regularization with respect to parameter choice .
the larger @xmath171 is , the more precise the parameter @xmath61 has to be chosen in order to guarantee exact support recovery .
the main motive behind the study and application of multi - penalty regularization is the problem that @xmath0-regularization is often not capable to identify the support of signal correctly ( see @xcite and references therein ) . including the additional @xmath172-regularization term , however
, might lead to an improved performance in terms of support recovery , because we can expect that the @xmath172-term takes care of all the small noise components . in order to verify this observation ,
a series of numerical experiments was performed , in which we illustrate for which parameters and gaussian matrices the conditions for support recovery derived in the previous section were satisfied .
in addition , we studied whether the inclusion of the @xmath172-term indeed increases the performance . in a first set of experiments , we have generated a set of 20 gaussian random matrices of different sizes and have tested for each three - dimensional subspace spanned by the basis elements whether the condition is satisfied , first for the single - penalty case , and then for the multi - penalty case with different values of @xmath62 .
the results for matrices of dimensions 30 times 60 and 40 times 80 , respectively , are summarized in table [ tb : cond ] and figure [ fi : cond ] .
as to be expected from the bad numerical performance of @xmath0-regularization in terms of support recovery , the inequality fails in a relatively large number of cases , especially when the discrepancy between the dimension @xmath173 of the vectors to be recovered and the number of measurements @xmath174 is quite large .
for instance , in the case @xmath175 and @xmath176 , the condition most of the time failed for more than half of the three - dimensional subspaces .
in contrast , the corresponding condition for multi - parameter regularization fails in the same situation only for about an eighth of the subspaces if @xmath177 , and in even fewer cases for @xmath178 . for other combinations of dimensionality of the problem and number of measurements ,
the situation is similar . introducing the additional @xmath172-penalty term
always allows for the exact support reconstruction on a larger number of subspaces than single - penalty regularization .
additionally , the results indicate that the number of recoverable subspaces increases with decreasing @xmath62 .
.[tb : cond ] percentage of 3-sparse subspaces for which the condition failed .
the condition was tested on samples of 20 gaussian random matrices of dimensions 30 times 60 ( upper table ) and 40 times 80 ( lower table ) .
other combinations of dimensionality and number of measurements showed qualitatively similar results . [ cols=">,^,^,^,^ " , ]
the figure reports the results of two different decoding procedures of the same problem , where the circles represent the noisy signal and the crosses represent the original signal .
_ upper figure : _ results with single - penalty regularization .
_ lower figure : _ results with multy - penalty regularization .
note that multi - penalty regularization allows for a better reconstruction of the support of the true signal @xmath28 . , title="fig:",scaledwidth=95.0% ] the figure reports the results of two different decoding procedures of the same problem , where the circles represent the noisy signal and the crosses represent the original signal .
_ upper figure : _ results with single - penalty regularization . _
lower figure : _ results with multy - penalty regularization .
note that multi - penalty regularization allows for a better reconstruction of the support of the true signal @xmath28 .
, title="fig:",scaledwidth=95.0% ] + | inspired by several real - life applications in audio processing and medical image analysis , where the quantity of interest is generated by several sources to be accurately modeled and separated , as well as by recent advances in regularization theory and optimization , we study the conditions on optimal support recovery in inverse problems of unmixing type by means of multi - penalty regularization .
we consider and analyze a regularization functional composed of a data - fidelity term , where signal and noise are additively mixed , a non - smooth , convex , sparsity promoting term , and a quadratic penalty term to model the noise .
we prove not only that the well - established theory for sparse recovery in the single parameter case can be translated to the multi - penalty settings , but we also demonstrate the enhanced properties of multi - penalty regularization in terms of support identification compared to sole @xmath0-minimization . we additionally confirm and support the theoretical results by extensive numerical simulations , which give a statistics of robustness of the multi - penalty regularization scheme with respect to the single - parameter counterpart .
eventually , we confirm a significant improvement in performance compared to standard @xmath0-regularization for compressive sensing problems considered in our experiments . |
ultrashort pulse characterization @xcite is a key technology underpinning much of ultrafast science .
progress in the field continues to occur on many fronts extending the range of measurable pulse parameters , such as duration , wavelength , and complexity , improving the sensitivity and single - shot capability , and providing spatial , in addition to temporal , resolution .
less easily quantified but equally important issues are precision , robustness , ease of use and the diagnosis , identification and quantification of errors .
innovations on this front have included simplified geometrical setups @xcite , theoretical analysis of the influence of various errors @xcite , and the exploitation of redundant data to identify misalignment or violations of the assumptions on the pulse properties @xcite .
spectral shearing interferometry ( ssi ) , a popular class of techniques whose best - known example is spectral phase interferometry for direct electric field reconstruction ( spider ) @xcite is widely used for its efficient and direct retrieval acquisition and computation requirements , which scale linearly with the time - bandwidth product of the unknown pulse , and its natural suitability to broadband optical input .
its direct retrieval formula makes analysis of the influence of various errors a straightforward matter and its one dimensional data encoding makes the acquisition of multiple shots , and subsequent analysis of their shot - to - shot variations , convenient and rapid .
geometrically simplified arrangements have also been developed @xcite .
the common feature of all ssi techniques is the interference of two spectrally sheared replicas of the unknown pulse .
this introduces two parameters , the shear and the interferometric reference phase , both of which must be calibrated to a precision which increases with the time - bandwidth product of the unknown pulse .
novel geometries for the convenient determination of these parameters @xcite are therefore of interest , supplementing the well - established techniques @xcite .
additionally , whilst typical raw ssi data contains some redundant information @xcite enabling self - consistency checking , this can be supplemented by taking several measurements at different shears which should result in identical retrieved fields .
most currently known spider variants @xcite are implemented as single shear devices .
changing the shear involves moving parts which requires additional calibration and prevents single shot multiple shear operation .
a significant recent development in this context is chirped ancillae arrangement- ( car-)spider @xcite , in which an upconverted replica of the unknown pulse with spatially varying frequency is produced through the phasematching properties of a long crystal .
this replica is split , and two - dimensional spectral interferometry is performed between itself and a laterally inverted copy , yielding an interferogram in which the shear varies with position .
the zero - shear line encodes the reference phase , whilst the shear calibration can be performed by blocking each arm individually .
car - spider is therefore an attractive device in the light of the previous discussion .
however , the collinear long - crystal geometry limits both the measureable bandwidth and the range of pulse lengths .
also in the long - crystal arrangement the spatial carrier of the interferogram depends on frequency , requiring an additional calibration for broadband pulses . in this paper
, we present a device which we dub spatially encoded arrangement for car - spider ( sea - car - spider ) , in which spatially chirped ancillae are produced by a diffraction grating and split before undergoing upconversion with the test pulse .
this removes the restriction on the pulse parameters imposed by the long - crystal geometry and also renders the spatial carrier independent of frequency , so that the reference phase is truly self - calibrated .
furthermore , the arrangement permits a convenient determination of all calibration parameters in a single step with much greater precision than car - spider , as well as a direct verification of the ancillae monochromaticity which is important for accurate measurement of long pulses .
the test pulse @xmath0 $ ] , propagating along the @xmath1-axis , undergoes sum - frequency generation with two spatially chirped ancillae , one of which has been laterally inverted about the @xmath2-axis , so that their local frequency is @xmath3 , where @xmath4 is the upconversion frequency common to both ancillae at @xmath5 .
the ancillae wavevectors also form an angle @xmath6 in the @xmath7-@xmath1-plane , symmetrically about the @xmath1-axis .
the resulting interferogram , re - imaged onto two - dimensional spectrometer , is of the form @xmath8
@xmath9)i(\omega-\alpha x)}\exp \left [ i \gamma(\omega , x ) \right]\ ] ] @xmath10 where @xmath11 is the positive - spatial - frequency ac sideband with spatial carrier @xmath12 , @xmath13 is the interferogram phase , and @xmath14 is the zero - shear calibration phase , which is theoretically zero due to symmetry but may be present due to imperfect imaging .
note that the spatial carrier @xmath15 is independent of the signal frequency @xmath16 because the spatial tilt is applied to the ancillae , rather than the upconverted replica as in car - spider . at all @xmath7-positions @xmath17 ,
the interferogram phase contains the finite differences @xmath18 i.e. spectral shearing interferometry data of shear @xmath19 . the reconstruction procedure is as follows : the phase - gradient @xmath13 , contained in the ac sideband at spatial frequency @xmath12 , is extracted by 2d fourier filtering , as with sea - spider @xcite
. then the calibration phase must be determined .
if the spectrum has no nulls , then @xmath13 is well defined along @xmath20 and @xmath14 can be obtained directly at this position and subtracted from @xmath13 @xcite .
alternatively , in the case of spectral nulls at @xmath5 , or for increased precision , the calibration phase can also be obtained by weighted averaging along the @xmath7 axis .
we use the fact that , except for the calibration phase , @xmath13 is an odd function about @xmath5 .
therefore by averaging accross a symmetric interval about @xmath5 the calibration phase can be isolated . in practice
this requires a method , which is robust against @xmath21 phase discontinuities .
we use @xmath22\ ] ] where @xmath23 is the principal value of the argument and @xmath24 denotes a one - dimensional phase unwrapping routine which removes @xmath21 discontinuities . after subtracting the calibration phase ,
the standard concatenation procedure can be performed at any @xmath7 as desired .
the spectral amplitude of the unknown pulse , modulated by the phasematching efficiency and the spectrometer response , is also obtained from the @xmath5 signal .
therefore , all data is recorded near the upconversion frequency , enabling a specificially optimized spectrometer to be used . even though a single sea - car - spidergram contains tens of independent datasets ,
online reconstruction of each of these is still possible owing to the fast algebraic reconstruction algorithm of spider .
we note that as with filter - spider @xcite and car - spider , our new device does not rely on temporal stretching of the ancillae to ensure the quasi - monochromaticity .
thus in sea - car - spider , which is free from the pulse length restrictions of the long - crystal arrangement , strongly chirped pulses can be characterized without the need for correction of the reconstructed phase @xcite .
the spatial carrier in sea - car - spider enables an intuitive interpretation of the fringe pattern , because , as we now show , the fringe contours follow the phase derivative of the unknown pulse .
let @xmath25 be a fringe contour , defined by @xmath26 .
then @xmath27}{{\mathrm{d}}\omega}= \frac{\partial \gamma[\omega , x{\textsubscript{con}}(\omega)]}{\partial \omega } + \frac{\partial \gamma[\omega , x{\textsubscript{con}}(\omega)]}{\partial x } \frac{\partial x{\textsubscript{con}}(\omega)}{\partial \omega}=0.\ ] ] in ( [ eq : gamma ] ) we approximate the finite differences with a derivative , and write @xmath28 the spatial derivative is @xmath29 the second term on the right hand side of represents the spatial carrier and is much larger than the first term otherwise the sidebands would not be distinct in the fourier domain .
we neglect the first term and substitute ( [ eq : derivative2 ] ) into ( [ eq : fringecontourdef ] ) , yielding @xmath30 therefore the fringe contours follow the spectral phase gradient scaled by the local shear and the inverse of the spatial carrier .
this is illustrated in fig .
[ fig : seacartracestheory ] , which shows calculated traces for a transform - limited pulse and for pulses with quadratic , cubic and quartic polynomial spectral phases .
note that for a transform - limited pulse ( fig .
[ fig : seacartracestheory ] ( a ) ) the fringe contours are flat , reflecting the frequency independence of the spatial carrier .
the fringe pattern provides a simple and useful visual tool for aligning a laser qualitatively without the need for running a reconstruction algorithm .
the experimental implementation of sea - car - spider is depicted in fig . [
fig : schematicsetup ] . the test pulse tp is split with beamsplitter bs1 and focussed into the nonlinear crystal @xmath31 with a cylindrical lens cl with focal length @xmath32{mm}$ ] .
the delay - line dl1 is only for coarse adjustment of the delay between ancillae and test pulse and only requires initial adjustment .
the ancillae are prepared as follows : the reflected beam from bs1 is dispersed by a grating of pitch @xmath33{mm^{-1}}$ ] and angle of incidence @xmath34{^{\circ}}$ ] .
the spherical lens l1 is positioned one focal length @xmath35{mm}$ ] away from the grating and the crystal is mounted @xmath36 further downstream . between the lens and the crystal the ancilla beam
is split at 50/50 beamsplitter bs2 and one arm is spatially inverted .
the coarse delay line dl2 is used to achieve temporal overlap between the ancillae and needs no subsequent adjustment .
the @xmath31 crystal ( bbo ) is placed in the back fourier - plane of l1 , where the spatial chirp of the ancillae is @xmath37 } \omega{\textsubscript{up}}^2 / ( 2\pi c f{\textsubscript{1 } } \lambda ) $ ] where @xmath38 is the diffracted angle of the beam from the grating . a half - wave retarder @xmath39 is introduced into the ancilla arm to ensure type - ii upconversion with the ancilla polarization aligned to the @xmath40-axis and the tp polarization to the @xmath41-axis of the crystal , respectively .
after spatial filtering ( sf ) the sum - frequency beams are re - imaged onto a 2d imaging spectrometer by lens l2 with focal length @xmath42{mm}$ ] .
the spectrometer s entrance slit is oriented in the dispersion plane of the ancillae and the signal is internally dispersed perpendicular to this plane .
the spectrometer provides astigmatism - free imaging over wide spectral and spatial extent @xcite .
the signal is recorded with a @xmath43{pixel}$ ] cmos detector with adc resolution .
single shot data is recorded at a rate of limited by memory and data - transfer capacities .
it should be noted that sea - car - spider can be operated as zero additional phase ( zap ) device by using a cylindrical mirror and reflection of the test - pulse off bs1 rather than transmission .
in order to obtain a single - step calibration of the shear slope @xmath44 , as well as the central upconversion frequency @xmath4 and the @xmath7-origin , the tp arm was blocked and the spatial filter sf temporarily removed . in this configuration , the second harmonic of the ancillae , as well as their sum - frequency signal , are obtained .
[ fig : calibration](a ) shows the calibration data obtained from the average of 50 single shot acquisitions .
a linear fit of the slopes of the two ancillae provides @xmath45{mrad / fs / mm}$ ] . from the crossing point of the two ancillae the @xmath7-origin and @xmath4
are obtained .
we determine the @xmath7-origin to be pixel @xmath46 .
the upconversion frequency is found to be @xmath47{rad
all errors are estimated using the covariances of the individual fit parameters for both ancillae .
the sum - frequency mixing between the ancillae appears as the vertical line at frequency @xmath4 in fig . [ fig : calibration](a ) .
its position with respect to the crossing point is a useful check of the axial alignment of the crystal - to - spectrometer reimaging system .
if l2 is correctly relay imaging the crystal plane to the spectrometer entrance slit , then all three lines cross at the same point .
the spectral width of the ancillae can also be obtained from the calibration data of fig .
[ fig : calibration](a ) .
this is useful in verifying that the ancillae are sufficiently monochromatic , a basic requirement of spectral shearing interferometry . in the case of sea - car - spider , where many shears are available
, the spectral width of the ancillae determines the minimum shear from which a meaningful reconstruction can be performed . here
, we obtain a monochromaticity of @xmath48{mrad / fs}$ ] fwhm ( in the second harmonic or @xmath49{mrad / fs}$ ] ( in the fundamental ) , granting a maximal time window for test pulses of @xmath50{ps}$ ] .
the ancillae spectral width is @xmath51 , where @xmath52 is the fwhm spot size produced by lens l1 .
the monochromaticity can therefore be further increased by using a higher pitched grating and a tighter focus .
we note that a similar calibration procedure was performed in the original car - spider arrangement . however , because the upconversion was performed using the asymmetric phasematching in a long crystal it is impossible to block the test pulse and hence the shear calibration was performed using the upconverted signal pulses .
as we now illustrate , the sea - car - spider procedure offers greater precision because the second harmonics of the ancillae are much more narrowband than the test pulse . in car - spider
, the calibration would be obtained by recording the individual signals separately and obtaining their displacement by cross - correlation , disregarding regions of low intensity .
one such fit is shown in fig .
[ fig : calibration](b ) .
the residuals of the slope fits are directly compared in fig .
[ fig : calibration ] ( c ) . using this method
we get shear slopes of @xmath53{mrad / fs / mm}$ ] . from the signal
fits we further determine the @xmath7-origin to be @xmath54{pixels}$ ] .
( the upconversion frequency is not determined by this method ) .
as expected due to the greater spectral width of the signals the errors in determining the calibration parameters are significantly larger compared to the new calibration procedure presented above .
this emphasizes the precision advantage of the calibration offered by the ancilla preparation in sea - car - spider . .
( b ) position of @xmath7-origin ( zero shear row ) determined from the crossing point .
( c ) upconversion frequency @xmath4 . ]
we investigated the sensitivity of the calibration against fluctuations of the beam - pointing and displacement .
our device preserves vertical symmetry ( @xmath2-axis in our coordinate system ) and relay images all beams .
it turns out that beam - pointing and displacement changes in the vertical direction do not affect the device calibration .
vertical beam pointing fluctuations cause the signal to move off the slit reducing the signal intensity .
horizontal beam displacements cause a change in the spatial carrier , which does not affect the calibration or reconstruction .
a beam - pointing change @xmath55 in the horizontal plane causes the two spatially chirped ancilla fields to move into opposite directions due to the lateral inversion in the ancilla preparation setup .
this leads to a shift in the upconversion frequency of @xmath56 to first order , whilst the zero shear position is not affected .
due to the low ( near normal ) diffraction angle from the grating the variation of the shear slope @xmath44 is negligible .
we tested these conclusions by recording 50 single shot calibration traces and determining the calibration parameters shear slope @xmath44 , zero shear row and upconversion frequency @xmath4 as shown in figure [ fig : calmultishots ] . confirming our findings we find that @xmath44 , and the @xmath7-origin are constant within the uncertainty of our fits .
the upconversion frequency @xmath4 varies by consistent with typical beam pointing fluctuations of our chirped pulse amplification ( cpa ) system .
in this section we show the characterization of pulses from a home - built cpa system .
the laser delivers pulses centered around at a repetition rate of @xmath57{khz}$ ] . after filtering ( @xmath21 scale ) .
the spatial carrier has been removed for clarity.,scaledwidth=90.0% ] a typical single shot sea - car - spidergram is shown in fig .
[ fig : trace](a ) .
the regions of low intensity are caused by dead rows in our detector and imperfections in the upconversion crystal .
confirming the calibration of the upconversion frequency , the signal is centered around @xmath58{nm}$ ] which is equivalent to @xmath59{rad / fs}$ ] or @xmath60{nm}$ ] where @xmath61 is the pulse centre frequency . to aid isolating the interferometric component from the baseband term in the fourier - filtering procedure we use a high spatial carrier of @xmath62{rad / pix}$ ] ( where @xmath63 is the nyquist rate ) resulting in fine fringes .
a lineout at @xmath58{nm}$ ] is shown in fig .
[ fig : trace](b ) . the two - dimensional discrete fourier transform of these data is shown in fig .
[ fig : trace](c ) ; the black rectangle indicates the filter region .
the phase of the filtered sideband is shown in fig .
[ fig : trace](d ) with the spatial carrier removed for clarity .
the symmetry of the signal about @xmath5 is evident .
{ mrad / fs}$ ] .
( a ) spectral domain with amplitude extacted from the trace at @xmath5 and the 3 phases correspondiong to each shear . ( b )
temporal domain : temporal intensities and phases for each shear . ] after subtracting the calibration phase and extracting the amplitude as described in section [ sec : sea - car - spider ] we obtain the spectral phase for each @xmath64 present in the data .
the resulting reconstructed spectral phase @xmath65 is displayed in fig . [
fig : exampleshears1d ] ( a ) for a subset of the shears present in the sea - car - spidergram .
the reconstructions show good agreement which we quantify below .
[ fig : exampleshears1d ] ( b ) shows the corresponding temporal intensities .
in this section we investigate the uncertainty of our measurements , showing that sea - car - spider is accurate and precise .
several features of the sea - car - spider enable a convenient and empirical evaluation of the uncertainty .
its single - shot nature enables the random errors to be evaluated by studying the statistics of an ensemble of sequential acquisitions , whilst the presence of systematic errors caused by spatial nonuniformities is revealed by variation between the reconstructions from different shears .
the magnitude of the errors resulting from various misalignments , instabilities or miscalibrations can also be conveniently calculated due to the simple analytic reconstruction algorithm .
we first evaluated the magnitude of the random errors by examining the statistics over @xmath66 single shot traces .
we reconstruct each shot independently at @xmath67 different shears @xmath68 , ranging from to , yielding a set of reconstructed phases @xmath69 from which shot - to - shot statistics can be obtained .
we calculated the mean square shot - to - shot spectral phase variation @xmath70 ^ 2\rangle_n$ ] , where @xmath71 denotes deviation from the mean value , and angle braces denote the expectation value over an ensemble which is specified by the subscript .
we find that @xmath72 is less than across the full - width at 10% maximum bandwidth , but did not vary greatly between shears .
it is shown for @xmath73{mrad / fs}$ ] by the black dash - dotted line in fig .
[ fig : multipleshears ] ( c ) . the shot - to - shot variation of the temporal rms pulse duration was less than for the different shears with smaller variations for larger shears .
we also calculated the rms field variation @xmath74 @xcite for each shear , which was typically around . 1 standard deviation ( light blue region ) across the shears .
( b ) temporal intensity ; mean ( red ) and @xmath75 1 standard deviation interval ( light red region ) across the shears .
( c ) rms phase variations magnified : over all shots at a single shear ( black dash - dotted ) , over all 12 shears ( blue , solid ) , and expected variation in the reconstructed phase resulting from fluctuations in the upconversion frequency ( green dashed).,scaledwidth=85.0% ] we attribute this shot - to - shot fluctuations to at least two sources .
the phase error resulting from camera shot noise can be evaluated by the following procedure : the noise density in the fourier domain is evaluated by examining a region with no signal , as shown by the green rectangle in fig .
[ fig : trace](c ) . the total noise energy in the filtered signal is then estimated by multiplying this density by the area of the sideband filter . by parseval s theorem ,
this is equal to the noise energy in @xmath11 after inverse fourier transforming . assuming the noise is uniformly spread over the @xmath76 domain , the noise variance @xmath77 ^ 2\rangle$ ] on the samples of @xmath11 is then estimated . here
, we find a peak snr of @xmath78 .
by contrast , before the filtering , the peak snr was 49 .
the noise on @xmath79 is then @xmath80 @xcite because in regions of low intensity the phase is less well defined . assuming these errors are independent
, the variance on the reconstructed phase resulting from the concatenation algorithm is the sum of their variances at the sampled points @xmath81 for @xmath82 .
therefore , the variance of the reconstructed phase difference between @xmath83 and @xmath84 is @xmath85\}^2\rangle=\sum_{k = n}^{m-1}\eta_{\gamma(\omega_k+\omega/2,x)}^2 $ ] . as an indicative measure
we choose @xmath86 and @xmath87{rad / fs}$ ] as the lower and upper frequencies at half the maximum spectral intensity .
here we find @xmath88{rad}$ ] , consistent with the observed shot - to - shot fluctuations .
another source of shot - to - shot variation was in @xmath4 , as described in section [ sec : calibration ] .
an error @xmath89 in this quantity leads to a shift in the reconstructed spectrum , giving a phase error of @xmath90 . using the rms value of @xmath91 obtained from the data in fig .
[ fig : calmultishots](c ) , we computed the resulting phase errors which are shown by the green dashed line in fig .
[ fig : multipleshears](c ) .
these are consistent with the observed phase fluctuations ( black dash - dotted line ) .
we now examine the systematic errors .
some of these are revealed by inconsistencies between the reconstructions from different shears .
we computed the mean - square variation over the different shears @xmath92 ^ 2 \rangle_k$ ] where @xmath93 is the shot - averaged phase reconstructed using shear @xmath68 .
we have plotted @xmath94 as the blue solid line in fig .
[ fig : multipleshears](c ) .
it is of similar magnitude to the shot - to - shot fluctuations .
however , in sea - car - spider , disagreement between the reconstructions from different shears arises from spatial inhomogeneities in the beam .
we verified this by recording a 2d spectrum of the test - pulse and extracting the variation of the spectral moments accross the beam @xmath7-axis .
this can also be directly extracted from the data trace by comparing the extracted amplitudes at different shear positions with the expected amplitudes @xmath95 , which can be calculated from the spectrum @xmath96 extracted at the @xmath7-origin .
since all temporal characterization techniques involve either spatial averaging or spatial selection , the ability to quantify the spatial variation is advantageous .
the final systematic errors we considered arose from the accuracy of our calibration of @xmath44 and the @xmath7-origin .
an error @xmath97 leads to a fractional shear error of @xmath98 which we measure as , indicated by the error bars in fig .
[ fig : calmultishots](a ) .
the resulting phase error is also fractional , scaled by the same amount , and is negligible for the data presented here .
the error in the @xmath7-origin is more difficult to analyse , as it affects both the shear and the calibration phase .
however , performing the reconstruction algorithm on this data with a range of different values for the @xmath7-origin did not affect the reconstructed phase .
we note that had we used the calibration method using the broadband signals , these errors would be significant .
since averaging of the statistically independent shot - to - shot fluctuations reduced the random errors to significantly below the systematic errors , the latter define the ultimate precision of our measurement . for an intuitive visualization of their magnitude , we plot the temporal profile ( shown in fig . [
fig : multipleshears ] ( b ) on logarithmic scale ) corresponding to the grand average phase @xmath99 with the variation between the shears indicated by the shaded area .
we achieve a temporal intensity dynamic range of @xmath100 .
we also verified the accuracy of the reconstruction by measuring the spectral phase introduced by a long piece of bk7 glass which we inserted into the beam .
we measured a quadratic phase of @xmath101{fs^2}$ ] , which differs from the theoretical value @xmath102{fs^2}$ ] by @xmath57{fs^2}$ ] .
we have demonstrated a novel self - referencing and easily calibrated ultrafast pulse characterization technique .
a full set of data , consisting of the calibration phase , the spectral amplitude of the unknown pulse , and the spectral phase encoded with multiple shears , is recorded in a single shot .
the redundant information contained in each single shot data trace provides useful material for consistency checks .
a high precision shear and upconversion frequency calibration is straightforward and can be done in one single step for all shears present without the need of any manual or motorized adjustments .
we envisage that this device will be beneficial for its inherent reliability through a lack of moving parts , its improved precision , and the redundant information in the trace which assists in error determination .
this research was supported in part by the european commission through the research training network xtra ( contract mrtn - ct-2003 - 505138 ) and by epsrc er / s24015/oj .
dra was supported by an oxford university clarendon fund scholarship .
iaw acknowledges support from the royal society . | we report a new version of spectral phase interferometry for direct electric field reconstruction ( spider ) , which enables consistency checking through the simultaneous acquisition of multiple shears and offers a simple and precise calibration method . by mixing the test pulse with two spatially chirped ancilla fields we generate a single - shot interferogram which contains multiple shears , the spectral amplitude of the test pulse , and the reference phase , which is accurate for broadband pulses .
all calibration parameters shear , upconversion - frequency and reference phase position can be accurately obtained from a single calibration trace . 10 i. a. walmsley and c. dorrer , `` characterization of ultrashort electromagnetic pulses , '' adv .
opt .
photon .
* 1 * , 308437 ( 2009 ) . p. oshea , m. kimmel , x. gu , and r. trebino , `` highly simplified device for ultrashort - pulse measurement , '' opt . lett . * 26 * , 932934 ( 2001 ) . a. s. radunsky , i. a. walmsley , s
.- p .
gorza , and p. wasylczyk , `` compact spectral shearing interferometer for ultrashort pulse characterization , '' opt . lett . *
32 * , 181183 ( 2007 ) . c. dorrer ,
`` influence of the calibration of the detector on spectral interferometry , '' j. opt .
soc . am .
b * 16 * , 11601168 ( 1999 ) . c. dorrer and i. walmsley ,
`` precision and consistency criteria in spectral phase interferometry for direct electric - field reconstruction , '' j. opt .
soc . am .
b * 19 * , 10301038 ( 2002 ) .
r. trebino , _ frequency - resolved optical gating : the measurement of ultrashort laser pulses _ ( springer , 2002 ) .
c. iaconis and i. a. walmsley , `` spectral phase interferometry for direct electric - field recontruction of ultrashort optical pulses , '' opt .
lett .
* 23 * , 792794 ( 1998 ) .
e. kosik , a. radunsky , i. walmsley , and c. dorrer , `` interferometric technique for measuring broadband ultrashort pulses at the sampling limit , '' opt . lett . * 30 * , 326328 ( 2005 ) . j. r. birge , r. ell , and f. x. krtner , `` two - dimensional spectral shearing interferometry for few - cycle pulse characterization , '' opt . lett . * 31 * , 20632065 ( 2006 ) . t. witting , d. r. austin , and i. a. walmsley , `` improved ancilla preparation in spectral shearing interferometry for accurate ultrafast pulse characterization , '' opt
. lett . * 34 * , 881883 ( 2009 ) .
a. monmayrant , m. joffre , t. oksenhendler , r. herzog , d. kaplan , and p. tournois , `` time - domain interferometry for direct electric - field reconstruction by use of an acousto - optic programmable filter and a two - photon detector , '' opt . lett . * 28 * , 278280 ( 2003 ) .
p. baum , s. lochbrunner , and e. riedle , `` zero - additional - phase spider : full characterization of visible and sub-20-fs ultraviolet pulses , '' opt .
lett .
* 29 * , 210212 ( 2004 ) .
m. lelek , f. louradour , a. barthelemy , and c. froehly , `` time resolved spectral interferometry for single shot femtosecond characterization , '' opt .
commun . * 261 * , 124129 ( 2006 ) .
s .- p .
gorza , p. wasylczyk , and i. a. walmsley , `` spectral shearing interferometry with spatially chirped replicas for measuring ultrashort pulses , '' opt . express * 15 * , 1516815174 ( 2007 ) .
j. wemans , g. figueira , n. lopes , and l. cardoso , `` self - referencing spectral phase interferometry for direct electric - field reconstruction with chirped pulses , '' opt
. lett . * 31 * , 22172219 ( 2006 ) .
d. r. austin , t. witting , and i. a. walmsley , `` broadband astigmatism - free czerny - turner imaging spectrometer using spherical mirrors , '' appl . opt . * 48 * , 38463853 ( 2009 ) .
c. dorrer and i. a. walmsley , `` accuracy criterion for ultrashort pulse characterization techniques : application to spectral phase interferometry for direct electric field reconstruction , '' j. opt .
soc . am .
b * 19 * , 10191029 ( 2002 ) .
d. r. austin , t. witting , and i. a. walmsley , `` high precision self - referenced phase retrieval of complex pulses with multiple - shearing spectral interferometry , '' j. opt .
soc . am .
b * in print * ( 2009 ) . |
one of the major issues in ongoing electron scattering studies of many - body nuclei is the study of short - range nucleon - nucleon ( @xmath10 ) correlations . at present ,
the major research efforts that aim at studying @xmath10 correlations with the aid of the electromagnetic probe proceed along two lines .
the most direct source of information is the simultaneous detection of the two ( correlated ) nucleons that are ejected after the absorption of one single photon .
a more indirect way of possibly probing the short - range correlations , is the semi - exclusive @xmath9 reaction provided that appropriate kinematic regions are sampled and background contributions from two - body currents @xcite and multiple - scattering effects can be kept under control @xcite .
numerous theoretical investigations that have addressed the semi - exclusive @xmath11 process , have adopted a factorized approach in which the electronuclear part and the information on the energy and momentum distributions of nucleons in nuclei are strictly separated in determining the differential cross sections @xcite @xmath12 here , the electronuclear part is contained in the factor @xmath13 which is an off - shell extrapolation of the electron - proton cross section .
the spectral function @xmath14 contains the nuclear - structure information and yields the joint probability distribution of removing a nucleon with momentum @xmath15 from the target system and finding the residual @xmath16 nucleus at an excitation energy @xmath17 .
the distorted spectral function @xmath18 that appears in the above expression corrects @xmath14 for final - state interaction ( fsi ) effects which the ejectile undergoes .
a factorized approach of the above form is subject to several uncertainties .
not only does it assume that the @xmath10 `` correlations '' and final - state interactions are equally acting in the longitudinal and transverse contributions to the cross sections , it fails to include the strength from competing multi - body mechanisms like for example meson - exchange currents .
there is accumulating evidence for enhanced @xmath9 strength of non - single particle origin in the transverse response at deep missing energies @xcite .
the fact that a comparable enhancement remains unobserved in the longitudinal channel alludes to important physical phenomena that fall beyond the effects implemented within factorized approaches based upon the eq .
( [ eq : factor ] ) . in ref .
@xcite an unfactorized model for the calculation of semi - exclusive @xmath9 cross sections was outlined .
the model included the effect of central ( or , jastrow ) nucleon - nucleon correlations as well as genuine two - body meson - exchange ( mec ) and isobar two - body currents ( ic ) that are often sources of @xmath1 strength that could be mistakingly interpreted as signals from the short - range correlations .
our method of calculating the semi - exclusive @xmath9 differential cross sections is based upon the assumption that the short - range correlations will predominantly manifest themselves as two - nucleon knockout even if only one of the ejectiles is detected . in this context
we found it appropriate to set up a calculational framework that aims at providing a unified description of exclusive @xmath19 and semi - exclusive @xmath9 processes . in computing semi - exclusive @xmath9 strength
we explicitly evaluate two - nucleon knockout cross sections and integrate over the complete phase space of the undetected nucleons . in calculating the two - nucleon knockout cross sections we go beyond the mean - field approach by implementing corrections for short - range nucleon - nucleon correlations upon wavefunctions determined in a mean - field model . in this paper , our framework for computing semi - exclusive @xmath9 strength
is extended to include also spin - dependent nucleon - nucleon correlations .
in addition , we address the issue how the effect of the correlations manifests itself as a function of the bjorken @xmath20 scaling variable .
moreover , we consider higher momentum transfer regions which are now accessible at the thomas jefferson national accelerator facility ( tjnaf ) . in doing this
we aim at finding out the most favorable conditions under which the mec and ic contributions to the semi - exclusive @xmath9 processes can be kept under control .
the outline of this paper is as follows .
the basic assumptions underlying our @xmath9 and @xmath21 calculations are sketched in sections [ sec : eep ] and [ sec : eepn ] .
section [ sec : effectiveop ] discusses the spin - dependent correlations and the way they have been implemented in the calculations .
technical details are given in section [ sec : appendix ] .
results of the numerical @xmath0o@xmath1 calculations for various kinematical regions in @xmath2 ( @xmath4(gev / c)@xmath5 ) and bjorken @xmath3 values in the range ( 0.15 @xmath22 @xmath3 @xmath22 2 ) are presented in section [ sec : results ] .
finally , the summary and conclusions are given in section [ sec : conclusions ] .
in determining the transition matrix elements that correspond to the contribution from the @xmath10 correlations to the semi - exclusive @xmath11 reaction @xmath23 we adopt the view that whenever a photon hits a `` correlated '' nucleon both the latter and its `` correlated '' partner will be ejected from the target system . as a consequence ,
the @xmath16 system for which the characteristics are determined in a semi - exclusive @xmath9 measurement is of the form @xmath24 adopting such a reaction picture , the semi - exclusive @xmath9 cross section can be computed by integrating over the phase space of the undetected but emitted nucleon @xmath25 thus , the problem of computing semi - exclusive @xmath9 cross sections expands into the question of calculating two - nucleon knockout cross sections .
this question is the subject of the next subsection . in the standard fashion
, the differential cross section for exclusive @xmath26 processes @xmath27 can be cast in the form @xmath28 \ ; , \end{aligned}\ ] ] where the mott cross section @xmath29 and the recoil factor @xmath30 are defined as @xmath31 further , @xmath32 denote the polar and azimuthal angle of the ejectile @xmath33 and @xmath34 is the angle between the directions of the two ejected nucleons .
the above differential cross section refers to the `` exclusive '' situation in which the residual @xmath35 nucleus is created at a well - defined excitation energy .
we consider a reference frame in which the momentum transfer @xmath36 is aligned along the @xmath37 axis and the @xmath38plane coincides with the electron scattering plane .
the structure functions @xmath39 are expressed in terms of the transition matrix elements for the two transverse and longitudinal photon polarizations in the usual manner @xmath40 \ ; , \\
w_{lt } \left ( \theta_a , \theta_b , \phi _ a - \phi_b , \frac { \phi _ a + \phi_b } { 2 } \right ) & = & 2 \re \biggl\ { \sum_{m_{s_a } , m_{s_b},m_r } \bigg [ \left ( m_{f}^{fi } \left ( \lambda = 0 \right ) \right)^ { \ast } \left ( m_{f}^{fi } \left ( \lambda = -1 \right ) \right ) \nonumber\\ & & - \left ( m_{f}^{fi } \left ( \lambda = 0 \right ) \right)^{\ast } \left ( m_{f}^{fi } \left ( \lambda = + 1 \right ) \right ) \biggr ] \biggr\ } \ ; , \\ w_{tt } \left ( \theta_a , \theta_b , \phi _ a - \phi_b , \frac{\phi _ a + \phi_b } { 2 } \right ) & = & 2 \re \biggl\ { \sum_{m_{s_a } , m_{s_b},m_r } \left ( m_{f}^{fi } \left ( \lambda = -1 \right ) \right)^{\ast } \left ( m_{f}^{fi } \left ( \lambda = + 1 \right ) \right ) \biggr\ } \ ; , \end{aligned}\ ] ] where the sum extends over the spins of the three fragments in the final state and the transition matrix elements are given by @xmath41 here , @xmath42 ( @xmath43 ) are the transverse ( longitudinal ) components of the spatial nuclear current density . a convenient and numerically stable way of dealing with the high dimensionality of transition matrix elements with three hadronic objects ( @xmath44 ) in the final state , is performing a partial wave expansion for the ejectile s wave functions @xcite
. this method asks for a parallel multipole decomposition of the spatial current density ( @xmath45 ) in terms of the conventional coulomb , electric and magnetic operators @xmath46 \ ; , \end{aligned}\ ] ] where @xmath47 .
each of the multipole components can be expressed as spherical tensor operators according to @xmath48 \cdot \vec{j } \left ( \vec{r } \right ) \ ; , \nonumber \\
t_{jm}^{mag } \left(q \right ) & = & \int d \vec{r } \
j_j \left(qr \right ) \vec{y}_{j \left(j,1 \right)}^m \left ( \omega_r \right ) \cdot \vec{j } \left ( \vec{r } \right ) \ ; .
\label{eq : multiop}\end{aligned}\ ] ] the transition matrixelements @xmath49 contained in eq .
( [ eq : transitionmat ] ) can be cast in a closed form after specifying the overlap between the final and initial state wave functions . as outlined in refs .
@xcite , within the adopted framework that involves partial - wave expansions for the nuclear wave functions and multipole decompositions for the operators , the calculation of the cross sections with genuine two - body currents ( like mec and ic ) is eventually reduced to evaluating standard two - body reduced matrix elements @xmath50 .
here , @xmath51 is one of the multipole operators from eq .
( [ eq : multiop ] ) and the quantum numbers @xmath52 refer to either a bound @xmath53 or a continuum eigenstate @xmath54 of the mean - field potential . in the forthcoming subsection
we outline a technique that allows to evaluate the matrix elements corresponding with the electromagnetic coupling to a correlated nucleon pair . over the years , various techniques to correct independent particle model ( ipm ) wave functions for correlations have been developed .
all of these techniques , though , face the complications that arise when going beyond a slater determinant approach . in a correlated basis function ( cbf )
theory , the correlated wave functions @xmath55 are constructed by applying a many - body correlation operator to the uncorrelated wave functions @xmath56 from a mean - field potential @xmath57 with the normalisation factor @xmath58 .
the correlation operator @xmath59 reflects a similar operatorial structure as the standard one - boson exchange parametrizations of the nucleon - nucleon force and contains several terms @xmath60 \ ; , \ ] ] where @xmath61 and @xmath62 is the symmetrisation operator . apart from ( relatively small )
spin - orbit terms the following operators are usually considered in constructing @xmath63 @xmath64 where @xmath65 is the tensor operator @xmath66 .
often , it has been reported that of all of the above components the central ( or jastrow ) @xmath67 , the spin - isospin @xmath68 and the tensor @xmath69 term cause the biggest correlation effects in the nuclear system @xcite .
so , for the remainder of the paper we consider a correlation operator of the form @xmath70 \ ; .\ ] ] in many finite - nuclei calculations , only the ( state - independent ) central correlations @xmath67 are retained . in this paper
, we aim at developing a practical method in which both the effect of the central and of the two major spin - dependent correlations on two- and single - nucleon knockout processes can be evaluated .
it is worth stressing that most of the cluster expansion techniques to calculate the electromagnetic response of nuclei that are outlined in literature refer to the `` inclusive '' case @xcite .
hereby , closure properties are maximally exploited and the computed responses usually refer to responses integrated over all excitation energies of the final system . for the present purposes , in which we aim at explicitly determining the effect of the correlations on specific _ exclusive _ and _ semi - exclusive _ channels ,
closure relations can not be exploited and the response to well - determined excited states of the residual system has to be evaluated . as a consequence , for our purposes
most of the cluster expansion techniques which are available in literature are not directly applicable . at the same time
, the scope of our theoretical calculations is different . indeed
, our calculational framework is meant to bridge the gap between the many - body calculations that come up with predictions for the dynamical correlation functions contained in @xmath71 and exclusive electronuclear reactions , where the predictions can be put to a test .
usually , evaluating transition matrixelements between correlated states is far from a trivial task .
we rely on an `` effective operator approach '' that was outlined in more detail in ref .
a formal formulation of the calculational scheme was given by feldmeier _
_ in ref .
@xcite where the method was introduced as the _ unitary correlation operator method ( ucom)_. in essence , the technique is based on an operator expansion that is pursued after combining the correlation operators acting on the nuclear states and the different terms contained in the photon - nucleus interaction hamiltonian .
formally , this procedure amounts to formally rewriting the transition matrix element between _ correlated _ states @xmath72 as a transition between _ uncorrelated _ states @xmath73 where the effect of the central , spin - isospin and tensor correlations is implemented in an effective transition operator that combines the effect of @xmath10 correlations and the photon - nucleus hamiltonian @xmath74 \right)^ { \dagger } \widehat{\omega } \left ( \prod_{k < l=1}^a \left [ 1 - \widehat{g } \left ( k , l \right ) + \widehat{s } \left ( k , l \right ) + \widehat{t } \left ( k , l \right ) \right ] \right)\ ; . \label{eq : lameff}\end{aligned}\ ] ] in the above expression the following shorthand notation for the central , spin - isospin and tensor correlation operator is introduced @xmath75 within the present context , the operator @xmath76 stands for the photon - nucleus interaction hamiltoninan and contains apart from the standard one - body part of the impulse approximation also two - body terms @xmath77 } \left ( i \right ) + \sum_{i < j=1 } ^{a } \widehat { \omega}^ { \left [ 2 \right ] } \left ( i , j \right ) \ ; .\ ] ] after expanding the effective transition operator of eq .
( [ eq : lameff ] ) and retaining solely the terms that are linear in any of the correlation operators @xmath78 or @xmath79 one obtains an expression of the type @xmath80 } \left ( i \right ) + \sum_{i < j=1 } ^{a } \widehat { \omega}^ { \left [ 2 \right ] } \left ( i , j \right ) \right ) + \left ( \widehat { \omega}^ { \left [ 1 \right ] , in } + \ \widehat { \omega}^ { \left [ 2 \right ] , in } + \ \widehat
{ \omega}^ { \left [ 1 \right ] , fi } + \ \widehat { \omega}^ { \left [ 2 \right ] , fi } \right ) \ ; , \label{eq : effop}\ ] ] where the index @xmath81 @xmath82 refers to the initial ( final ) state correlations and the following operators were introduced @xmath83 , in } & = & \sum_{i < j } \left [ \widehat { \omega}^ { \left [ 1 \right ] } \left ( i \right ) + \widehat { \omega}^ { \left [ 1 \right ] } \left ( j \right ) \right ] \widehat{l } \left ( i , j \right ) \nonumber \\ & & + \sum_{i < j < k } \left [ \widehat { \omega}^ { \left [ 1 \right ] } \left ( i \right ) \widehat{l } \left ( j , k \right ) + \widehat { \omega}^ { \left [ 1 \right ] } \left ( j \right ) \widehat{l } \left ( i , k \right ) + \widehat { \omega}^ { \left [ 1 \right ] } \left ( k \right ) \widehat{l } \left ( i , j \right ) \right ] \ ; , \nonumber \\ & & \nonumber \\ \widehat { \omega } ^ { \left [ 2 \right ] , in } & = & \sum_{i < j } \ ; \widehat { \omega}^ { \left [ 2 \right ] } \left ( i , j \right ) \widehat{l } \left ( i , j \right ) \nonumber \\ & & + \sum_{i < j < k } \left [ \widehat { \omega}^ { \left [ 2 \right ] } \left ( i , j \right ) \widehat{l } \left ( i , k \right ) + \widehat { \omega}^ { \left [ 2 \right ] } \left ( i , j \right ) \widehat{l } \left ( j , k \right ) + \widehat { \omega}^ { \left [ 2 \right ] } \left ( i , k \right ) \widehat{l } \left ( i , j \right ) \right . \nonumber \\ & & + \left . \widehat { \omega}^ { \left [ 2 \right ] } \left ( i , k \right ) \widehat{l } \left ( j , k \right ) + \widehat { \omega}^ { \left [ 2 \right ] } \left ( j , k \right ) \widehat{l } \left ( i , j \right ) + \widehat { \omega}^ { \left [ 2 \right ] } \left ( j , k \right ) \widehat{l } \left ( i , k \right ) \right ] \nonumber \\ & & + \sum_{i < j < k < m } \left [ \widehat { \omega}^ { \left [ 2 \right ] } \left ( i , j \right ) \widehat{l } \left ( k , m \right ) + \widehat { \omega}^ { \left [ 2 \right ] } \left ( i , k \right ) \widehat{l } \left ( j , m \right ) + \widehat { \omega}^ { \left [ 2 \right ] } \left ( i , m \right ) \widehat{l } \left ( j , k \right ) \right . \nonumber \\ & & + \left . \widehat { \omega}^ { \left [ 2 \right ] } \left ( j , k \right ) \widehat{l } \left ( i , m \right ) + \widehat { \omega}^ { \left [ 2 \right ] } \left ( j , m \right ) \widehat{l } \left ( i , k \right ) + \widehat { \omega}^ { \left [ 2 \right ] } \left ( k , m \right ) \widehat{l } \left ( i , j \right ) \right ] , \nonumber \\ & & \nonumber \\
\widehat { \omega } ^ { \left [ 1 \right ] , fi } & = & \left ( \widehat { \omega } ^ { \left [ 1 \right ] , in } \right ) ^{\dagger } \nonumber \\ \widehat { \omega } ^ { \left [ 2 \right ] , fi } & = & \left ( \widehat { \omega } ^ { \left [ 2 \right ] , in } \right ) ^{\dagger } \ ; .\end{aligned}\ ] ] the operator @xmath84 is a shorthand notation for the sum of the central , spin - isospin and tensor correlation operator @xmath85 in the absence of initial and final - state correlations only the first two terms in the expansion of eq .
( [ eq : effop ] ) would not vanish . at large internucleon distances ( @xmath86 4 fm )
we have @xmath87 and the operator @xmath88 heals to the uncorrelated operator @xmath89 .
it is worth remarking that the ground- and final - state correlations are to be treated on the same footing in order to guarantee that the effective operator formalism produces hermitian transition operators . even in the lowest - order expansion of eq .
( [ eq : effop ] ) the effective operator @xmath90 contains up to four - body operators . in
exclusive two - nucleon knockout processes that sample two - body kinematics the three- and four - body operators are expected to produce very small corrections .
additionally , the probability of finding three and more correlated nucleons at normal nuclear densities is generally conceived to be extremely small . in the `` single pair approximation '' ( spa ) , all terms up to two - body operators in the expansion of eq .
( [ eq : effop ] ) are retained . substituting the one - body hadronic current @xmath91}$ ] for @xmath92}$ ] and the two - body current @xmath93}$ ] for @xmath94}$ ] one
is left with the following effective transition operator @xmath95 } \left ( i \right ) + \sum _ { i < j=1}^a j_{\lambda}^ { \left [ 2 \right ] } \left ( i , j \right ) \right\ } + \sum _ { i < j=1}^a j_{\lambda}^ { \left [ 1 \right],in } \left ( i , j \right ) + \sum _ { i < j=1}^a j_{\lambda}^ { \left [ 1 \right],fi } \left ( i , j \right ) \ ; \ ; \ ; \ ; \ ; \left ( \lambda \ ! = \ ! 0 , \pm 1 \right ) \ ; .
\label{eq : spa}\ ] ] the physical interpretation of the various terms appearing in eq .
( [ eq : spa ] ) is straightforward .
the first two terms reflect the electromagnetic interaction hamiltonian for uncorrelated mean - field states .
the @xmath96 } \left ( i , j \right)$ ] are the mesonic and isobaric currents for which our model assumptions can be found in refs @xcite .
for the one - body charge and current density operator we consider the standard form of the non - relativistic impulse approximation .
then , the `` correlation '' terms that occur in eq .
( [ eq : spa ] ) are effective two - body operators and read @xmath97,in } \left ( i , j \right ) & = & e \frac{\omega } { q } \biggl [ \delta \ ! \left ( \vec{r } - \vec{r}_i \right ) \frac{1 + \tau_{z , i}}{2 } + \delta \
! \left ( \vec{r } - \vec{r}_j \right ) \frac{1 + \tau_{z , j}}{2 } \biggr ] \left[\widehat{s } \left ( i , j \right ) + \widehat{t } \left ( i , j \right ) - \widehat{g } \left ( i , j \right ) \right ] \ ; , \nonumber \\ \vec{j}_{\lambda= \pm1}^{\ \left [ 1 \right],in } \left ( i , j \right ) % \vec{j}_{\lambda= \pm1 , corr}^{\ \left [ 2 \right ] , conv } \left % ( i , j \right ) + \vec{j}_{\lambda= \pm1 , corr}^{\ \left [ 2 \right ] , magn } % \left ( i , j \right ) % \nonumber \\ & = & \frac { e \hbar } { 2 m_n } \biggl [ \frac{1 + \tau_{z , i}}{2i } \left [ \vec{\nabla}_i , \delta \left ( \vec{r } - \vec{r}_i \right ) \right]_+ + \frac{1 + \tau_{z , j}}{2i } \left [ \vec { \nabla}_j , \delta \left ( \vec{r } - \vec{r}_j \right ) \right]_+ \nonumber \\ & & + \mu_i \delta \left ( \vec{r } - \vec{r}_i \right ) \vec { \nabla}_i \times \vec { \sigma}_i + \mu_j \delta \ !
\left ( \vec{r } - \vec{r}_j \right ) \vec { \nabla}_j \times \vec { \sigma}_j \biggr ] \left [ \widehat{s } \left ( i , j \right ) + \widehat{t } \left ( i , j \right ) - \widehat{g } \left ( i , j \right ) \right ] \ ; , \nonumber \\
\vec{j}_{\lambda}^{\ \left [ 1 \right],fi } & = & \left(\vec{j}_{\lambda}^{\ \left [ 1 \right],in } \right)^{\dagger } \ ; , \label{eq : jeff}\end{aligned}\ ] ] where we have introduced the charge density operator in the longitudinal component of the vector current , to impose current conservation at the one - body level .
the anticommutator @xmath98_+ \equiv \widehat{a}\widehat{b}+\widehat{b}\widehat{a}$ ] is defined in the standard fashion .
the reduced matrix elements corresponding with the effective currents operators of eq .
( [ eq : jeff ] ) are given in appendix a. it is worth stressing that the tensor term can only be evaluated at a large computational cost . strictly speaking , the effective operator of eq .
( [ eq : spa ] ) has an additional term of the form @xmath99 } \left(i , j\right ) \widehat{l } \left(i , j\right ) + h.c .
\right)$ ] .
this term regularizes the two - body currents for dynamical correlations in the nuclear wave functions .
the short - range behaviour of two - body currents is commonly regularized in a semi - phemonological manner through introducing hadronic form factors of the monopole form . in our calculations
we have adopted this procedure with a cut - off formfactor of @xmath100=1250 mev .
the cumulative effect of hadronic form factors , which seriously cut on the short - range part of the two - body currents , and dynamical short - range correlations in the wave functions was shown to be small @xcite and is neglected in our calculations .
further , we have adopted electromagnetic nucleon form factors in the standard dipole parametrization .
in the forthcoming , @xmath0o@xmath1 results will be presented .
although our methods are applicable to all even - even target nuclei , we have selected a light target nucleus for several reasons .
a first reason is that the amount of computing time required to determine the differential cross sections , depends on the number of nucleon pairs that can be formed in the target nucleus .
second , unlike for heavy target nuclei , for a @xmath101 nucleus coulomb distortion effects in the electron waves can be safely ignored .
a third consideration concerns the fact that light nuclei are far more transparent for nucleon emission than heavy nuclei @xcite . as a consequence ,
the attenuation of the emitted protons and neutrons is a serious complication in heavy nuclei .
after all , the physics behind @xmath10 `` correlations '' is conceived to be rather mass independent and more favorable conditions to study them with the electromagnetic probe are probably reached when selecting light target nuclei . in constructing the slater determinants for the initial and final states we use a realistic set of single - particle wave functions obtained from a hartree - fock calculation with an effective skyrme force .
the use of other sets of realistic single - particle wave functions does not significantly alter our conclusions .
as both the ground - state wave function for the target nucleus @xmath0o and the scattering states for the two ejectiles are constructed from the same hartree - fock potential , the transition matrix elements are free from spurious contributions related to non - orthogonality deficiencies . in implementing the effects of ground - state correlations , we use the central correlation function @xmath102 from the g - matrix calculations in nuclear matter with the reid potential by w.h .
dickhoff and c. gearhart @xcite .
this correlation function is shown in figure [ fig : corfunc ] and heals to zero at @xmath103 2.5 @xmath104 .
it has a hard core at short internucleon distances @xmath105 , guaranteeing that the nucleons repel each other strongly enough when they come too close . using this correlation function
, we achieved a fair agreement with the available data sets for @xmath106c@xmath107 @xcite and @xmath0o@xmath107 @xcite .
for the spin - isospin and tensor correlation functions we use those that were obtained by s. pieper _ et al .
_ in a variational calculation for the ground state of @xmath0o with the argonne @xmath108 nn potential @xcite .
it is worth stressing that the correlation functions are conceived to constitute a general feature of atomic nuclei and that the correlation functions are predicted to exhibit a very small @xmath109 dependence . from fig .
[ fig : corfunc ] it is clear that in coordinate space the spin - isospin and tensor correlation function appear considerably weaker than the central one . in momentum space , though , a very different picture emerges and for momenta in the range 200 - 400 mev / c the correlation functions related to the spin - dependent terms are of comparable magnitude than the central correlation function . on the basis of the behaviour of the correlation functions in momentum space , one could expect important contributions from the tensor correlations at intermediate missing ( or , relative pair ) momenta .
all the @xmath0o@xmath1 differential cross sections presented below are obtained by incoherently adding the separately computed @xmath0o@xmath110 and @xmath0o@xmath107 strengths . in this procedure ,
two - nucleon knockout from all possible shell - combinations ( viz .
@xmath111 and @xmath112 ) is incorporated .
we have carried out @xmath0o@xmath1 calculations for three kinematical conditions .
the first corresponds with a small bjorken scaling variable @xmath113 , the second with @xmath114 ( quasi - elastic conditions ) and the third with @xmath8 where some typical quasi - deuteron like structures are expected to occur .
the quasi - elastic case coincides with the kinematics of a recent tjnaf experiment ( e89 - 003 ) @xcite .
figures [ fig : central]-[fig : centralx2 ] summarize the results of the @xmath0o(@xmath115 ) calculations .
we first address the issue how the central , tensor and spin - isospin correlations manifest themselves in the @xmath1 differential cross sections . for
each of the three kinematical regions under consideration we study the differential @xmath0o@xmath1 cross section versus missing energy and proton angle @xmath116 in planar kinematics ( @xmath117 ) .
the observed functional dependence in @xmath118 allows to study the missing momentum variations of the cross sections . for the sake of convenience , for each electron kinematics we have added a panel with the variation of the missing momentum versus missing energy and proton angle .
roughly , the probed missing momentum @xmath119 increases with increasing @xmath116 and missing energy @xmath120 .
it is clear that by measuring the missing energy dependence of the @xmath1 cross section for a number of proton polar angles , access to broad regions in the @xmath121 plane is gained .
comparing the two upper panels in figures [ fig : central ] , [ fig : centralx1 ] and [ fig : centralx2 ] , a common qualitative feature emerges .
for all three considered regions in the bjorken scaling variable @xmath3 the strength attributed to the tensor correlations largely overshoots the strength from the central correlations .
this holds in particular for the small proton angles @xmath118 . at these angles ,
typically the smallest missing momenta are probed .
the effect of the spin - isospin correlations is found to be at the few percent level , and , therefore , negligible .
the central correlations are observed to manifest themselves in a wider range of the ( @xmath122 ) space than the tensor correlations do .
the effect of the tensor correlations is seemingly confined to a region of relatively low and moderate missing momenta , whereas the contribution from the central correlations extends to higher proton angles @xmath123 where typically higher missing momenta are probed .
this qualitative behaviour of the calculated differential cross sections reflects the fact that central correlations are the most important correlations in the spectral function at really high missing momenta , whereas the intermediate range is usually dominanted by the tensor correlations @xcite .
such a qualitative behaviour can also be inferred from fig .
[ fig : corfunc ] . another interesting feature of how the ground - state correlations manifest themselves in the @xmath121 plane is that the peak of the @xmath0o@xmath1 differential cross sections shifts to higher missing energies @xmath120 as one gradually moves out of parallel kinematics and higher @xmath116 angles ( and consequently , missing momenta ) are probed .
this observation is a manifestation of a well - known feature of the correlated part of the spectral function .
indeed , the average missing energy @xmath124 is predicted to increase quadratically in the missing momentum : @xmath125 , where @xmath126 is the threshold energy for two - nucleon knockout and @xmath127 the average excitation energy of the @xmath128 system that was created after two nucleons escaped from the orbits characterized by the quantum numbers @xmath129 and @xmath130 .
moreover , the strength in the correlated part of the spectral function is often predicted to be localized in a rather narrow region on both sides of this central @xmath120 value ( the so - called `` ridge '' in the spectral function ) .
the upper panels in figs . [ fig : central]-[fig : centralx2 ] clearly illustrate that the largest fraction of the calculated @xmath0o@xmath1 strength is localized on both sides of the `` ridge '' . for the sake of convenience
, a panel with the exact location of the ridge in the @xmath121 plane was added to figure [ fig : central ] .
the calculated strength appears in a rather wide band of missing energies around this ridge , though .
despite the fact that our calculations are unfactorized in nature , the above observations with regard to the qualitative behaviour of the calculated differential cross sections , illustrate that our unfactorized results exhibit the same qualitative features than what could be expected to happen in a factorized approach based upon the equation ( [ eq : factor ] ) using a realistic `` correlated '' spectral function .
one of the major advantages of our unfactorized approach to the semi - exclusive @xmath9 reaction , is that also the strength from the mec and ic can be evaluated in the same calculational framework in which also the contributions from ground - state correlations are determined .
this allows to calculate some of the additional reaction - mechanism effects that break the direct link between the measured cross sections @xmath1 differential cross sections and the spectral function .
the predictions for the differential cross sections when including both the ground - state correlations and the mec / ic currents are contained in figs .
[ fig : central]-[fig : centralx2 ] . for the low @xmath3 kinematics , the calculated contribution from the mec and ic overshoots the calculated strength from the @xmath10 correlations .
this confirms earlier findings for the @xmath106c@xmath1 @xcite and the @xmath131he@xmath1 @xcite reaction . even in the quasi - elastic situation ( @xmath114 ) of fig .
[ fig : centralx1 ] the calculated @xmath0o@xmath1 strength that is created through mec and ic is substantially larger than the strength attributed to the nucleon - nucleon correlations .
this is particularly the case at backward nucleon angles @xmath118 where the two - body currents dominate the @xmath1 strength in the @xmath132 plane .
an interesting feature is observed for the low @xmath3 case of fig .
[ fig : central ] .
the mec / ic create almost an order of magnitude more proton - knockout strength than the @xmath10 correlations do .
a striking observation , though , is that in this particular case the mec / ic generate the @xmath1 strength in the same regions of the @xmath133 ) plane as the @xmath10 correlations do .
thus , the mere observation of a `` ridge '' in the continuous part of the @xmath1 spectrum does not necessarily imply that signals directly pointing to the @xmath10 correlations are detected .
better conditions to isolate the @xmath10 correlations with the aid of the @xmath1 reaction , are predicted for the @xmath8 case ( fig .
[ fig : centralx2 ] ) . despite the fact that in the considered kinematics reasonably high missing momenta ( @xmath134 mev / c ) are probed ,
the role played by the mec and ic is moderate in comparison with the role played by the nucleon - nucleon correlations .
it should be stressed that the @xmath135 region is only accessible at facilities with sufficient inital electron energy , like tjnaf , and implies relatively small cross sections ( for the typical example considered for the results of fig .
[ fig : centralx2 ] the cross sections are of the order 0.01 pb / mev@xmath5/sr@xmath5 ) . a recent account of the @xmath1 program at tjnaf is given in ref .
an accepted proposal to measure @xmath106c@xmath136 at @xmath8 at tjnaf kinematics is described in ref .
@xcite .
a powerful tool in studies with electromagnetic probes is the separation of the various structure functions .
we study the longitudinal and transverse structure functions in parallel kinematics ( @xmath137 ) for all three kinematical situations which were discussed earlier in this sections .
not only do solely two structure functions ( @xmath138 and @xmath139 ) contribute to the cross section in the @xmath137 case , parallel kinematics has often been shown to create favorable conditions when it comes to controlling the fsi effects @xcite . in high - energy processes , for example , the eikonal approximation ( or its multiple scattering extension , usually referred to as glauber theory ) is expected to be most accurate at small proton angles @xmath140 .
one could wonder whether parallel kinematics is also favorable when it comes to suppressing the background of mec and ic in @xmath1 . to that purpose
we have evaluated the ratio of the calculated @xmath0o@xmath1 differential cross section with and without mec / ic .
the results are shown in fig .
[ fig : superratio ] .
they illustrate that parallel kinematics is not a bad choice with the eye on suppressing the contribution from the genuine two - body currents ( mec and ic ) . in the @xmath8 case
the @xmath10 correlations and two - body currents are each responsible for about half of the calculated strength at forward proton angles .
as one moves to higher proton angles the @xmath10 correlations gradually lose in importance . for the @xmath141 case
the @xmath10 correlations account for a mere 5% in parallel kinematics and at @xmath118=90@xmath142 their predicted impact is virtually non - existent in comparison with the contributions from the two - body currents .
the longitudinal and transverse part in the differential cross sections for parallel kinematics are shown in figs .
[ fig : parlowx]-[fig : parx2 ] . for all three kinematical conditions considered , the central and tensor @xmath10 correlations exhibit a similar qualitative behaviour in the longitudinal and transverse part of the differential cross section . for both structure functions
, the tensor term strongly dominates the strength attributed to @xmath10 correlations . for @xmath143
the central correlations produce only small effects . for higher values of x ,
when higher missing momenta are probed , the central correlations gain in relative importance with respect to the tensor correlations .
the transverse response , though , is dramatically increased after including the meson - exchange and isobar currents .
qualitatively , the resulting missing energy dependence ( solid curves ) of the transverse cross section differs from the variant without two - body currents ( dot - dashed curves ) insignificantly .
in summary , we have reported unfactorized calculations for the @xmath0o@xmath1 differential cross section at deep missing energies and discussed its sensitivity to @xmath10 correlations .
the calculations are performed in a complete unfactorized theoretical framework in which both state - dependent and state - independent @xmath10 correlations are incorporated , as well as competing processes related to genuine two - body photoabsorption mechanisms like meson - exchange and isobar currents .
we have evaluated the effect of central , spin - isospin and tensor @xmath10 correlations .
of these three terms , the tensor correlations are by far the strongest source of @xmath1 strength .
their contribution often overshoots the strength from the central ( jastrow ) correlations by almost an order of magnitude .
the effect of the spin - isospin @xmath10 correlations upon the semi - exclusive @xmath9 differential cross sections was found to be practically negligible .
the results reveal further that meson - exchange and isobar two - body currents are strongly feeding the semi - exclusive @xmath9 channel at deep missing energies . at low values of bjorken
@xmath3 the genuine two - body currents ( mec / ic ) typically produce one or two orders of magnitude more @xmath1 strength above the two - nucleon knockout threshold than the @xmath10 correlations do . even under quasi - elastic conditions and four - momentum transfers of the order @xmath144(gev / c)@xmath5 , the meson - exchange and isobar currents generate substantially more @xmath9 strength at deep missing energies than the @xmath10 correlations do .
the @xmath10 correlations are found to equally feed the longitudinal and transverse part of the differential cross sections .
whenever the meson - exchange and isobar currents start dominating , the continuous part of the @xmath1 strength becomes almost exclusively transverse . at `` quasi - deuteron '' like values of the bjorken variable ( @xmath145 ) the effect of the two - body currents is observed to be sufficiently suppressed to make a direct link between the measured @xmath1 differential cross sections and the spectral function feasible .
we close with remarking that our investigations reveal that the pion degrees of freedom tend to dominate the @xmath1 strength that is not of single - particle origin .
indeed , both the tensor @xmath10 correlations and the meson - exchange / isobar currents which we find to dominante the semi - exclusive proton knockout process , are intimately connected to the pion - exchange term in the nucleon - nucleon force .
further , from our investigations it appears hard to gain access to central ( jastrow ) correlations with the aid of the @xmath1 reaction .
triple coincidence @xmath146 measurements offer better perspectives in this respect .
this work was supported by the fund for scientific research of flanders under contract no .
4.0061.99 and the university research council .
[ sec : appendix ]
in section [ sec : effectiveop ] it was pointed out that the standard ( one - body ) electromagnetic coupling of a virtual photon to a tensor and spin - isospin correlated nucleon pair can be formally treated as effective two - body current operators that are composed of a `` correlation operator '' and a one - body hadronic current . here
, we collect the expressions for the two - body matrixelements that emerge when considering the electromagnetic coupling through the standard one - body operator of the impulse approximation to a tensor and spin - isospin correlated pair .
the expressions for the central ( state - independent ) correlation term can be found in ref . @xcite .
as we rely on a partial - wave technique to deal with the high dimensionality of the three - fragment final state , the longitudinal and transverse multipole components are separately constructed .
we first consider the matrix elements related to the tensor ( section [ sec : tensor ] ) and thereafter the spin - isospin ( section [ sec : spiniso ] ) correlations . in describing the matrix elements
, we use the shorthand notation @xmath147 .
if an other set of quantumnumbers is involved , they are given explicitly .
the effective operator that accounts for the coupling of a longitudinally polarized virtual photon to a tensor correlated pair reads @xmath148 } \left ( i , j \right ) \equiv \frac{q}{\omega } j_{\lambda=0 , t \tau}^ { \left [ 1 \right],in } \left ( i , j \right ) = e \biggl [ \delta \ ! \left ( \vec{r } - \vec{r}_i \right ) \frac{1 + \tau_{z , i}}{2 } + \delta \ ! \left ( \vec{r } - \vec{r}_j \right ) \frac{1 + \tau_{z , j}}{2 } \biggr ] \ \widehat{t } \left ( i , j \right ) \;.\ ] ] we define the partial wave components of the tensor correlation function @xmath149 @xmath150 and remark that the multipole components @xmath151 corresponding with the above operator can be written in the form @xmath152 } \left ( i , j \right ) \right ] = \sum_{l_1 l_2 } \sum_{l } \sum_{j_3 j_4 } \frac { 4 \sqrt{6 } } { \sqrt { \pi } } e { \cal x } ^ { t \tau } \ ! \left ( l_1 , l_2 , r_i , r_j \right ) \widehat{l}_1 \widehat{l}_2 \widehat{l } \widehat{j}_3 \widehat { j}_4 \langle l_1 \ 0 \ l_2 \ 0 \mid 2 \ 0 \rangle \left\ { \begin{array}{ccc } 1 & 1 & 2 \nonumber \\ l_1 & l_2 & j_3 \end{array } \right\ } i^{l_1 + l_2 } \left ( \vec { \tau}_i \cdot \vec { \tau}_j \right ) \nonumber \\ & & \times \biggl\ { \frac{1 + \tau_{z , i}}{2}\ \widehat{l}_1 \left ( \begin{array}{ccc } l & j & l_1 \nonumber \\ 0 & 0 & 0 \end{array } \right ) \left\ { \begin{array}{ccc } l & j & l_1 \nonumber \\ j_3 & 1 & j_4 \end{array } \right\ } \left ( -1 \right)^{j+1 } j_j \left ( qr_i \right ) \biggl [ \left [ y_l \left ( \omega_i \right ) \otimes \vec { \sigma}_i \right]_{j_4 } \otimes \left [ y_{l_2 } \left ( \omega_j \right ) \otimes \vec { \sigma}_j \right]_{j_3 } \biggr]_j^m \nonumber \\ & & + \frac{1 + \tau_{z , j}}{2}\ \widehat{l}_2 \left ( \begin{array}{ccc } l & j & l_2 \\ 0 & 0 & 0 \end{array } \right ) \left\ { \begin{array}{ccc } l & j & l_2 \\ j_3 & 1 & j_4 \end{array } \right\ } \left ( -1 \right)^{j_3+j_4 + 1 } j_j \left ( qr_j \right ) \biggl [ \left [ y_{l_1 } \left ( \omega_i \right ) \otimes \vec { \sigma}_i \right]_{j_3 } \otimes \left [ y_l \left ( \omega_j \right ) \otimes \vec { \sigma}_j \right]_{j_4 } \biggr]_j^m \biggr\ } \;.
\end{aligned}\ ] ] after straightforward algebraic manipulations , the reduced two - body matrix element corresponding with this operator reads @xmath153 } \left ( 1,2 \right ) \right ] \parallel cd ; j_2 \rangle = \sum_{l_1 l_2 } \sum_{l } \sum_{j_3 j_4 } \int \ ! dr_1 \int \ ! dr_2 \frac { 4 \sqrt{6 } } { \sqrt { \pi } } e { \cal x } ^ { t \tau } \ ! \left ( l_1 , l_2 , r_1 , r_2 \right ) \nonumber \\ & & \times \widehat{l}_1 \widehat{l}_2 \widehat{l } \widehat{j}_3 \widehat { j}_4 \widehat{j}_1 \widehat{j}_2 \widehat{j } \nonumber \langle l_1 \ 0 \ l_2 \ 0 \mid 2 \ 0 \rangle \left\ { \begin{array}{ccc } 1 & 1 & 2 \nonumber \\ l_1 & l_2 & j_3 \end{array } \right\ } i^{l_1 + l_2 } \nonumber \\ & & \times \biggl\ { \delta_{ac , p}\ \widehat{l}_1 \left ( \begin{array}{ccc } l & j & l_1 \nonumber \\ 0 & 0 & 0 \end{array } \right ) \left\ { \begin{array}{ccc } l & j & l_1 \nonumber \\ j_3 & 1 & j_4 \end{array } \right\ } \left\ { \begin{array}{ccc } j_a & j_b & j_1 \nonumber \\ j_c & j_d & j_2 \nonumber \\ j_4 & j_3 & j \end{array } \right\ } \left ( -1 \right)^{j+1 } \nonumber \\ & & \times \langle \ a \parallel j_j \left ( qr_1 \right ) \left [ y_l \left ( \omega_1 \right ) \otimes \vec { \sigma}_1 \right]_{j_4 } \parallel c \
\rangle_{r_1 } \langle \ b \parallel \left [ y_{l_2 } \left ( \omega_2 \right ) \otimes \vec { \sigma}_2 \right]_{j_3 } \parallel d \
\rangle_{r_2 } \nonumber \\ & & + \delta_{bd , p}\ \widehat{l}_2 \left ( \begin{array}{ccc } l & j & l_2 \nonumber \\ 0 & 0 & 0 \end{array } \right ) \left\ { \begin{array}{ccc } l & j & l_2 \nonumber \\ j_3 & 1 & j_4 \end{array } \right\ } \left\ { \begin{array}{ccc } j_a & j_b & j_1 \nonumber \\ j_c & j_d & j_2 \nonumber \\ j_3 & j_4 & j \end{array } \right\ } \left ( -1 \right)^{j_3+j_4 + 1 } \nonumber \\ & & \times \langle \ a \parallel \left [ y_{l_1 } \left ( \omega_1 \right ) \otimes \vec { \sigma}_1 \right]_{j_3 } \parallel c \ \rangle_{r_1 } \langle \ b \parallel j_j \left ( qr_2 \right ) \left [ y_l \left ( \omega_2 \right ) \otimes \vec { \sigma}_2 \right]_{j_4 } \parallel d
\ \rangle_{r_2 } \biggr\ } \ ; , \end{aligned}\ ] ] where the radial transition densities @xmath154 are defined as @xmath155 we introduce @xmath156 and remark that it suffices to determine the matrix elements of this spherical tensor operator as both the electric and magnetic multipole operators can be expressed in terms of @xmath157 @xmath158 in what follows , we treat the convective and magnetization component of the transverse one - body current separately . electromagnetic coupling through the magnetization ( @xmath159 ) current to a tensor - correlated pair @xmath160,magn } \left ( i , j \right ) = \frac { e \hbar } { 2 m_n } \left [ \mu_i \delta \left ( \vec{r } - \vec{r}_i \right ) \vec { \nabla}_i \times \vec { \sigma}_i + \mu_j \delta \ ! \left ( \vec{r } - \vec{r}_j \right ) \vec { \nabla}_j \times \vec { \sigma}_j \right ] \ \widehat{t } \left ( i , j \right)\ ] ] leads to the following @xmath157 operator @xmath161,magn } \left ( i , j \right ) \right ] = \sum_{l_1 l_2 } \sum_{j_3 j_4 } \sum_{l l_3 } \sum _ { \eta= \pm1 } \frac{24 } { \sqrt { \pi } } \frac { \hbar eq}{2 m_n } \widehat { l}_1 \widehat { l}_2 \widehat{j}_3
\widehat{j}_4 \widehat{l } \widehat{l}_3 \widehat { j\!\!+\!\ ! \kappa \!\!+\!\ !
\eta } \left ( \vec { \tau}_i \cdot \vec { \tau}_j \right ) i^{l_1+l_2 + 1 } \nonumber \\ & & \times \sqrt{j+ \kappa+ \delta _ { \eta,+1 } } { \cal x } ^ { t \tau } \ !
\left ( l_1 , l_2 , r_i , r_j \right ) \langle
l_1 \ 0 \ l_2 \ 0 \mid 2 \ 0 \rangle \left\ { \begin{array}{ccc } 1 & 1 & 2 \nonumber \\ l_2 & l_1 & j_3 \end{array } \right\ } \left\ { \begin{array}{ccc } 1 & j & j\!\!+\!\ ! \kappa \nonumber \\
! \kappa \!\!+\!\ !
\eta & 1 & 1 \end{array } \right\ } \left\ { \begin{array}{ccc } j & 1 & j\!\!+\!\ ! \kappa \!\!+\!\ !
\eta \nonumber \\ j_4 & j_3 & l_3 \end{array } \right\ } \nonumber \\ & & \times \biggl\ { \left [ \delta_{i , p } \mu_p + \delta_{i , n } \mu_n \right ] \widehat{l}_1 \left ( \begin{array}{ccc } l & l_1 & j\!\!+\!\ ! \kappa \!\!+\!\ !
\eta \nonumber \\ 0 & 0 & 0 \end{array } \right ) \left\ { \begin{array}{ccc } l & j\!\!+\!\ ! \kappa \!\!+\!\ ! \eta & l_1
\nonumber \\ j_3 & 1 & j_4 \end{array } \right\ } \left(-1 \right)^ { l_3 - j_3 + \kappa } \nonumber \\ & & \times j_{j + \kappa + \eta } \left(qr_i \right ) \biggl [ \biggl [ \left [ y_l \left ( \omega_i \right ) \otimes \vec { \sigma}_i \right]_{j_4 } \otimes \vec { \sigma}_i \biggr]_{l_3 } \otimes \left [ y_{l_2 } \left ( \omega_j \right ) \otimes \vec { \sigma}_j \right]_{j_3 } \biggr]_j^m \nonumber \\ & & + \left [ \delta_{j , p } \mu_p + \delta_{j , n } \mu_n \right ] \widehat{l}_2 \left ( \begin{array}{ccc } l & l_2 & j\!\!+\!\ ! \kappa
\eta \nonumber \\ 0 & 0 & 0 \end{array } \right ) \left\ { \begin{array}{ccc } l & j\!\!+\!\ ! \kappa \!\!+\!\ ! \eta & l_2 \nonumber \\ j_3 & 1 & j_4 \end{array } \right\ } \left(-1 \right)^ { j + \kappa } \nonumber \\ & & \times j_{j + \kappa + \eta } \left(qr_j \right ) \biggl [ \left [ y_{l_1 } \left ( \omega_i \right ) \otimes \vec { \sigma}_i \right]_{j_3 } \otimes \biggl [ \left [ y_l \left ( \omega_j \right ) \otimes \vec { \sigma}_j \right]_{j_4 } \otimes \vec { \sigma}_j \biggr]_{l_3 } \biggr]_j^m \biggr\ } \ ; . \end{aligned}\ ] ] with the aid of standard racah algebra techniques it can be shown that the reduced matrix element corresponding with this operator can be cast in the form @xmath162,magn } \left ( 1,2 \right ) \right ] \parallel cd ; j_2 \rangle = \sum_{l_1 l_2 } \sum_{j_3 j_4 } \sum_{l l_3 } \sum _ { \eta= \pm1 } \sum_j { \int \ ! dr_1 \int \ !
dr_2 } \frac{24 \sqrt{6 } } { \sqrt { \pi } } \frac { \hbar eq}{2 m_n } \langle l_1 \ 0 \ l_2 \ 0 \mid 2 \ 0 \rangle \nonumber \\ & & \times \widehat { l}_1 \widehat { l}_2 \widehat{j}_3 \widehat{j}_4 \widehat{l } \left ( \widehat{l}_3 \right)^2 \widehat { j\!\!+\!\ ! \kappa \!\!+\!\ ! \eta } \widehat{j}_1 \widehat{j}_2 \widehat{j } \widehat{j } i^{l_1+l_2 + 1 } \sqrt{j+ \kappa+ \delta _ { \eta,+1 } } { { \cal x}^{t \tau } \ ! \left ( l_1 , l_2 , r_1 , r_2 \right ) } \nonumber \\ & & \times \left\ { \begin{array}{ccc } 1 & 1 & 2 \nonumber \\
l_2 & l_1 & j_3 \end{array } \right\ } \left\ { \begin{array}{ccc } 1 & j & j\!\!+\!\ !
\\ j\!\!+\!\ ! \kappa \!\!+\!\ !
\eta & 1 & 1 \end{array } \right\ } \left\ { \begin{array}{ccc } j & 1 & j\!\!+\!\ ! \kappa \!\!+\!\ !
\eta \nonumber \\ j_4 & j_3 & l_3 \end{array } \right\ } \nonumber \\ & & \times \biggl\ { \left [ \delta_{ac , p } \mu_p + \delta_{ac , n } \mu_n \right ] \widehat{l}_1 \widehat{j}_c \left ( \begin{array}{ccc } l & l_1 & j\!\!+\!\ ! \kappa \!\!+\!\ ! \eta \nonumber \\ 0 & 0 & 0 \end{array } \right ) \left\ { \begin{array}{ccc } l & j\!\!+\!\ ! \kappa \!\!+\!\ ! \eta & l_1 \nonumber \\ j_3 & 1 & j_4 \end{array } \right\ } \nonumber \\ & & \left\ { \begin{array}{ccc } j_4 & 1 & l_3 \nonumber \\ j_c & j_a & j \end{array } \right\ } \left\ { \begin{array}{ccc } l_c & 1/2 & j \nonumber \\ 1 & j_c & 1/2 \end{array } \right\ } \left\ { \begin{array}{ccc } j_a & j_b & j_1 \nonumber \\ j_c & j_d & j_2 \nonumber \\
l_3 & j_3 & j \end{array } \right\ } \left(-1 \right)^ { j_a + j_c + l_c + j - j_3 + \kappa + 3/2 } \nonumber\\ & & \times \langle \ a \parallel j_{j+ \kappa+ \eta } \left ( qr_1 \right ) \left [ y_l \left ( \omega_1 \right ) \otimes \vec { \sigma}_1 \right]_{j_4 } \
\parallel n_c l_c \frac{1}{2 } j \ \rangle_{r_1 } \langle \ b \parallel \left [ y_{l_2 } \left ( \omega_2 \right ) \otimes \vec { \sigma}_2 \right]_{j_3 } \parallel d \ \rangle_{r_2 } \nonumber \\ & & + \left [ \delta_{bd , p } \mu_p + \delta_{bd , n } \mu_n \right ] \widehat{l}_2 \widehat{j}_d \left ( \begin{array}{ccc } l & l_2 & j\!\!+\!\ ! \kappa \!\!+\!\ !
\eta \nonumber \\ 0 & 0 & 0 \end{array } \right ) \left\ { \begin{array}{ccc } l & j\!\!+\!\ ! \kappa \!\!+\!\ ! \eta & l_2 \nonumber \\ j_3 & 1 & j_4 \end{array } \right\ } \nonumber \\ & & \left\ { \begin{array}{ccc } j_4 & 1 & l_3 \nonumber \\ j_d & j_b & j \end{array } \right\ } \left\ { \begin{array}{ccc } l_d & 1/2 & j \nonumber \\ 1 & j_d & 1/2 \end{array } \right\ } \left\ { \begin{array}{ccc } j_a & j_b & j_1 \nonumber \\ j_c & j_d & j_2 \nonumber \\ j_3 & l_3 & j \end{array } \right\ } \left(-1 \right)^ { j_b + j_d + l_d + j + l_3 + j + \kappa + 3/2 } \nonumber \\ & & \times \langle \ a \parallel \left [ y_{l_1 } \left ( \omega_1 \right ) \otimes \vec { \sigma}_1 \right]_{j_3 } \parallel c \ \rangle_{r_1 } \langle \ b \parallel j_{j+ \kappa+ \eta } \left ( qr_2 \right ) \left [ y_l \left ( \omega_2 \right ) \otimes \vec { \sigma}_2 \right]_{j_4 } \parallel n_d l_d \frac{1}{2 } j \ \rangle_{r_2 } \biggr\ } \;. \end{aligned}\ ] ] the combination of the electromagnetic coupling through the convection current ( @xmath163 ) with a tensor correlated pair @xmath164,conv } \left ( i , j \right ) = \frac { e \hbar } { 2 m_n i } \left [ \frac{1 + \tau_{z , i}}{2 } \left\ { \vec{\nabla}_i , \delta \left ( \vec{r } - \vec{r}_i \right ) \right\ } + \frac{1 + \tau_{z , j}}{2 } \left\ { \vec { \nabla}_j , \delta \left ( \vec{r } - \vec{r}_j \right ) \right\ } \right ] \
\widehat{t } \left(i , j \right)\ ] ] leads to the following spherical tensor operator @xmath165,conv } \left ( i , j \right ) \right ] = \sum_{l_1 l_2 } \sum_{j_3 j_4 } \sum_{j_5 l } \sqrt{\frac{24}{\pi } } \frac { \hbar e}{m_n } \widehat { l}_1 \widehat { l}_2 \widehat{j}_3 \widehat{j}_4 \widehat{j}_5 \widehat{l } \widehat { j\!\!+\!\ ! \kappa } \left ( \vec { \tau}_i \cdot \vec { \tau}_j \right ) \nonumber \\ & & \times i^{l_1+l_2 - 1 } { \cal x } ^ { t \tau } \ ! \left(l_1 , l_2 , r_i , r_j \right ) \langle l_1 \ 0 \ l_2 \ 0 \mid 2 \ 0 \rangle \left\ { \begin{array}{ccc } 1 & 1 & 2 \nonumber \\ l_2 & l_1 & j_3 \end{array } \right\ } \nonumber \\ & & \times \biggl\ { \left ( \frac{1 + \tau_{z , i}}{2 } \right ) \widehat{l}_1 \left ( \begin{array}{ccc } l & l_1 & j\!\!+\!\ ! \kappa \nonumber \\ 0 & 0 & 0 \end{array } \right ) \left\ { \begin{array}{ccc } 1 & j & j\!\!+\!\ !
\kappa \nonumber \\
l_1 & l & j_4 \end{array } \right\ } \left\ { \begin{array}{ccc } 1 & j_3 & l_1 \nonumber \\ j & j_4 & j_5 \end{array } \right\ } \left ( -1 \right)^{j } \nonumber \\ & & \times
j_{j + \kappa } \left(qr_i \right ) \biggl [ \biggl [ \left [ y_l \left ( \omega_i \right ) \otimes \left ( \vec{\nabla}_i - \vec{\nabla}_i ' \right ) \right]_{j_4 } \otimes \vec { \sigma}_i \biggr]_{j_5 } \otimes \left [ y_{l_2 } \left ( \omega_j \right ) \otimes \vec { \sigma}_j \right]_{j_3 } \biggr]_j^m \nonumber \\ & & + \left ( \frac{1 + \tau_{z , j}}{2 } \right ) \widehat{l}_2 \left ( \begin{array}{ccc } l & l_2 & j\!\!+\!\ ! \kappa \nonumber \\ 0 & 0 & 0 \end{array } \right ) \left\ { \begin{array}{ccc } 1 & j & j\!\!+\!\ ! \kappa \nonumber \\ l_2 & l & j_4 \end{array } \right\ } \left\ { \begin{array}{ccc } 1 & j_3 & l_2 \nonumber \\
j & j_4 & j_5 \end{array } \right\ } \left(-1 \right)^{j_3 + j_5 } \nonumber \\ & & \times j_{j + \kappa } \left(qr_j \right ) \biggl [ \left [ y_{l_1 } \left ( \omega_i \right ) \otimes \vec { \sigma}_i \right]_{j_3 } \otimes \biggl [ \left [ y_l \left ( \omega_j \right ) \otimes \left ( \vec{\nabla}_j - \vec{\nabla}_j ' \right ) \right]_{j_4 } \otimes \vec { \sigma}_j \biggr]_{j_5 } \biggr]_j^m \biggr\ } \ ; .\end{aligned}\ ] ] in deriving the above expressions , the terms containing derivatives of the correlation function are neglected .
the notation @xmath166 denotes a gradient operator acting to the left .
the reduced matrix element corresponding with the above operator reads @xmath167,conv } \left ( 1,2 \right ) \right ] \parallel cd ; j_2 \rangle = \sum_{l_1 l_2 } \sum_{j_3 j_4 } \sum_{j_5 l } \int \ ! dr_1 \int \
! dr_2 \frac{12 } { \sqrt { \pi } } \frac { \hbar e } { m_n } \widehat { l}_1 \widehat { l}_2 \widehat{j}_1 \widehat{j}_2 \widehat{j}_3 \widehat{j}_4 \left ( \widehat{j}_5 \right)^2 \widehat{l } \widehat { j\!\!+\!\ ! \kappa } \widehat{j } \nonumber \\ & & \times \langle l_1 \ 0 \ l_2 \ 0 \mid 2 \ 0 \rangle \left\ { \begin{array}{ccc } 1 & 1 & 2 \nonumber \\ l_2 & l_1 & j_3 \end{array } \right\ } i^{l_1+l_2 - 1 } { \cal x } ^ { t \tau } \ ! \left ( l_1 , l_2 , r_1 , r_2 \right ) \nonumber \\ & & \times \biggl\ { \delta_{ac , p } \ ; \widehat{l}_1 \widehat{j}_a \widehat{j}_c \left ( \begin{array}{ccc } l & l_1 & j\!\!+\!\ ! \kappa \nonumber \\ 0 & 0 & 0 \end{array } \right ) \left\ { \begin{array}{ccc } 1 & j & j\!\!+\!\ ! \kappa \nonumber \\ l_1 & l & j_4 \end{array } \right\ } \left\ { \begin{array}{ccc } 1 & j_3 & l_1 \nonumber \\
j & j_4 & j_5 \end{array } \right\ } \left\ { \begin{array}{ccc } j_a & j_b & j_1 \nonumber \\
j_c & j_d & j_2 \nonumber \\
j_5 & j_3 & j \end{array } \right\ } \left\ { \begin{array}{ccc } l_a & 1/2 & j_a \nonumber \\
l_c & 1/2 & j_c \nonumber \\
j_4 & 1 & j_5 \end{array } \right\ } \left(-1 \right)^{j } \nonumber \\ & & \times \langle \ n_a l_a \parallel j_{j+ \kappa } \left ( qr_1 \right ) \left [ y_l \left ( \omega_1 \right ) \otimes \left ( \vec { \nabla}_1 - \vec { \nabla}_1 ' \right ) \right]_{j_4 } \parallel n_c l_c \rangle_{r_1 } \langle \ b \parallel \left [ y_{l_2 } \left ( \omega_2 \right ) \otimes \vec { \sigma}_2 \right]_{j_3 } \parallel d \
\rangle_{r_2 } \nonumber \\ & & + \delta_{bd , p } \ ; \widehat{l}_2 \widehat{j}_b \widehat{j}_d \left ( \begin{array}{ccc } l & l_2 & j\!\!+\!\ ! \kappa \nonumber \\ 0 & 0 & 0 \end{array } \right ) \left\ { \begin{array}{ccc } 1 & j & j\!\!+\!\ !
\kappa \nonumber \\ l_2 & l & j_4 \end{array } \right\ } \left\ { \begin{array}{ccc } 1 & j_3 & l_2 \nonumber \\
j & j_4 & j_5 \end{array } \right\ } \left\ { \begin{array}{ccc } j_a & j_b & j_1 \nonumber \\ j_c & j_d & j_2 \nonumber \\ j_3 & j_5 & j \end{array } \right\ } \left\ { \begin{array}{ccc } l_b & 1/2 & j_b \nonumber \\
l_d & 1/2 & j_d \nonumber \\
j_4 & 1 & j_5 \end{array } \right\ } \left(-1 \right)^{j_3 + j_5 } \nonumber \\ & & \times \langle \ a \parallel \left [ y_{l_1 } \left ( \omega_1 \right ) \otimes \vec { \sigma}_1 \right]_{j_3 } \parallel c \ \rangle_{r_1 } \langle \ n_b l_b \parallel j_{j+ \kappa } \left ( qr_2 \right ) \left [ y_l \left ( \omega_2 \right ) \otimes \left ( \vec { \nabla}_2 - \vec { \nabla}_2 ' \right ) \right]_{j_4 } \parallel n_d l_d \
\rangle_{r_2 } \biggr\ } \;.\end{aligned}\ ] ] in a completely analogous manner as outlined for the tensor correlations in section [ sec : tensor ] , the effective two - body operators and matrix elements corresponding with the coupling of a virtual photon to a spin - isospin correlated dinucleon , i.e. a nuclear pair correlated through the operator @xmath168 , can be constructed .
the effective coulomb operator of rank @xmath169 and corresponding reduced two - body matrix element for longitudinal electromagnetic coupling to a spin - isospin correlated pair becomes @xmath170 } \left ( i , j \right ) \right ] & = & % \nonumber \\ \sum_{ll } \sum_{j_3 j_4 } \sqrt { 4 \pi}e \frac { \widehat{l } \widehat{j}_3 \widehat{j}_4 } { \widehat{l } } \ { \cal x}^{\sigma \tau } \left(l , r_i , r_j \right ) \left ( \vec { \tau}_i \cdot \vec { \tau}_j \right ) \left ( \begin{array}{ccc } j & l & l \nonumber \\ 0 & 0 & 0 \end{array } \right ) \left\ { \begin{array}{ccc } j_3 & l & 1 \nonumber \\
l & j_4 & j \end{array } \right\ } \nonumber \\ & & \times \biggl\ { \left ( -1 \right)^{j_4+l } j_j \left ( q r_i \right ) \biggl [ \biggl [ y_l \left ( \omega_i \right ) \otimes \vec { \sigma}_i \biggr]_{j_3 } \otimes \biggl [ y_l \left ( \omega_j \right ) \otimes \vec { \sigma}_j \biggr]_{j_4 } \biggr]_j^m \frac{1 + \tau_{z , i}}{2 } \nonumber \\ & & + \left ( -1 \right)^{j_3+l+j } j_j \left ( q r_j \right ) \biggl [ \biggl [ y_l \left ( \omega_i \right ) \otimes \vec { \sigma}_i \biggr]_{j_4 } \otimes \biggl [ y_l \left ( \omega_j \right ) \otimes \vec { \sigma}_j \biggr]_{j_3 } \biggr]_j^m \frac{1 + \tau_{z , j}}{2 } \biggr\ } \;,\end{aligned}\ ] ] @xmath171 } \left ( 1,2 \right ) \right ] \parallel cd ; j_2 \rangle = \sum_{ll } \sum_{j_3 j_4 } \int \ ! d r_1 \int \ ! d
r_2 \sqrt { 4 \pi}e \frac { \widehat { j}_1 \widehat { j}_2 \widehat { j}_3 \widehat { j}_4 \widehat{l } } { \widehat{l } } \langle l \ 0 \ l \ 0 \mid j \ 0 \rangle \
\left\ { \begin{array}{ccc } j_3 & l & 1 \nonumber \\
l & j_4 & j \end{array } \right\ } { \cal x}^{\sigma \tau } \left ( l , r_1 , r_2 \right ) \nonumber \\ & & \times \biggl [ \delta_{ac , p } \left ( -1 \right)^{l + j_4 } \left\ { \begin{array}{ccc } j_a & j_b & j_1 \nonumber \\ j_c & j_d & j_2 \nonumber \\ j_3 & j_4 & j \end{array } \right\ } \langle \ a \parallel j_j \left ( q r_1 \right ) \biggl [ y_l \left ( \omega_1 \right ) \otimes \vec { \sigma}_1 \biggr]_{j_3 } \parallel c \ \rangle_{r_1 } \langle \ b \parallel \biggl [ y_l \left ( \omega_2 \right ) \otimes \vec { \sigma}_2 \biggr]_{j_4 } \parallel d
\ \rangle_{r_2 } \nonumber \\ & & + \delta_{bd , p } \left ( -1 \right)^{l+j+j_3 } \left\ { \begin{array}{ccc } j_a & j_b & j_1 \\ j_c & j_d & j_2 \\ j_4 & j_3 & j \end{array } \right\ } \langle \ a \parallel \biggl [ y_l \left ( \omega_1 \right ) \otimes \vec { \sigma}_1 \biggr]_{j_4 } \parallel c \ \rangle_{r_1 } \langle \ b
\parallel \biggl [ j_j \left ( q r_2 \right ) y_l \left ( \omega_2 \right ) \otimes \vec { \sigma}_2 \biggr]_{j_3 } \parallel d \ \rangle_{r_2 } \biggr ] \;.\end{aligned}\ ] ] the combination of the transverse magnetization current and the spin - isospin correlation operator @xmath174 leads to the following transverse multipole operator @xmath175,magn } \left ( i , j \right ) \right ] = \sum_{ll } \sum_{j_3 j_4 } \sum_{j_5 j_6 } \sum _ { \eta= \pm1 } \frac { i e \hbar q}{2 m_n } \sqrt{24 \pi } \frac { \widehat{l } \widehat{j}_3 \widehat{j}_4 ( \widehat{j}_5)^2 \widehat{j}_6 } { \widehat{l}}\ { \cal x}^{\sigma \tau } \left(l , r_i , r_j \right ) \sqrt { j + \kappa + \delta _ { \eta , + 1 } } \nonumber \\ & & \times \langle l \ 0 \ l \ 0 \mid j\!\!+\!\ ! \kappa \!\!+ \!\ ! \eta \ 0
\rangle \left\ { \begin{array}{ccc } 1 & 1 & 1 \nonumber \\
j\!\!+\!\ ! \kappa & j\!\!+\!\ ! \kappa\!\ !
+ \!\!\eta & j \end{array } \right\ } \left\ { \begin{array}{ccc } l & l & j\!\!+\!\ ! \kappa\!\ ! + \!\!\eta \nonumber \\ 1 & j & j_5 \end{array } \right\ } \left\ { \begin{array}{ccc } j_5 & 1 & l \nonumber \\ j_3 & 1 & j_6 \end{array } \right\ } \left\ { \begin{array}{ccc } j_6 & j_5 & 1 \nonumber \\
l & j_4 & j \end{array } \right\ } \nonumber \\ & & \times \biggl\ { \left [ \mu_p \delta_{i , p } + \mu_n \delta_{i , n } \right ] \biggl [ \biggl [ \left [ y_l \left ( \omega_i \right ) \otimes \vec { \sigma}_i \right]_{j_3 } \otimes \vec { \sigma}_i \biggr]_{j_6 } \otimes \left [ y_l \left ( \omega_j \right ) \otimes \vec { \sigma}_j \right]_{j_4 } \biggr]_j^m j _ { j+\kappa + \eta } \left ( qr_i \right ) \left ( -1 \right)^ { l+j_3+j_4+j_5+j+\kappa+1 } \nonumber \\ & & + \left [ \mu_p \delta_{j , p } + \mu_n \delta_{j , n } \right ] \biggl [ \left [ y_l \left ( \omega_i \right ) \otimes \vec { \sigma}_i \right]_{j_4 } \otimes \biggl [ \left [ y_l \left ( \omega_j \right ) \otimes \vec { \sigma}_j \right]_{j_3 } \otimes \vec { \sigma}_j \biggr]_{j_6 } \biggr]_j^m j _ { j+\kappa + \eta } \left ( qr_j \right ) \left ( -1 \right)^ { l+j_3+j_5+j_6+\kappa+1 } \biggr\ } \ ; .\end{aligned}\ ] ] the reduced matrix element of this multipole operator becomes : @xmath176,magn } \left ( 1,2 \right ) \right ] \parallel cd ; j_2 \rangle = \sum_{ll } \sum_{j_3 j_4 } \sum_{j_5 j_6 } \sum_{j } \sum _ { \eta = \pm1 } \int \ !
d r_1 \int \ ! d r_2 \frac { i e \hbar q}{2 m_n } 12 \sqrt { \pi }
\frac { \widehat{l } \widehat{j}_1 \widehat{j}_2 \widehat{j}_3 \widehat{j}_4 ( \widehat{j}_5)^2 ( \widehat{j}_6)^2 \widehat{j } \widehat{j } } { \widehat{l } } { \cal x}^{\sigma \tau } \left ( l , r_1 , r_2 \right ) \nonumber \\ & & \times \sqrt { j \!+ \
! \kappa \!+ \ !
\delta _ { \eta , + 1 } } \
\langle l \ 0 \ l \ 0 \mid j\!\!+\!\ ! \kappa \!\!+ \!\ ! \eta \ 0
\rangle \left\ { \begin{array}{ccc } 1 & 1 & 1 \nonumber \\
j\!\!+\!\ ! \kappa & j\!\!+\!\ ! \kappa\!\
! + \!\!\eta & j \end{array } \right\ } \left\ { \begin{array}{ccc } l & l & j\!\!+\!\ ! \kappa\!\ ! + \!\!\eta \nonumber \\ 1 & j & j_5 \end{array } \right\ } \left\ { \begin{array}{ccc } j_5 & 1 & l \nonumber \\ j_3 & 1 & j_6 \end{array } \right\ } \left\ { \begin{array}{ccc } j_6 & j_5 & 1 \nonumber \\
l & j_4 & j \end{array } \right\ } \nonumber \\ & & \times \biggl [ \left [ \mu_p \delta_{ac , p } + \mu_n \delta_{ac , n } \right ] \widehat{j}_c \left ( -1 \right)^{l+j_3+j_4+j_5+j_6+j+\kappa+j+j_a+j_c+l_c+1/2 } \nonumber \\ & & \times \left\ { \begin{array}{ccc } j_3 & 1 & j_6 \nonumber \\ j_c & j_a & j \end{array } \right\ } \left\ { \begin{array}{ccc } l_c & 1/2 & j \nonumber \\ 1 & j_c & 1/2 \end{array } \right\ } \left\ { \begin{array}{ccc } j_a & j_b & j_1 \nonumber \\ j_c & j_d & j_2 \nonumber \\ j_6 & j_4 & j \end{array } \right\ } \nonumber \\ & & \times \langle \ a \parallel j_{j+ \kappa+ \eta } \left ( qr_1 \right ) \left [ y_l \left ( \omega_1 \right ) \otimes \vec { \sigma}_1 \right]_{j_3 } \parallel n_c l_c \frac{1}{2 } j \ \rangle_{r_1 } \langle \ b
\parallel \left [ y_l \left ( \omega_2 \right ) \otimes \vec { \sigma}_2 \right]_{j_4 } \parallel d \
\rangle_{r_2 } \nonumber \\ \nonumber \\ & & + \left [ \mu _ { \pi } \delta_{bd , \pi } + \mu _ { \nu } \delta_{bd , \nu } \right ] \widehat{j}_d \left ( -1 \right)^{l+j_3+j_5+\kappa+j+j_b+j_d+l_d+1/2 } \nonumber \\ & & \times \left\ { \begin{array}{ccc } j_3 & 1 & j_6 \nonumber \\ j_d & j_b & j \end{array } \right\ } \left\ { \begin{array}{ccc } l_d & 1/2 & j \nonumber \\ 1 & j_d & 1/2 \end{array } \right\ } \left\ { \begin{array}{ccc } j_a & j_b & j_1 \nonumber \\ j_c & j_d & j_2 \nonumber \\ j_4 & j_6 & j \end{array } \right\ } \nonumber \\ & & \times \langle \ a \parallel \left [ y_l \left ( \omega_1 \right ) \otimes \vec { \sigma}_1 \right]_{j_4 } \parallel c \ \rangle_{r_1 } \langle \
b \parallel j_{j+ \kappa+ \eta } \left ( qr_2 \right ) \left [ y_l \left ( \omega_2 \right ) \otimes \vec { \sigma}_2 \right]_{j_3 } \parallel n_d l_d \frac{1}{2 } j \ \rangle_{r_2 } \biggr ] \;.\end{aligned}\ ] ] for both the central and the tensor correlation term the numerical calculations indicate that the transverse strength is almost exclusivly due to the magnetization current , the convection current producing very small matrix elements . for that reason we have neglected the convection current contribution when evaluating the spin - spin correlations .
99 l.j.h.m .
et al . _ ,
b366 ( 1996 ) 44 .
frankfurt , m.m .
sargsian and m.i .
strikman , phys .
c 56 ( 1997 ) 1124 .
i. sick in proceedings of the workshop on electron nucleus scattering , eds omar benhar and adelchi fabrocini ( edizioni ets , pisa , 1997 ) p. 445 .
p. demetriou , a. gil , s. boffi , c. giusti , e. oset and f.d .
pacati , nucl .
a650 ( 1999 ) 199 .
a. gil , j. nieves and e. oset , nucl .
a627 ( 1997 ) 599 .
o. benhar , s. fantoni and g.i .
lykasov , eur .
phys . j a 5 ( 1999 ) 137 .
h. morita , c. ciofi degli atti and d. treleani , phys .
c 60 ( 1999 ) 034603 .
claudio ciofi degli atti and daniele treleani , phys .
c 60 ( 1999 ) , 024602 .
et al . _ ,
59 ( 1987 ) 2259 .
lourie , w. bertozzi , j. morrison and l.b .
weinstein , phys .
c 57 ( 1993 ) r444 .
n. liyanage , ph.d .
thesis , mit ( 1999 ) .
m. holtrop _ et al .
c 58 ( 1998 ) 3205 .
d. dutta , ph.d .
thesis , northwestern university ( 1999 ) .
j. ryckebusch , v. van der sluys , k. heyde , h. holvoet , w. van nespen , m. waroquier and m. vanderhaeghen , nucl .
phys . a 624 ( 1997 ) 581 .
j. ryckebusch , m. vanderhaeghen , l. machenil and m. waroquier , nucl .
phys . a 568 ( 1994 ) 828 - 854 .
pieper , r.b .
wiringa and v.r .
pandharipande , phys .
c 46 ( 1992 ) 1741 .
o. benhar and v.r .
pandharipande , rev .
modern phys .
65 ( 1993 ) 817 .
r. guardiola , p.i .
moliner , j. navarro , r.f .
bishop , a. puente and niels r. walet , nucl .
phys . a 609 ( 1996 ) 218 .
pandharipande and r.b .
wiringa , rev .
51 ( 1979 ) 821 . s. fantoni and v.r .
pandharipande , nucl .
phys . a 473 ( 1987 ) 234 .
h. feldmeier , t. neff , r. roth and j. schnack , nucl .
phys . a 632 ( 1998 ) 61 .
gerald a. miller and james e. spencer , annals of physics 100 ( 1976 ) 562 .
m. vanderhaeghen , l. machenil , j. ryckebusch and m. waroquier , nucl .
phys . a 580 ( 1994 ) 551 .
d. abbott _
et al . _ ,
80 ( 1998 ) 5072 .
c.c.gearhart , phd thesis , washington university ( st . louis , 1994 ) , unpublished and w. dickhoff , private communication .
et al . _ ,
b 421 ( 1998 ) 71 .
jan ryckebusch , dimitri debruyne , wim van nespen and stijn janssen , phys .
c 60 ( 1999 ) 034604 .
et al . _ ,
81 ( 1998 ) 2213 .
r. starink em et al . , in press and ph.d .
thesis , free university of amsterdam ( 1999 ) , unpublished .
d. van neck , l. van daele , y. dewulf and m. waroquier , phys .
c 56 ( 1997 ) 1398 .
van leeuwe _
et al . _ ,
phys . a 631 ( 1998 ) 593c .
templon , nucl - ex/9902008 .
tjnaf experiment e97 - 106 `` studying the internal small - distance structure of nuclei via the triple coincidence ( @xmath177 ) measurement '' ( spokesperson e. piasetzky ) .
a. bianconi , s. jeschonnek , n.n .
nikolaev and b.g .
zakharov , phys .
b 363 ( 1995 ) 217 . | the effect of central , tensor and spin - isospin nucleon - nucleon correlations upon semi - exclusive @xmath0o@xmath1 reactions is studied for @xmath2 and bjorken @xmath3 values in the range @xmath4(gev / c)@xmath5 and 0.15 @xmath6 @xmath3 @xmath6 2 .
the fully unfactorized calculations are performed in a framework that accounts not only for the dynamical coupling of virtual photons to correlated nucleon pairs but also for meson - exchange and @xmath7-isobar currents .
tensor correlations are observed to produce substantially larger amounts of semi - exclusive @xmath0o@xmath1 strength than central correlations do and are predominantly manifest in the proton - neutron knockout channel . with the exception of the @xmath8 case ,
in all kinematical situations studied the meson - exchange and isobar currents are a strong source of @xmath9 strength at deep missing energies .
this feature gives the @xmath9 strength at deep missing energies a pronounced transverse character .
_ pacs : _
25.30.-c,24.10.-i,25.30.fj _ keywords _ : nucleon - nucleon correlations , semi - exclusive @xmath1 |
we begin by defining the new matroid invariant . for matroid terminology
undefined here , we refer the reader to some of the standard references , such as @xcite .
let @xmath1 be a matroid on ground set @xmath2 , with bases @xmath3 .
let @xmath4 be the positive integers .
we will say that a weighting function @xmath5 is _ @xmath6-generic _ if the minimum @xmath7-weight @xmath8 among all bases @xmath9 of @xmath6 is achieved by a _ unique _ base @xmath10 .
for example , it is a standard exercise in matroid theory ( see , e.g. ( * ? ? ? * exer .
1.8.4 ) ) to show that @xmath7 is @xmath6-generic if @xmath7 is _ injective _ , that is , if @xmath7 assigns all distinct weights .
[ definition - of - f ] given a matroid @xmath6 as above , define a power series @xmath11 in countably many variables @xmath12 as the generating function for @xmath6-generic weighting functions @xmath7 according to number of times @xmath7 takes on each value in @xmath13 .
that is , @xmath14 where @xmath15 .
one of the defining properties of a matroid ( * ? ? ?
* theorem 1.8.5 ) is that an @xmath7-minimizing base may be found by _
( kruskal s ) greedy algorithm _ : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ construct a sequence of independent sets @xmath16 by defining @xmath17 where @xmath18 is any element in @xmath2 having minimum weight @xmath19 among those for which @xmath20 is independent .
then @xmath21 is an @xmath7-minimizing base of @xmath6 .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
we recall @xcite , @xcite what it means for a power series @xmath22 in a linearly ordered variable set @xmath12 to be _ quasisymmetric _ : @xmath7 must have bounded degree , and for any fixed composition @xmath23 in @xmath24 , the coefficient of the monomials @xmath25 with @xmath26 in @xmath7 are all the same . put differently , @xmath7 is quasisymmetric if and only if it is a ( finite ) linear combination of the _ monomial _ quasisymmetric functions and monomial quasisymmetric functions @xmath27 , the difference will always be clear by the context . ] indexed by compositions @xmath28 : @xmath29 [ quasisymmetry ] for any matroid @xmath6 , the power series @xmath11 is _
quasisymmetric_. this follows from the fact that the @xmath7-minimum bases can all be found by the greedy algorithm , and this algorithm makes all of its decisions based only on the _ relative ordering _ and equality of various weights @xmath19 , not on their actual values .
[ tiny - examples ] when @xmath30 , there is only one matroid @xmath31 , having rank @xmath32 and exactly one base , the empty base @xmath33 . as there is only one function @xmath7 from the empty set @xmath2 into @xmath13 , and this @xmath7
has no coordinates ( ! ) , we should decree @xmath34 ( as the empty product is @xmath35 ) . hence @xmath36 .
there are two matroids with @xmath37 , namely @xmath38 of rank @xmath35 having a single base @xmath39 , and @xmath40 of rank @xmath32 having a single base @xmath33 .
every @xmath41 is generic for either of these , so that @xmath42 the enumerative information recorded in @xmath11 is data about optimizing weight functions on the bases of @xmath6 .
an obvious specialization counts @xmath6-generic weight functions that take on only a limited number of distinct weight values . for a positive integer @xmath43 ,
let @xmath44:=\{1,2,\ldots , m\}$ ] , and define @xmath45 \}| . \end{aligned}\ ] ] since @xmath11 is a power series of bounded degree , @xmath46 is a polynomial function of @xmath43
. when @xmath43 is large , almost all weight functions @xmath47 $ ] are injective and hence @xmath6-generic , so the polynomial expansion of @xmath46 begins @xmath48 where @xmath49 . in section [ antipodes ]
, our analysis of the behavior of @xmath11 under the hopf algebra antipode on quasisymmetric functions will imply an interesting reciprocity result for the polynomial @xmath46 . in the remainder of this paper
, we will suppress the @xmath50 in @xmath11 and write @xmath51 unless there is a need to consider the variables .
there is a known hopf algebra structure built from matroids @xcite and a perhaps better - known hopf algebra of quasisymmetric functions @xcite .
the goal of this section is to show that the invariant @xmath51 defines a hopf morphism between them .
let @xmath52 be the free @xmath53-module consisting of formal @xmath53-linear combinations of basis elements @xmath54 $ ] indexed by isomorphism classes of matroids @xmath6 .
endow @xmath52 with a product and coproduct extended @xmath53-linearly from the following definitions on basis elements : @xmath55 \cdot [ m_2 ] & : = [ m_1 \oplus m_2 ] \\
\delta [ m ] : & = \sum_{a \subseteq e } [ m|_a ] \otimes [ m / a ] \end{aligned}\ ] ] where @xmath56 is the _ direct sum _ of the matroids @xmath57 , and @xmath58 denote the _ restriction _ of @xmath6 to @xmath59 and the _ contraction _ ( or _ quotient _ ) of @xmath6 by @xmath59 , respectively .
one has a @xmath53-module direct sum decomposition @xmath60 , where @xmath61 denotes the submodule spanned by the basis elements @xmath54 $ ] for which the ground set @xmath2 of @xmath6 has cardinality @xmath62 .
one can then easily check that this product and coproduct make @xmath52 into a graded , connected hopf algebra over @xmath53 which is commutative , but non - cocommutative . here
the unit is @xmath63 $ ]
. let @xmath64 ( or @xmath65 ) denote the hopf algebra of quasisymmetric functions in the linearly ordered variable set @xmath12 and having coefficients in @xmath53 .
the product in @xmath64 is inherited from the formal power series ring @xmath66 $ ] .
the coproduct may be described as follows .
a quasisymmetric function @xmath22 defines a unique quasisymmetric function @xmath67 in the linearly ordered variable set @xmath68 by insisting that @xmath69 . in other words , for any @xmath70 and @xmath71 , the coefficient of @xmath72 in @xmath67 is defined to be the coefficient of @xmath73 in @xmath22 .
consider the injective map @xmath74 \otimes { { \mathbb z}}[[y_1,y_2,\ldots ] ] \rightarrow { { \mathbb z}}[[x_1,x_2,\ldots , y_1,y_2,\ldots]]\ ] ] which sends @xmath75 to @xmath76 .
the image @xmath77 contains the quasisymmetric functions @xmath78 , that is , there is a unique expansion @xmath79 for any quasisymmetric function @xmath67 .
this defines the coproduct @xmath80 .
grading @xmath64 by the usual notion of degree , one can check that @xmath64 becomes a graded , connected hopf algebra over @xmath53 which is commutative , but non - cocommutative .
[ morphism - theorem ] the map @xmath81 @xmath82 & \mapsto & f(m ) \end{matrix}\ ] ] is a morphism of hopf algebras .
example [ tiny - examples ] shows that @xmath81 sends the unit @xmath31 of @xmath52 to the unit @xmath35 of @xmath64 .
the fact that @xmath81 preserves degree shows that it preserves the counit .
the fact that @xmath81 preserves the product structures follows because the bases of @xmath56 are the disjoint unions @xmath83 of a base @xmath84 from each .
this implies that @xmath85 is @xmath86-generic if and only if @xmath87 is @xmath88-generic for @xmath89 .
the fact that @xmath81 preserves the coalgebra structure is somewhat more interesting . unravelling the definitions ,
this amounts to checking the following identity : @xmath90 the left side of has the following interpretation .
linearly order the disjoint union @xmath91 as follows : @xmath92 given a weight function @xmath93 , define @xmath94 where @xmath95 then @xmath96 on the other hand , the right side of expands to @xmath97 , where the sum ranges over all triples @xmath98 in which 1 .
@xmath59 is a subset of @xmath2 , 2 .
@xmath99 is @xmath100-generic , and 3 .
@xmath101 is @xmath102-generic .
there is an obvious association @xmath103 defined by @xmath104 it only remains to check that @xmath7 is @xmath6-generic if and only if @xmath105 and @xmath106 are @xmath100 and @xmath102-generic , respectively .
this follows from the sequential nature of the greedy algorithm : because the primed values @xmath107 are bigger than all the unprimed values @xmath108 , when the greedy algorithm finds @xmath7-minimizing bases for @xmath6 , it must first find @xmath105-minimizing bases for @xmath100 by trying to use only @xmath18 s with unprimed values for as long as it can , and then proceed to find @xmath109-minimizing bases for @xmath102 using primed values .
lack of uniqueness in the @xmath7-minimizing bases of @xmath6 can only occur if it occurs in one of these two steps , leading either to lack of uniqueness in the @xmath105-minimizing bases of @xmath100 or in the @xmath106-minimizing bases of @xmath102 .
conversely , lack of uniqueness in either step will lead to lack of uniqueness for the whole computation .
it turns out that the hopf morphism @xmath110 is _ not _ surjective if one works over @xmath53 , but becomes surjective after tensoring with the rationals .
the somewhat technical proof of this surjectivity is given in the appendix ( section [ appendix ] ) .
the proof involves the construction of two new @xmath53-bases for @xmath64 , which may be of independent interest .
( _ on combinatorial hopf algebras _ )
+ definition [ definition - of - f ] for @xmath51 immediately implies that for any composition @xmath111 of @xmath49 , the coefficient @xmath112 in the unique expansion @xmath113 has the following interpretation : @xmath112 is the number of @xmath6-generic @xmath41 in which @xmath114 for @xmath115 .
the work of aguiar , bergeron and sottile @xcite on combinatorial hopf algebras also offers an interpretation for @xmath112 , using the fact that @xmath81 is a hopf morphism , as we explain here . in their theory , the _ character _ (= multiplicative linear functional ) @xmath116 defined by @xmath117 plays a crucial role , making @xmath64 into what they call a _ combinatorial hopf algebra_. the hopf morphism @xmath118 then allows one to uniquely define a character @xmath119 , via @xmath120 , so that @xmath81 becomes a _
morphism of combinatorial hopf algebras_. it is not hard to see directly ( or one can appeal to corollary [ zeta - computation ] below ) the following more explicit description of the character @xmath121 . say that a matroid @xmath6 _ splits completely _ if it is a direct sum of matroids on @xmath35 element , that is , a direct sum of loops and isthmuses , or equivalently , if it has only one base @xmath9 .
then for any matroid @xmath6 @xmath122 _ q([m ] ) = 1 & @xmath6 + 0 & @xmath122 using this , ( * ? ? ? * theorem 4.1 ) immediately implies another interpretation for the coefficient @xmath112 in .
given a flag @xmath123 of subsets @xmath124 where @xmath2 is the ground set for the matroid @xmath6 , let @xmath125 be the composition of @xmath49 defined by @xmath126 .
[ abs - computation ] the coefficient @xmath112 in is the number of flags @xmath123 of subsets of @xmath2 having @xmath127 and for which each subquotient @xmath128 splits completely .
the equivalence of these two interpretations of @xmath112 is easy to understand .
any @xmath5 with @xmath114 for @xmath115 gives rise to a flag @xmath123 of subsets as in with @xmath127 , by letting @xmath129 .
in other words , @xmath7 takes on the constant value @xmath108 on each of the set differences @xmath130 .
one can then readily see ( e.g. from the greedy algorithm ) that @xmath7 will be @xmath6-generic if and only if each of the subquotients @xmath128 has only one base , that is , if and only if each such subquotient splits completely .
a consequence of this equivalence is that ( * ? ? ?
* theorem 4.1 ) gives an alternate proof of theorem [ morphism - theorem ] above .
given this discussion , the existence of canonical _ odd and even subalgebras _ inside any combinatorial hopf algebra ( see @xcite ) naturally suggests the following question .
[ eulerian - question ] what is the odd subalgebra of the combinatorial hopf algebra @xmath52 ? does it contain any elements of the form @xmath54 $ ] for a single matroid @xmath6 , or does it contain only nontrivial sums @xmath131 $ ] ? for any such @xmath54 $ ] in the odd subalgebra of @xmath52 , it will follow from ( * ? ? ? * propositions 5.8e and 6.5 ) that @xmath51 will lie in the _ peak subalgebra _ of @xmath64 ( see @xcite for definitions ) .
one can use proposition [ abs - computation ] to compute some more examples of @xmath51 . if @xmath6 is a rank @xmath35 matroid on ground set @xmath132 in which @xmath133 are parallel elements , then there are exactly two flags @xmath123 having all subquotients that split completely : @xmath134 both of these flags have @xmath135 and hence @xmath136 . similarly , if @xmath6 is a rank @xmath35 matroid on ground set @xmath137 in which @xmath138 are all parallel , then there are two kinds of flags @xmath123 having all subquotients that split completely : 1 .
@xmath139 flags of the form @xmath140 where @xmath141 is some permutation of @xmath142 , all having @xmath143 , and 2 .
@xmath144 flags of the form @xmath145 where @xmath146 , all having @xmath147 .
consequently , @xmath148 .
recall that if @xmath6 is a matroid on ground set @xmath2 with bases @xmath149 , then its _ dual _ ( or _ orthogonal _ ) matroid @xmath150 has the same ground set @xmath2 but bases @xmath151 where @xmath152 is called the _ cobase _ of @xmath150 corresponding to the base @xmath9 of @xmath6 .
[ duality - proposition ] @xmath153 if and only if @xmath154 where @xmath155 is the reverse composition to @xmath156 .
we check that for any composition @xmath157 , the coefficient of @xmath158 in @xmath51 is the same as the coefficient of @xmath159 in @xmath160 .
the former coefficient counts the set of @xmath6-generic @xmath41 for which @xmath161 .
the latter coefficient counts the set of @xmath150-generic @xmath162 for which @xmath163 .
we exhibit a bijection between these sets as follows .
if @xmath9 is a base of @xmath6 with cobase @xmath164 of @xmath150 , then the equation @xmath165 shows that @xmath9 is @xmath7-minimizing if and only if @xmath164 is @xmath7-maximizing .
now define @xmath166 , so that one has @xmath167 this equation shows that @xmath164 is @xmath7-maximizing if and only if @xmath164 is @xmath168-minimizing . since @xmath169 if and only if @xmath170 , the map @xmath171 restricts to the desired bijection .
quasisymmetric functions were originally introduced by gessel @xcite ( building on work of stanley ) as enumerators for @xmath0-partitions .
we review this here , and explain how it leads to an expansion of @xmath51 as a sum of @xmath0-partition enumerators . a _ labelled poset
_ @xmath172 on @xmath173 elements is a poset @xmath0 together with a bijective labelling function @xmath174:=\{1,2,\ldots , n\}$ ] .
a _ @xmath172-partition _ is a function @xmath175 such that @xmath176 it will sometimes be more convenient for us to refer only to a labelled poset @xmath0 on @xmath177 $ ] ( suppressing the extra labeling function @xmath178 ) , by which we mean a partial order @xmath179 on the set @xmath177 $ ] . using this terminology ,
a @xmath0-partition is a function @xmath180 \rightarrow { { \mathbb p}}$ ] satisfying @xmath181 for example , every permutation @xmath182 of @xmath177 $ ] can be regarded as a labelled poset on @xmath177 $ ] which is totally ordered : @xmath183 .
let @xmath184 denote the set of @xmath172-partitions , and let @xmath185 be their weight enumerator : @xmath186 a basic result of stanley tells how @xmath187 expands in terms of another basis for @xmath64 indexed by compositions @xmath156 , known as the _ fundamental quasisymmetric functions _
say that a permutation @xmath189 in the symmetric group @xmath190 is a _ linear extension _ of @xmath172 if @xmath191 in @xmath0 implies @xmath192 .
jordan - hlder set _ of @xmath172 is the set @xmath193 of all linear extensions of @xmath172 .
the _ descent composition _ for the permutation @xmath194 is the composition @xmath195 of @xmath173 which gives the lengths of the maximal increasing consecutive subsequences ( _ runs _ ) of @xmath194 .
it is not hard to check that , regarding @xmath194 as a totally ordered labelled poset on @xmath177 $ ] as above , one has @xmath196 .
the basic result about @xmath0-partitions is the following expansion .
@xcite , @xcite , ( * ? ? ?
( 1 ) ) [ stanleys - p - partition - result ] @xmath197 it turns out that every base @xmath9 of a matroid @xmath6 leads to a certain labelled poset @xmath198 , whose @xmath0-partition enumerator is relevant for expanding @xmath51 ; see theorem [ poset - expansion ] below . given a base @xmath9 of a matroid @xmath6 on ground set @xmath2 ,
let @xmath199 be the corresponding cobase of @xmath150
. for each @xmath200 the _ basic bond for @xmath18 in @xmath164 _ is the set of @xmath201 for which @xmath202 is another base of @xmath6 .
dually , for each @xmath203 the _ basic circuit for @xmath18 in @xmath9 _ is the set of @xmath201 for which @xmath204 is another base of @xmath6 . by definition
then , one has a symmetric relationship : @xmath205 lies in the basic bond for @xmath18 in @xmath164 if and only if @xmath18 lies in the basic circuit for @xmath205 in @xmath9 .
thus these relations can be encoded by a bipartite graph with vertex set @xmath2 , bipartitioned as @xmath206 .
define the poset @xmath198 to be the one whose hasse diagram is this bipartite graph , with edges directed upward from @xmath9 to @xmath164 .
say that a labelling @xmath178 of a poset @xmath0 is _ natural _ ( resp .
_ strict _ or _ anti - natural _ ) if @xmath207 ( resp .
@xmath208 ) whenever @xmath191 in @xmath0 .
[ poset - expansion ] for any matroid @xmath6 , @xmath209 where @xmath210 is any strict labelling of @xmath198 .
we will show that @xmath9 is the unique @xmath7-minimizing base of @xmath6 for some @xmath41 if and only if @xmath7 lies in @xmath211 .
first assume that @xmath7 does not lie in @xmath211 , that is , there exists some @xmath212 in @xmath198 for which @xmath213 .
by definition of @xmath198 , this means that @xmath18 lies in @xmath9 , @xmath205 does not lie in @xmath9 , and @xmath214 is another base of @xmath6 .
however @xmath215 implies @xmath216 , so that @xmath9 can not be the unique @xmath7-minimizing base .
now assume that @xmath9 is not the unique @xmath7-minimizing base of @xmath6 .
this means that there exists another base @xmath217 of @xmath6 having @xmath216 . by convexity
, we may assume that the pair @xmath218 corresponds to an edge of the _ matroid base polytope _ @xmath219 , which is defined to be the convex hull in @xmath220 of all characteristic @xmath221-vectors of bases of @xmath6 ( see section [ decomposition - section ] below ) . a well - known fact from matroid theory ( * ? ? ?
* , theorem 1 ) says that all edges of @xmath219 take the form @xmath218 in which @xmath222 differ by a single _ basis exchange _ : there exists some @xmath200 and @xmath223 such that @xmath224 .
thus @xmath212 in @xmath198 .
since @xmath216 forces @xmath215 , this means @xmath7 is not in @xmath211 .
aguiar has pointed out that theorem [ poset - expansion ] shows the hopf morphism @xmath225 factoring through the _ hopf algebra @xmath226 of ( labelled ) posets _ , which is described ( for unlabelled posets ) in ( * ? ? ?
* example 2.3 ) .
more precisely , one has a hopf morphism @xmath227 & \mapsto & \sum_{b \in { { \mathcal b}}(m)}[(p_b,\gamma_b ) ] \end{matrix}\ ] ] and the usual @xmath172-partition enumerator hopf morphism @xmath228 & \mapsto & f(p,\gamma,{{\mathbf x } } ) .
\end{matrix}\ ] ] then @xmath225 is the composite of these two morphisms .
[ last - coefficient ] let @xmath229
. then 1 .
the coefficients @xmath230 are nonnegative , 2 .
their sum @xmath231 is @xmath232 where @xmath49 , and 3 .
the coefficient @xmath233 of @xmath234 is the number of bases of @xmath6 .
everything will follow from proposition [ stanleys - p - partition - result ] and theorem [ poset - expansion ] .
assertion ( i ) is immediate .
assertion ( ii ) follows because each of the @xmath232 linear orderings @xmath235 of @xmath2 is a linear extension for exactly one of the posets @xmath198 , namely the one indexed by the unique @xmath7-minimizing base @xmath9 when @xmath236 .
assertion ( iii ) follows because any strictly ( anti - naturally ) labelled poset @xmath172 has the reversing permutation @xmath237 in @xmath193 , and @xmath238 is the _ only _ permutation having descent composition @xmath239 .
corollary [ last - coefficient ] gives a combinatorial interpretation for the coefficient @xmath233 .
it would be nice to have such an interpretation for every coefficient @xmath230 .
the next result at least tells us how to interpret the coefficients `` at the other end '' of the @xmath240 expansion , namely @xmath241 where @xmath156 has at most two parts , in terms of some basic matroid invariants of @xmath6 . recall that an element @xmath18 in @xmath2 is a _ loop _ in @xmath6 if it appears in _ no _ bases of @xmath6 , and it is a _
coloop _ ( or _ isthmus _ ) if it appears in _ every _ base of @xmath6 .
[ first - coefficients ] let @xmath6 be matroid having 1 .
rank @xmath242 , 2 .
corank @xmath243 , 3 .
number of loops equal to @xmath244 , 4 .
number of coloops equal to @xmath245 , and 5 .
number of bases @xmath246 .
then @xmath247 for some nonnegative coefficients @xmath248 . here
@xmath249 denotes the number of parts in the composition @xmath250 .
equivalently , if @xmath251 is the matroid obtained from @xmath6 by removing all loops and coloops , so that @xmath251 has @xmath252 then @xmath253 for some nonnegative integer coefficients @xmath254 .
the second assertion follows from the @xmath255 case of the first , applying the multiplicative property @xmath256 to the decomposition of @xmath6 as a direct sum of @xmath251 with @xmath257 loops and isthmuses . for the first assertion
, we apply theorem [ poset - expansion ] . for each base @xmath9 ,
the poset @xmath198 will have height one , and decompose into three sets : 1 .
the set @xmath258 of @xmath257 loops and coloops , which are all both minimal and maximal in @xmath198 , 2 .
the set @xmath259 of @xmath260 non - coloop elements in @xmath9 , each of which is minimal but not maximal in @xmath198 , and 3 .
the set @xmath261 of @xmath262 non - loop elements in @xmath164 , each of which is maximal but not minimal in @xmath198 . we are free to choose the strict labelling @xmath210 so that the elements in @xmath259 all have the highest labels , the elements in @xmath261 all have the lowest labels , and the elements in @xmath258 have the labels in between .
how then can one choose a linear extension @xmath194 in @xmath263 so that its descent composition @xmath195 has at most two parts ?
this means that @xmath194 has at most two increasing runs , separated by a unique descent .
because of our chosen labelling of @xmath9 , such a @xmath194 will have the first run of length at least @xmath260 , and the second run of length at least @xmath262 .
furthermore , for any integer @xmath264 in the range @xmath265 $ ] , one can check that there are @xmath266 ways to choose such a @xmath194 in @xmath267 so that it starts with an increasing run of length @xmath268 , followed by its unique descent , and then ends with an increasing run of length @xmath269 : one must place the elements of @xmath259 together with any @xmath264 elements chosen from @xmath258 before the unique descent , and place the elements of @xmath261 together with other @xmath270 elements of @xmath258 after the unique descent .
in particular , the previous proposition tells us the coefficient of @xmath240 in @xmath51 when @xmath156 has only @xmath35 part .
recall from section [ hopf - section ] that a matroid @xmath6 is said to split completely if @xmath6 is a direct sum of loops and isthmuses .
[ zeta - computation ] for @xmath6 a matroid on ground set @xmath2 of size @xmath62 , the expansion of @xmath51 in the @xmath240 ( resp .
@xmath158 ) basis for @xmath64 has the coefficient of @xmath271 ( resp .
@xmath272 ) equal to @xmath35 if @xmath6 splits completely , and @xmath32 otherwise .
the assertion for the coefficient of @xmath271 follows from proposition [ first - coefficients ] .
then the assertion for the coefficient of @xmath272 follows from the expansion of @xmath240 into @xmath273 s .
part of the structure of a hopf algebra is an involutive anti - automorphism known as its _
antipode_. for the hopf algebra of quasisymmetric functions , the antipode @xmath274 is known to be related to combinatorial reciprocity results @xcite .
it turns out to have an interesting effect on @xmath51 , transforming it into a different sort of enumerator for weight functions @xmath5 .
we begin by reviewing how the antipode relates to reciprocity .
the antipode @xmath275 has the following effect on the @xmath240-basis ( * ? ? ?
* corollary 2.3 ) : @xmath276 where @xmath277 denotes the _ weight _ of the composition @xmath156 , and @xmath278 corresponds to the subset @xmath279 \backslash s$ ] if @xmath156 corresponds to the subset @xmath280 of @xmath281 $ ] ( _ i.e. _ @xmath280 is the set of partial sums of @xmath156 ) .
stanley s reciprocity theorem for @xmath0-partitions ( * ? ? ?
* theorem 4.5.7 ) tell us that if @xmath282 are natural and strict labellings of the same poset @xmath0 , then @xmath283 upon specializing @xmath284 to @xmath285 , that is , @xmath286 one obtains @xmath287 where @xmath28 .
then the equality @xmath288 leads immediately to the following reciprocity fact ( cf .
@xcite ) .
if two homogeneous quasisymmetric functions @xmath289 of degree @xmath173 are related by @xmath290 , then their specializations @xmath291 satisfy @xmath292 we can now identify the image of @xmath51 under the antipode @xmath293 in @xmath64 .
define a power series in @xmath12 @xmath294 also define a polynomial in @xmath43 @xmath295 } |\ { f\textrm{-minimizing bases of } m\}| .
\end{aligned}\ ] ] one could argue that these two enumerators @xmath296 are at least as natural to consider as our original @xmath297 .
for example , the expected number of @xmath7-minimizing bases of @xmath6 attained when using at most @xmath43 distinct values for the weights is exactly @xmath298 .
[ reciprocity - theorem ] for any matroid @xmath6 on @xmath173 elements , @xmath299 and consequently , @xmath300 theorem [ poset - expansion ] implies @xmath301 note that since @xmath302 are related by the antipode @xmath293 , they carry equivalent information , a fact which is not completely obvious from their definitions .
the same goes for @xmath46 and @xmath303 .
the goal of this section is to show that the matroid invariant @xmath51 behaves like a valuation on the associated matroid base polytopes @xmath219 , and apply this to the subtle problem of detecting decompositions of these polytopes . by the _ matroid base polytope _ we mean the convex polytope @xmath304 where @xmath305 denotes the @xmath306 standard basis vector in @xmath307 .
this polytope @xmath219 is a face of a polytope first studied by edmonds @xcite , which took as vertices the indicator functions of _ all _ independent sets in @xmath6 ( subsets of bases ) .
we are interested in the existence or non - existence of certain polytopal decompositions of @xmath219 .
a _ matroid base polytope decomposition _ of @xmath219 is a decomposition @xmath308 where 1 .
each @xmath309 is a matroid base polytope for some matroid @xmath88 , and 2 . for each @xmath310 , the intersection
@xmath311 is a face of both @xmath309 and of @xmath312 .
we call such a decomposition a _ hyperplane split _ of @xmath219 if @xmath313 .
we say that @xmath219 is _ decomposable _ if it has a matroid base polytope decomposition with @xmath314 , and _ indecomposable _ otherwise .
we say that the decomposition is _ coherent _ if the @xmath309 are exactly the maximal domains of linearity for some @xmath315-valued piecewise - linear convex function on @xmath219 .
for example , hyperplane splits are always coherent .
coherent matroid base polytope decompositions arise in work of lafforgue @xcite on compactifications of the fine schubert cell of the grassmannian corresponding to the matroid @xmath6 , and in related work by keel and tevelev @xcite , and by hacking , keel and tevelev @xcite . in particular
, lafforgue s work implies that for a matroid @xmath6 represented by vectors in @xmath316 , if @xmath219 is indecomposable , then @xmath6 will be _ rigid _ , that is , @xmath6 will have only finitely many realizations , up to scaling and the action of @xmath317 .
we will need a version of a theorem of lawrence ( * ? ? ? * theorem 16 ) ( see also ( * ? ? ? * corollary iv.1.6 ) ) about polarity , which can be proved by a minor adjustment to the proof of ( * ? ? ?
* theorem iv.1.5 ) .
let @xmath318 denote the usual inner product on @xmath319 .
if @xmath59 is a convex set in @xmath319 , then denote by @xmath320 $ ] its indicator function and by @xmath321 the convex set @xmath322 recall that a closed convex cone @xmath323 is said to be _ pointed _ if it contains no lines . in this case , its _ polar cone _
@xmath324 has a nonempty interior .
for a _ nonzero _ pointed cone @xmath325 , @xmath326 is the interior of @xmath327 . we show that the function @xmath328 acts as a valuation on nonempty closed convex sets .
[ polar_relation ] let @xmath329 be a finite family of nonempty closed convex sets .
if @xmath330=0\ ] ] for real numbers @xmath331 , then @xmath332 = 0.\ ] ] the proof is as in @xcite , except that in theorem iv.1.5 , one defines @xmath333 in this case , the limiting argument of @xcite ( and @xcite ) is not necessary . as in @xcite the association @xmath334 \mapsto [ i(a)]\ ] ] is the specialization to indicator functions @xmath320 $ ] of a linear map @xmath335 where @xmath336 is the algebra of indicator functions of closed convex sets in @xmath337 ( see ( * ? ? ?
i.7.3 ) ) .
the map @xmath338 may be defined as follows : for a function @xmath339 , the value of @xmath340 on a point @xmath341 is given by @xmath342 here @xmath343 denotes the euler characteristic linear functional on @xmath336 ; its value on a function @xmath344 is determined uniquely from knowing that it takes the value @xmath35 on indicator functions of closed convex sets .
we now wish to apply this to a decomposition @xmath345 of matroid base polytopes .
necessarily the @xmath346 will be _ weak images _ ( _ degenerations _ ) of @xmath6 , that is , @xmath347 . if @xmath348 and @xmath349 are matroids of the same rank on the same set @xmath2 , then we define @xmath350 .
we write @xmath351 if @xmath352 ( as opposed to @xmath353 ) .
note that even when @xmath354 , it is not usually a matroid take , for example , the rank @xmath355 matroids @xmath57 having bases @xmath356 so that @xmath357 .
however , when @xmath358 and @xmath359 meet along a common face ( as in a matroid base polytope decomposition ) , and that face is nonempty , the intersection @xmath360 _ will _ be a matroid . [
matroid - intersection ] if @xmath361 and @xmath362 are matroids of the same rank @xmath242 and @xmath363 is a nonempty common face of @xmath358 and @xmath359 , then @xmath364 is a matroid of rank @xmath242 and @xmath365 nonempty faces of matroid base polytopes are matroid base polytopes ( * ? ? ?
* theorem 2 ) , and so the common face @xmath363 must be a matroid base polytope .
the vertices of @xmath363 correspond to common bases of @xmath361 and @xmath362 , that is , to elements of @xmath366 .
suppose @xmath367 is the vertex of @xmath219 corresponding to the base @xmath9 of @xmath6 .
we denote by @xmath368 the closed convex cone generated by the minkowski sum ( translate ) @xmath369 .
its polar @xmath370 is the _ normal cone _ to @xmath219 at @xmath371 .
notice that by the proof of theorem [ poset - expansion ] , the expansion of @xmath51 given there can be written @xmath372 with this , one can prove that @xmath51 acts as a valuation over subdivisions of @xmath219 .
[ valuation ] the association @xmath373 is a valuation on the class of matroid polytopes : if @xmath219 can be subdivided into finitely many matroid polytopes @xmath374 , then @xmath375 with the sum over @xmath376 such that @xmath377 .
any decomposition of @xmath219 induces , for each @xmath378 , a decomposition of @xmath368 into @xmath379 where @xmath380 .
( for notational convenience , we include all @xmath381 and set @xmath382 when @xmath383 . )
this , in turn , leads to an inclusion - exclusion relation ( see , for example , ( * ? ? ?
* lemma i.7.2 ) ) @xmath384 & = \sum_{j}(-1)^{{j-1}}\sum_{i_{1}<i_{2}<\cdots < i_{j } } \left[k_{b}(m_{i_{1}})\cap\cdots\cap k_{b}(m_{i_{j}})\right]\\ & = \sum_{j}(-1)^{{j-1}}\sum_{i_{1}<i_{2}<\cdots < i_{j } } \left[k_{b}(m_{i_{1}}\cap\cdots\cap m_{i_{j}})\right],\end{aligned}\ ] ] with the second equality following from proposition [ matroid - intersection ] .
clearly , we can restrict these sums to those @xmath385 for which @xmath386 , in which case @xmath387 .
thus , by proposition [ polar_relation ] , we have the relation @xmath388 = \sum_{j}(-1)^{{j-1}}\sum_{i_{1}<i_{2}<\cdots
< i_{j } } \left[i\left(k_{b}(m_{i_{1}}\cap\cdots\cap m_{i_{j}})\right)\right].\end{aligned}\ ] ] the assertion now follows from .
it turns out that all of the terms with @xmath389 in the summation of theorem [ valuation ] involve matroids which are _ disconnected_. this will allow us to deduce a corollary ( corollary [ sumresolution ] below ) which ignores these terms , and leaves a sum with _ positive _ coefficients . to this end
, recall that a nonempty subset @xmath390 is called a _
separator _ of @xmath6 if it leads to a direct sum decomposition of matroids : @xmath391 the whole ground set @xmath2 is itself a separator , and the collection of separators is closed under intersection . hence @xmath2 can be written as a disjoint union of inclusion - minimal separators of @xmath6 .
denote by @xmath392 the number of minimal separators of @xmath6 .
the following is ( * ? ? ?
* , proposition 4 ) .
the dimension of the matroid polytope @xmath219 is @xmath393 . considering @xmath64 as a graded @xmath53-algebra , its maximal ( homogeneous )
ideal is @xmath394 .
given an element @xmath395 , let @xmath396 denote its image in the quotient ring @xmath397 .
if @xmath398 and the dimension of @xmath219 is less than @xmath399 , then @xmath51 lies in the square @xmath400 of the maximal ideal @xmath401 . in other words , @xmath402 in @xmath397 .
if @xmath219 has dimension less than @xmath399 then @xmath403 , so there exists at least one proper separator @xmath404 .
since @xmath225 is an algebra morphism , one has @xmath405 , and hence @xmath51 lies in @xmath400 . since @xmath406 , to study decomposability of matroid polytopes @xmath219 , it is enough to restrict attention to _ connected _ matroids @xmath6 , that is , those with @xmath407 .
for these , the maximal cells in any decomposition @xmath408 will have dimension @xmath399 and so will also correspond to connected matroids .
all their proper intersections , however , will be lower - dimensional and so correspond to matroids with non - trivial separators .
[ sumresolution ] if a matroid polytope @xmath219 can be subdivided into finitely many matroid polytopes @xmath309 , then in @xmath397 one has @xmath409 this corollary interacts nicely with a result of hazewinkel ( * ? ? ?
* theorem 8.1 ) , confirming a conjecture of ditters which says that the @xmath53-algebra structure on @xmath64 is that of a _
free _ commutative algebra , that is , a polynomial algebra .
consequently , @xmath410 is a _ free _ ( graded ) @xmath53-module , and hence each homogeneous component @xmath411 is a @xmath53-lattice @xmath412 of some finite rank[multiblock footnote omitted ] @xmath413 .
thus for matroids @xmath6 of rank @xmath242 on ground set @xmath2 of size @xmath173 , to understand the potential matroid base polytope decompositions of @xmath219 , it helps to examine the _ additive semigroup _ structure generated by the elements @xmath414 within the lattice @xmath412 .
say that @xmath414 is _ decomposable _ if there exist matroids @xmath88 with @xmath415 , and indecomposable otherwise .
say that @xmath414 is _ weak image decomposable _ if it is decomposable with each @xmath88 a weak image of @xmath6 , that is , @xmath416 .
in other words , @xmath414 is indecomposable if and only if it is an element of the unique _ hilbert basis _
* chapter 13 ) for the additive semigroup generated by the @xmath414 .
corollary [ sumresolution ] implies that @xmath219 is indecomposable unless @xmath414 is weak image decomposable .
however , decomposability ( and hence also weak image decomposability ) of @xmath414 is easily checked using computer algebra packages that can compute the _ toric ideal _ and/or the hilbert basis for the additive semigroup generated by the @xmath414 within @xmath397 ; see ( * ? ? ? * chapters 4 and 13 ) .
[ rank2 ] a ( loopless ) rank @xmath355 matroid @xmath6 on @xmath173 elements is determined up to isomorphism by the partition @xmath417 of @xmath173 that gives the sizes @xmath418 of its parallelism classes .
also , @xmath419 is a weak image of @xmath420 , up to isomorphism , if and only if the partition @xmath421 is refined by the partition @xmath422 .
note that @xmath6 is connected if and only if @xmath423 has at least @xmath144 parts .
hence the connected weak images of @xmath6 correspond to all coarsenings of @xmath417 with at least @xmath144 parts .
in particular , if @xmath417 has _ exactly _
@xmath144 parts then @xmath414 must be weak image indecomposable and @xmath219 must be indecomposable .
in fact , by computer calculations , we have verified for @xmath424 that the rank @xmath355 matroids @xmath6 for which @xmath417 has exactly @xmath144 parts form the hilbert basis for the semigroup generated by the @xmath414 , and those for which @xmath417 has more than @xmath144 parts all have @xmath219 decomposable ( and hence @xmath414 weak image decomposable ) .
the following question was left open in an earlier version of this paper , but has recently been resolved in the affirmative by work of luoto ( * ? ? ?
* corollary 6.7 ) : fix @xmath173 , and consider the semigroup generated by @xmath414 within @xmath397 as one ranges over all matroids @xmath6 of rank @xmath355 on @xmath173 elements . is the hilbert basis for this semigroup indexed by those @xmath6 for which @xmath417 has exactly @xmath144 parts ?
we should mention that a convenient parametrization of all rank @xmath355 matroid base polytope decompositions was given by kapranov @xcite , who showed that all decompositions in this ( rank @xmath355 ) setting can be achieved by a sequence of hyperplane splits .
[ rank3 ] considering all 15 connected rank 3 matroids @xmath6 with @xmath425 ( see , for example , ( * ? ? ?
2 ) ) , we found five for which @xmath414 is indecomposable .
these are illustrated in figure [ extremes - figure ] .
in particular , the two matroids @xmath361 and @xmath362 in figure [ isotutte - figure](a ) satisfy @xmath426 , which can be written three different ways as sums of these indecomposables @xmath427 for @xmath362 , all three of these additive decompositions correspond to matroid base polytope decompositions of @xmath359 , as does the first for @xmath361 .
however , since @xmath428 is not a weak image of @xmath361 , the second and third can not correspond to such decompositions of @xmath358 .
does @xmath414 being weak image decomposable in @xmath397 imply that @xmath219 is decomposable ?
we see no reason , a priori , for this to hold , but the matroids considered in examples [ rank2 ] and [ rank3 ] provide no counterexamples .
in fact , for all of the matroids @xmath6 in those examples , one has @xmath219 indecomposable if and only if @xmath414 is indecomposable if and only if @xmath6 is minimally connected ( _ i.e. _ , all weak images of @xmath6 have a nontrivial separator ) .
the connected rank three matroids on 6 points for which @xmath219 is indecomposable , represented as affine point configurations in the plane .
the first matroid contains a tripled point , and the second contains two pairs of doubled points . ]
it is worth noting that among the rank @xmath144 matroids with @xmath429 elements , one finds the first matroid base polytope decompositions which are _ not _ hyperplane splits .
for example , if @xmath6 is the rank 3 matroid on @xmath430 having every triple but @xmath431 , @xmath432 and @xmath433 as bases , then both @xmath414 and @xmath219 split into three indecomposable pieces , each isomorphic to the matroid @xmath434 in figure [ extremes - figure ] .
this subdivision of @xmath219 can not be obtained via hyperplane splits .
one might ask how fine a matroid invariant is @xmath51 .
that is , how well does it distinguish non - isomorphic matroids , say in comparison with well - studied matroid invariants like the tutte polynomial ? certainly the kernel of the hopf algebra map @xmath225 contains @xmath435-[m_{{\mathrm{loop}}}]$ ] by example [ tiny - examples ] , and hence contains the smallest hopf ideal @xmath436 generated by @xmath437 .
in fact , since @xmath437 is primitive ( as it is of degree @xmath35 ) , the hopf ideal @xmath436 which it generates coincides with the principal ideal consisting of all multiples of @xmath437 .
consequently @xmath81 factors through the quotient @xmath438 , that is , through the hopf algebra of matroids modulo `` loops = coloops '' . beyond this inability to distinguish loops from coloops
, one might ask how discriminating @xmath51 is .
the next two examples show that it certainly does nt distinguish all loopless and coloopless matroids up to isomorphism ( which would have been too much to ask ) , but it at least does better than the well - known tutte polynomial in some instances .
( a ) two matroids @xmath57 , represented by affine point configurations having the same tutte polynomials and the same quasisymmetric functions .
( b ) two matroids @xmath439 having the same tutte polynomials but different quasisymmetric functions . ]
[ six - element - example ] figure [ isotutte - figure](a ) depicts two matroids , represented as affine point configurations , having the same tutte polynomial ( because @xmath440 and @xmath441 , where @xmath18 is the labelled point in each case ) .
direct computer calculation ( using theorem [ poset - expansion ] ) shows that @xmath442 .
these two examples were borrowed from brylawski and oxley s survey on the tutte polynomial @xcite ; they are the smallest examples of non - isomorphic matroids with the same tutte polynomial .
[ seven - element - example ] figure [ isotutte - figure](b ) depicts two matroids @xmath439 having the same tutte polynomials ( since @xmath443 and @xmath444 ) , but different quasisymmetric functions : it turns out that the coefficient of @xmath445 in @xmath446 is @xmath447 , while in @xmath448 is @xmath449 .
these two examples were again taken from brylawski and oxley s survey @xcite , where they point out other features that @xmath439 share and do not share .
note that example [ seven - element - example ] rules out the possibility of computing @xmath51 purely in terms of @xmath450 . even though there is no direct deletion - contraction computation for @xmath51 for which one might have naively hoped , does this rule out other sorts of recursions ?
in particular , one is tempted to try the following .
theorem [ poset - expansion ] says @xmath451 can one better identify the two summands in this last equation ? are they instances of some quasisymmetric functions that should be associated to objects more general than matroids ?
lastly , we mention an invariant @xmath452 for a matroid @xmath6 ( representable over @xmath453 ) recently introduced by speyer @xcite , which shares some common features with @xmath51 . among other of its properties ,
this invariant @xmath452 is 1 . a polynomial in one variable t with integer coefficients ( conjecturally nonnegative ) , 2 .
multiplicative under direct sums : @xmath454 , 3 .
invariant under duality of matroids : @xmath455 , 4 .
additive under any decomposition of the matroid base polytope @xmath456 , where @xmath457 are all of the _ interior _ faces of the decomposition , in the sense that @xmath458 .
is @xmath452 related to ( some specialization of ) @xmath51 ? in personal communication , speyer has pointed out that all three invariants of a matroid @xmath6 discussed in this section , behave either _ valuatively _ or _ additively _ under matroid base polytope decompositions : 1 .
the quasisymmetric function @xmath459 behaves valuatively according to theorem [ valuation ] , 2 .
speyer has checked that the tutte polynomial @xmath460 also behaves valuatively ( via a small calculation using the _ corank - nullity _
formula for @xmath460 ; see also @xcite ) , and 3 .
his invariant @xmath452 behaves additively by property ( iv ) above .
speyer then used this to explain why all three invariants take the same value for the two matroids @xmath57 shown in figure [ isotutte - figure](a ) : either of the matroid base polytopes @xmath309 for @xmath89 can be obtained from the hypersimplex @xmath461 associated to the uniform matroid of rank @xmath144 on the same six elements , by splitting off with hyperplanes ( in any order ) two other polytopes @xmath462 , that is , @xmath463 furthermore , the @xmath464 are _ all isomorphic _ as matroids @xmath465 and have _ isomorphic intersections _ @xmath466 as a consequence , a matroid invariant @xmath467 will have @xmath468 for @xmath89 . in either case , this forces @xmath469 .
this strongly suggests trying to define a `` universal '' valuative invariant of matroids , following mcmullen s _ polytope algebra _ , and in particular his section @xcite dealing with valuations invariant under a finite group action .
build an abelian group starting with the free abelian group on basis elements @xmath54 $ ] indexed by matroids @xmath6 , imposing the valuation relation for each matroid base polytope decomposition of @xmath219 , and the relation @xmath54=[m']$ ] if @xmath6 and @xmath470 are isomorphic as matroids are required to be @xmath221-vectors whose coordinates sum to the rank @xmath471 . ] .
valuative matroid invariants are exactly the linear functionals on this abelian group .
study the structure of this abelian group .
are there special classes of special matroids which generate it ?
for example , a conjecture of speyer ( * ? ? ?
* conjecture 11.3 ) would follow if this abelian group were generated by the classes @xmath54 $ ] where @xmath6 runs over all direct sums of _ series - parallel matroids_.
it turns out that the proofs of proposition [ quasisymmetry ] , theorem [ poset - expansion ] , corollary [ last - coefficient ] , theorem [ reciprocity - theorem ] , and theorem [ valuation ] generalize in a straightforward way to give results about a general class of convex polytopes studied recently by postnikov @xcite ; see also @xcite and @xcite .
given a convex polytope @xmath472 in @xmath319 , the following conditions are well - known to be equivalent ( * ? ? ?
* proposition 7.12 ) : 1 .
every edge of @xmath472 lies in one of the directions @xmath473 .
the normal fan of @xmath472 in @xmath474 is refined by the usual _ braid arrangement _ ( or _ type @xmath475 weyl chamber fan _ ) .
the polytope @xmath472 is a minkowski summand of some realization of the _ permutohedron _ as a minkowski sum of line segments ( possibly of different lengths ) in the directions @xmath476 .
say that @xmath472 is a _
generalized @xmath173-permutohedron _ when any of these equivalent conditions hold .
[ matroid - base - polytope - example ] given a matroid @xmath6 on ground set @xmath477 $ ] , the matroid base polytope @xmath219 defined in section [ decomposition - section ] is a generalized @xmath173-permutohedron ( * ? ? ?
* , theorem 1 ) , a fact that played a crucial role in the proof of theorem [ poset - expansion ] . given a polytope @xmath472 in @xmath319 , say that a function @xmath478 \rightarrow { { \mathbb p}}$ ] ( which we think of as giving an element of @xmath474 ) is _
@xmath472-generic _ if @xmath7 maximizes over @xmath472 uniquely at a vertex . in other words , @xmath7 lies in the _ interior _ of an @xmath173-dimensional cone in the normal fan for @xmath472 .
one can then prove the following : [ permutohedron - summand - theorem ] if @xmath472 is a generalized @xmath173-permutohedron in @xmath319 , then 1 . the power series @xmath479 \rightarrow { { \mathbb p } } } } { { \mathbf x}}_f\ ] ] is quasisymmetric , with 2 .
an expansion in terms of @xmath0-partitions enumerators as @xmath480 where @xmath481 are certain strictly labelled posets indexed by the vertices of @xmath472 .
3 . furthermore ,
the coefficients @xmath482 in its expansion @xmath483 1 .
are nonnegative , 2 .
sum to @xmath232 , and 3 .
have @xmath484 equal to the number the number of vertices of @xmath472 .
4 . the antipode @xmath293 on @xmath64 satisfies @xmath485 where @xmath486 \rightarrow { { \mathbb p } } } |\ { f\textrm{-minimizing vertices of } q\}|.\ ] ] 5 . the two polynomials @xmath487 in the variable @xmath43 defined by specializing @xmath488 to @xmath285 satisfy @xmath489 6 . suppose @xmath490 is a decomposition of @xmath472 into finitely many permutohedron summands @xmath491 , in which @xmath492 is a common face of @xmath491 and @xmath493 for all @xmath494 .
then @xmath495 where the sum is over those terms in which @xmath496 is nonempty . in fact
, the posets @xmath497 appearing in the theorem have a very simple description : @xmath497 is the transitive closure of the binary relation on @xmath177 $ ] which has @xmath498 if there exists an edge of @xmath472 of the form @xmath499 with @xmath500 . in the remainder of this section , we discuss three naturally occurring families of generalized @xmath173-permutohedra that have occurred in the literature .
[ gen - perm - qsym ] study the quasisymmetric functions @xmath501 associated with any of these families of generalized permutohedra @xmath472 .
+ let @xmath502 be a simple graph on vertex set @xmath177 $ ] .
let @xmath503 denote the minkowski sum of line segments in the directions @xmath504 then @xmath503 is a generalized @xmath173-permutohedron ; the @xmath173-permutohedron itself equals @xmath505 where @xmath506 is the complete graph on @xmath173 vertices .
it is easy to see that a function @xmath478 \rightarrow { { \mathbb p}}$ ] is @xmath503-generic if and only if it is a _ proper coloring _ of the vertex set @xmath177 $ ] of @xmath502 .
one concludes that @xmath507 is the same as the _ chromatic symmetric function _
@xmath508 introduced by stanley @xcite , and studied further by others in recent years .
many of the results of this paper were inspired by his work , and in particular theorem [ permutohedron - summand - theorem ] generalizes a few of the facts about @xmath509 .
it is also known ( see ( * ? ? ?
* example 4.5 ) ) that the map @xmath510 can be interpreted as a hopf morphism between a certain hopf algebra of graphs and the hopf algebra @xmath511 of _ symmetric functions _ inside the quasisymmetric functions @xmath64 .
as far as we know , this morphism is of a different nature than our hopf morphism @xmath225 .
+ example [ matroid - base - polytope - example ] alludes to a famous resut of gelfand , goresky , macpherson , and serganova , characterizing matroids in terms of their matroid base polytopes , which we rephrase slightly here .
( see ( * ? ? ?
* , theorem 1 ) , ( * ? ? ? * theorem 1.11.1 ) ) let @xmath512 be a collection of @xmath242-subsets of @xmath177 $ ] , and @xmath472 the convex hull of their characteristic vectors in @xmath513
. then @xmath512 is the collection of bases @xmath149 for some matroid @xmath6 on ground set @xmath477 $ ] ( and @xmath514 is the associated matroid base polytope ) if and only if @xmath472 is a generalized @xmath173-permutohedron .
this led gelfand , goresky , macpherson , and serganova to the notion of _ coxeter matroids _
a coxeter matroid is the result of taking the characterization in the previous theorem and 1 . replacing @xmath242-subsets of @xmath177 $ ] , which can be thought of as the cosets of maximal parabolic subgroups in the coxeter group of type @xmath475 , with cosets of an arbitrary parabolic subgroup in an arbitrary finite coxeter group , 2 . replacing the characteristic vectors of @xmath242-subsets with @xmath515-translates of sums of fundamental dominant weights , 3 .
replacing generalized @xmath173-permutohedra with minkowski summands of the zonotopes generated by other root systems .
when the coxeter group is of type @xmath475 , considering arbitrary parabolic subgroups instead of just maximal ones leads to the notion of a _ flag matroid _ , and its _ flag matroid base polytope_. these will be generalized @xmath173-permutohedra generalizing the matroid base polytopes , whose vertices are vectors in @xmath516 that no longer necessarily sum to @xmath242 , but obey certain constraints on the sizes of their coordinates ; see @xcite .
generalizing in another direction , a _ discrete polymatroid base polytope of rank _
@xmath242 ( see @xcite ) is a generalized @xmath173-permutohedron , each of whose vertices has nonnegative integer coordinates summing to @xmath242 .
these polytopes were introduced by edmonds @xcite in the context of combinatorial optimization .
+ building on work of others ( de concini - procesi , davis - januszkiewicz - scott , and carr - devadoss ) , postnikov @xcite showed that the generalized @xmath173-permutohedra contain an interesting subclass of polytopes called _ graph - associahedra _ , indexed by simple graphs @xmath502 on vertex set @xmath177 $ ] . within this subclass ,
the associahedra and cyclohedra correspond to the cases where the graphs @xmath502 are paths and cycles , respectively .
we observe here that the morphism @xmath81 is definitely _ not _ surjective without extending scalars .
the image of the map @xmath518 on homogeneous components of degree @xmath355 is a sublattice of index @xmath355 within @xmath519 : there are only four non - isomorphic matroids on @xmath355 elements , whose images under @xmath81 are all either of the form @xmath520 or @xmath521
. our approach will be to define , for each degree @xmath173 , a family of @xmath522 matroids on ground set @xmath477 $ ] , whose images under @xmath81 span @xmath523 with rational coefficients .
it turns out that it will suffice to take a subfamily of a family of @xmath524 matroids which were called _
freedom matroids _ in @xcite , and which we will call _ @xmath525-matroids _ here .
they were considered in the context of face enumeration in @xcite and in @xcite , where they arose in the context of combinatorial operators on zonotopes . given a matroid @xmath6 ,
let @xmath526 be a single - element extension of @xmath6 by an isthmus .
let @xmath527 be a single - element extension of @xmath6 which is the _ principal extension of @xmath6 along the improper flat _ , that is , one adjoins a new element @xmath18 to the ground set , which is generic while obeying the constraint that it does not increase the rank .
say that @xmath6 is a _
@xmath525-matroid _ if it can be obtained from the empty matroid @xmath528 on @xmath529 by performing a sequence of repeated @xmath530 and/or @xmath531 operations .
it happens that every matroid with @xmath532 is isomorphic to a @xmath525-matroid .
let @xmath533 denote the collection of all binary strings @xmath534 that begin with a @xmath32 .
given @xmath535 in @xmath533 , let @xmath536 be the @xmath525-matroid built from this sequence beginning with an empty matroid , where one performs the @xmath436 operation for each @xmath32 and the @xmath0 operation for each @xmath35 in @xmath535 .
for example , the sequence @xmath537 would build the @xmath525-matroid @xmath538 of rank @xmath35 consisting of @xmath539 parallel elements ) .
the operation @xmath530 which adds an isthmus to @xmath6 has a predictable effect on @xmath51 : @xmath542 seeing this , one might hope to approach theorem [ refined - surjectivity - thm ] by understanding how @xmath543 relates to @xmath51 .
unfortunately , @xmath543 does not depend solely on @xmath51 via some operation in @xmath64 .
for example , the two matroids @xmath544 have @xmath545 , however @xmath546 _ step 1 .
_ introduce a family of posets @xmath547 on @xmath177 $ ] , also indexed by @xmath533 , and show that the expansion of the @xmath548 in terms of the _ strictly labelled _ @xmath0-partition enumerators for the @xmath547 is triangular in some ordering .
furthermore , the diagonal coefficients in this expansion are products of binomial coefficients that all divide @xmath232 . _ step 3 . _
show that the expansion of the _ naturally labelled _
@xmath0-partition enumerators for the @xmath547 in terms of the @xmath0-partition enumerators for the @xmath549 is unitriangular with respect to some ordering . from step 2
it then follows that the former @xmath0-partition enumerators also give a @xmath53-basis for @xmath64 .
step 1 is completed in the remainder of this subsection , while steps 2 and 3 are achieved in subsections [ q - basis - section ] and [ r - q - expansion - section ] .
as mentioned in an earlier footnote , luoto @xcite has recently found an alternative to steps 2 and 3 , by expanding the @xmath547 basis elements unitriangularly in terms of his `` matroid - friendly '' basis for @xmath64 .
each such @xmath535 also defines a partition of the set @xmath177 $ ] into intervals that we will call the _ blocks _ @xmath553 of @xmath535 , by breaking @xmath177 $ ] between the positions @xmath554 where @xmath555 .
we also define a vector @xmath556 associated to @xmath535 as follows : @xmath557 is the number of positions @xmath264 in the block @xmath558 for which @xmath559 .
it is not hard to see that one can recover @xmath535 uniquely from the blocks @xmath560 and the values @xmath556 .
let @xmath561 and let @xmath535 be the string in @xmath533 given by @xmath562 then @xmath547 is the labelled poset on @xmath563 $ ] in which the minimal elements are @xmath564 , the maximal elements are @xmath565 ( and @xmath566 ) , and the order relations are @xmath567 as illustrated in figure [ r - poset - figure ] .
[ shifted - characterization ] a labelled poset @xmath0 on @xmath177 $ ] is isomorphic to @xmath547 for some @xmath535 if and only if it has height at most one , and can be relabelled so that each minimal ( resp .
element has its upward ( resp .
downward ) neighbors in @xmath0 forming a final ( resp .
initial ) segment of @xmath177 $ ] .
[ pi - triangularity ] the lexicographic order @xmath572 on @xmath533 induced by @xmath573 makes the expansion of @xmath574 in terms of the strict @xmath0-partition enumerators @xmath575 triangular of the following form : @xmath576 where @xmath577 , and @xmath578 is any strict labelling of the poset @xmath579 .
furthermore , the diagonal coefficient @xmath580 can be expressed in terms of the blocks @xmath581 and vector @xmath556 associated to @xmath535 as follows : @xmath582 we use theorem [ poset - expansion ] and expand @xmath583 the bases @xmath9 of @xmath536 are easily analyzed in terms of the blocks @xmath581 and vector @xmath556 associated to @xmath535 ( cf .
* proposition 5.1 ) ) .
note that @xmath536 will have rank @xmath584 , and it has a distinguished chain of flats @xmath585\ ] ] in which @xmath586 .
bases @xmath9 of @xmath536 are then simply the @xmath242-subsets @xmath9 of @xmath177 $ ] that contain for each @xmath587 at most @xmath588 elements from the flat @xmath589 . given any base @xmath9 of @xmath536
, we claim that the poset @xmath198 is isomorphic to some @xmath579 .
to see this , we use proposition [ shifted - characterization ] . we know that @xmath198 has height at most @xmath35 . relabel its minimal ( resp .
maximal ) elements , that is , those in @xmath9 ( resp .
@xmath590 ) by an initial ( resp .
final ) segment of @xmath177 $ ] , with those lying in block @xmath558 coming earlier than those in block @xmath591 whenever @xmath592 .
it is then easy to check that any minimal ( resp .
maximal ) element of @xmath198 will have its upward ( resp .
downward ) neighbors in @xmath198 forming a final ( resp .
initial ) segment of @xmath177 $ ] .
the diagonal terms on the right side of come from bases @xmath9 of @xmath536 containing _ exactly _
@xmath557 elements of @xmath593 for each @xmath108 ; let us call these the _ diagonal _ bases of @xmath536 . for example , the lexicographically earliest base @xmath594 for @xmath536 is a diagonal base , and it is not hard to see that @xmath595 on the nose ; see figure [ diagonal - bases - figure ] for an example .
there are a total of @xmath596 diagonal bases @xmath9 for @xmath536 , and each has @xmath597 . for any non - diagonal base @xmath9
, there is some smallest index @xmath108 such that @xmath9 contains less than @xmath557 elements of @xmath593 .
it is not hard to see that such a @xmath9 will have @xmath598 for some @xmath599 that agrees with @xmath535 in the first @xmath600 positions , that is , in the positions indexed by their first @xmath601 blocks of @xmath535 ( or @xmath599 ) .
but then the @xmath306 block @xmath558 for @xmath599 indexes a @xmath221-substring of @xmath599 of the form @xmath602 starting with more zeroes than does the corresponding @xmath306 block @xmath558 for @xmath535 , so that @xmath603 . figure [ diagonal - bases - figure ] illustrates the previous proof . here
the matroid @xmath536 is drawn as an affine point configuration .
its associated chain of flats is @xmath605 the lexicographically first base @xmath606 of @xmath536 is a diagonal base , having poset @xmath607 which coincides with @xmath547 .
an example of a non - diagonal base @xmath608 is shown , with poset @xmath198 isomorphic to @xmath579 where where @xmath609 . here
the smallest index @xmath108 for which @xmath9 does not contain @xmath557 elements of @xmath610 is @xmath611 , and hence @xmath612 agree in their first @xmath613 positions
. however the second block @xmath614 in @xmath599 indexes a substring @xmath615 starting with two zeroes , while the second block @xmath616 in @xmath535 indexes a substring @xmath617 starting with only one zero .
hence @xmath603 . in this subsection , we complete step 2 of the proof of theorem [ refined - surjectivity - thm ] by exhibiting a new @xmath53-basis for @xmath64 that may be of independent interest .
this basis turns out to have a nice expansion property ( lemma [ positive - triangular - expansion - property ] ) when one multiplies one of its elements by @xmath620 .
this new basis comes from a family of ( non - naturally , non - strictly ) labelled posets @xmath549 on @xmath177 $ ] , indexed by @xmath535 in @xmath533 , which are defined recursively . before defining them
, we recall some standard labelled poset terminology .
let @xmath621 be labelled posets on label sets @xmath622 that disjointly decompose @xmath177 $ ] , that is , @xmath177=a_1
\sqcup a_2 $ ] . their _ disjoint sum _
@xmath623 is the labelled poset on label set @xmath177 $ ] keeping all order relations that were present in @xmath624 or in @xmath625 , with no new order relations between @xmath624 and @xmath625 .
their _ ordinal sum _
@xmath626 is obtained by from the disjoint sum by imposing further new order relations : @xmath627 for all @xmath628 . 1 . @xmath629 is the labelled poset on @xmath177 $ ] which is an antichain .
if @xmath535 ends with a @xmath35 , say @xmath630 , then @xmath631 where @xmath632 is a labelled poset with one element labelled @xmath633 .
if @xmath535 ends with a @xmath32 ( but is not _ all _ zeroes ) , say @xmath634 , then @xmath549 is obtained from @xmath635 by adding in a new element labelled @xmath633 , with only one new order relation @xmath636 ( plus all others generated by transitivity ) , and then swapping the labels of @xmath637 .
it is not hard to see that @xmath549 is always isomorphic to an iterated ordinal sum of a sequence of antichains .
for example , in the poset @xmath549 of example [ q - poset - example ] , these antichains are the induced subposets on these sets : @xmath642 for each positive integer @xmath43 , define an operation @xmath643 that takes labelled posets on @xmath177 $ ] to labelled posets on @xmath644 $ ] as follows .
given a labelled poset @xmath0 on @xmath173 , then @xmath645 where @xmath646 is a labelled poset with one element labelled @xmath647 , and @xmath59 is an @xmath648-element antichain with elements labelled @xmath649 . to describe @xmath549 in terms of these operations , uniquely decompose @xmath535 into an initial sequence of @xmath650 zeroes , and
then sequences of length @xmath651 of the form @xmath652 then @xmath653 given @xmath656 , let @xmath657 be the linear extension of the labelled poset @xmath549 obtained by reading each of the antichains discussed above in the _ reverse _ of their usual numerical order .
e.g. one has @xmath658 in the previous example , where we have indicated the positions of descents in @xmath657 by dots . 1 .
the descent set of @xmath657 can be read from @xmath535 as follows : @xmath659:\sigma_{i+1 } = 0\},\ ] ] and 2 .
every other linear extension @xmath194 in @xmath660 has @xmath661 because at least one of the antichains discussed above must not appear in reverse order in @xmath194 .
+ it turns out that the @xmath663 basis for @xmath64 has an interesting expansion property when one multiplies by @xmath664 .
the expansion is both nonnegative , and triangular in a certain sense ; see lemma [ positive - triangular - expansion - property ] below .
+ let @xmath0 be a labelled poset on @xmath173 integers @xmath665 .
then the _ standardization _
@xmath666 of @xmath0 is the labelled poset on @xmath177 $ ] obtained by replacing the label @xmath667 in @xmath0 with the integer @xmath108 for @xmath668 .
let @xmath0 be a labelled poset on @xmath177 $ ] which is an ordinal sum @xmath672 in which @xmath673 is the labelled poset with one element labelled @xmath173 , and @xmath624 have @xmath674 elements respectively ( so that @xmath675 ) .
let @xmath676 be the following labelled poset on @xmath177 $ ] .
first form the labelled poset @xmath677 on @xmath678 $ ] obtained from @xmath679 by adding @xmath680 to all of its labels .
define @xmath681 then @xmath682 [ psi - map - prop ] the @xmath53-linear map @xmath683 defined by sending @xmath684 for any permutation @xmath194 will also send @xmath685 and more generally , for any labelled poset @xmath0 on @xmath177 $ ] , sends @xmath686 in this case , @xmath173 is a top element of @xmath549 by construction , and we decompose the linear extensions @xmath194 in @xmath690 into three sets , based on the location of @xmath633 relative to @xmath173 : 1 . those @xmath194 with @xmath633 occurring second - to - last , just before @xmath173 .
2 . those @xmath194 with @xmath633 occurring last ,
just after @xmath173 .
3 . those @xmath194 remaining , in which @xmath633 occurs at least two positions before @xmath173 . letting @xmath693
denote the transposition that swaps the labels @xmath637 in a labelled poset on @xmath694 $ ] , a little thought shows @xmath695 also , if one applies @xmath696 to a linear extension in which @xmath637 are not adjacent , there is no effect on the descent set .
since this is true for every linear extension in @xmath697 , one knows that @xmath698 has the same distribution of descent sets as @xmath699 . 1 . those @xmath194 with @xmath633 at least two positions after @xmath173 .
2 . those @xmath194 with @xmath633
immediately after @xmath173 .
3 . those @xmath194 with @xmath633 immediately preceding @xmath173 .
4 . those @xmath194 with @xmath633 at least two positions before @xmath173 .
given @xmath535 in @xmath714 , one has @xmath720 where the third equality uses induction , and the last equality uses lemma [ positive - triangular - expansion - property ] . note that the last equality implies assertion ( ii ) . given @xmath535 in @xmath717 and @xmath718
, one has @xmath721 by induction using assertions ( i),(ii ) ( that is , theorem [ r - q - expansion ] ) .
applying @xmath643 to this equality gives @xmath722 where the last equality uses proposition [ psi - map - prop ] .
this gives assertion ( iii ) .
.1 in * proof of ( i ) .
* given @xmath535 in @xmath714 , let @xmath723 : \sigma_j=1\},\ ] ] so that the labelled poset @xmath724 has the element labelled @xmath173 above _ all _ of the elements in @xmath177-j$ ] , and above _ none _ of the element in @xmath725 .
this means that for every linear extension @xmath194 in @xmath726 , there is a unique subset @xmath727 consisting of those elements appearing later ( _ i.e. _ higher ) in @xmath194 than @xmath173 .
a little thought shows that this gives a decomposition @xmath728 where @xmath729 is a labelled poset on @xmath177 $ ] having @xmath730 the restriction of @xmath547 to its elements labelled by @xmath281-i$ ] , and @xmath731 an antichain labelled by the elements of @xmath436 .
consequently , @xmath732 where for each @xmath727 , the string @xmath733 is obtained from the string @xmath535 by removing all the ones that were in the positions indexed by @xmath436 . hence by induction using assertion ( iii ) one obtains @xmath734 with @xmath735 in @xmath713 .
note that for any @xmath727 one has @xmath736 and equality occurs if and only if @xmath737 .
hence assertion ( i ) follows from .
note that the @xmath710 basis for @xmath64 consists entirely of _ naturally labelled _ @xmath0-partition enumerators .
this answers affirmatively the question of whether @xmath64 is @xmath53-linearly spanned by naturally labelled @xmath0-partition enumerators ; note that neither of the usual @xmath53-bases for @xmath64 ( the @xmath158 or @xmath240 ) have this form . the same question was also answered ( affirmatively ) in recent work of stanley @xcite who , after being queried by the authors of the current paper , produced yet another @xmath53-basis for @xmath64 consisting of naturally labelled @xmath0-partition enumerators . given a composition @xmath28 of @xmath173 , he defined @xmath738 to be the naturally labelled poset which is the ordinal sum @xmath739 , in which @xmath558 is an antichain on @xmath740 elements for each @xmath115 .
these posets @xmath738 bear a close resemblance to the ( non - naturally ) labelled posets @xmath549 defined above , in that both have simple , unitriangular expansions of their @xmath0-partition enumerators in terms of the @xmath240-basis . in @xcite ,
stanley combinatorially interprets this upper unitriangular change - of - basis matrix between his basis and the @xmath240-basis , as well as providing a nice ( and remarkably similar ) combinatorial interpretation for the inverse change - of - basis matrix .
order the strings @xmath535 in @xmath533 with lex order , and order the compositions @xmath156 of @xmath173 also in lex order
. then the matrix @xmath743 expanding @xmath710 in terms of @xmath687 will be upper unitriangular ( by theorem [ r - q - expansion ] ) , while the matrix @xmath744 expanding @xmath687 in terms of @xmath240 will be lower unitriangular ( by the proof of proposition [ q - basis - prop ] ) . and @xmath745 .
for example , when @xmath746 , this looks like @xmath747 \quad = \ ] ] @xmath748 \left [ \begin{matrix } & r_{000 } & r_{001 } & r_{010 } & r_{011 } \\ q_{000 } & 1 & 0 & 0 & 0 \\ q_{001 } & & 1 & 1 & 0 \\ q_{010 } & & & 1 & 1 \\
q_{111 } & & & & 1 \end{matrix } \right]\ ] ] we have now encountered five @xmath53-bases for the hopf algebra @xmath64 of quasisymmetric functions , namely @xmath749 given any such basis @xmath750 , one might ask whether the structure constants @xmath751 from the unique expansion @xmath752 are always nonnegative . for the monomial basis @xmath158 and the fundamental bases @xmath240
, this property is well - known to hold and is straightforward .
unfortunately , this property fails for the remaining three bases @xmath753 .
they turn out to have some negative multiplication structure constants occurring already in ( relatively ) low degrees : @xmath754 l.j .
billera , r. ehrenborg and m. readdy , the @xmath755-index of zonotopes and arrangements , _ mathematical essays in honor of gian - carlo rota _ , b.e . sagan and r.p .
stanley , eds . , birkhuser , boston , 1998 , pp .
2340 .
j. edmonds , submodular functions , matroids , and certain polyhedra , _ combinatorial structures and their applications ( proc .
calgary internat .
conf . , calgary , alta . , 1969 ) _ , 6987 , gordon and breach , new york , 1970 ( reprinted in _
combinatorial optimization
eureka , you shrink ! _
, 1126 , lecture notes in comput .
, 2570 , springer , berlin , 2003 ) .
i. gessel , multipartite @xmath0-partitions and inner products of skew schur functions .
combinatorics and algebra ( boulder , colo . , 1983 ) , 289317 , _ contemp .
_ * 34 * , amer . math
. soc . , providence , ri , 1984 .
stanley , the descent set and connectivity set of a permutation .
_ j. integer sequences _ * 8 * , ( 2005 ) , article 05.3.8 .
b. sturmfels , grbner bases and convex polytopes .
_ university lecture series _ * 8 * , american mathematical society , providence , ri , 1996 . | a new isomorphism invariant of matroids is introduced , in the form of a quasisymmetric function .
this invariant 1 . defines a hopf morphism from the hopf algebra of matroids to the quasisymmetric functions , which is surjective if one uses rational coefficients , 2 .
is a multivariate generating function for integer weight vectors that give minimum total weight to a unique base of the matroid , 3 .
is equivalent , via the hopf antipode , to a generating function for integer weight vectors which keeps track of how many bases minimize the total weight , 4
. behaves simply under matroid duality , 5 .
has a simple expansion in terms of @xmath0-partition enumerators , and 6 .
is a valuation on decompositions of matroid base polytopes .
this last property leads to an interesting application : it can sometimes be used to prove that a matroid base polytope has no decompositions into smaller matroid base polytopes .
existence of such decompositions is a subtle issue arising in work of lafforgue , where lack of such a decomposition implies the matroid has only a finite number of realizations up to scalings of vectors and overall change - of - basis . . |
the concept of @xmath0-ray strength functions was introduced in the fundamental work of blatt and weisskopf @xcite .
they showed that the square of the @xmath0-transition matrix element connecting highly excited states is proportional to the level spacing @xmath7 of the initial states @xmath8 with equal spin and parity .
therefore , the ratio of the partial radiative width @xmath9 of the states @xmath8 ( which is connected to @xmath0 transitions with transition energy @xmath10 and populating some low lying levels ) and @xmath7 was suggested to be important for the description of @xmath0 transitions in the continuum . the corresponding model - independent definition of the @xmath0-ray strength function is given by @xmath11 , where @xmath12 is the multipolarity of the @xmath0 transition and x refers to the electric or magnetic character of the @xmath0 transition .
the @xmath0-ray strength function is now considered as a measure for the average electromagnetic properties of nuclei and is fundamental for the understanding of nuclear structure and reactions involving @xmath0 rays .
experimentally , the main information on the @xmath0-ray strength function has been obtained from the study of photoabsorption cross - sections @xcite .
it is commonly adopted that the e1 strength function is determined by the properties of the giant electric dipole resonance ( gedr ) around its resonance energy , typically @xmath13
1020 mev
. however , serious lack of information persists at lower @xmath0-ray energies .
it was assumed that the tail of the lorentzian describing the gedr determines the e1 strength function at these energies .
the only experimental data on the e1 @xmath0-ray strength function between compound states with @xmath0-ray energies below 2 mev have been obtained using the @xmath14nd(n,@xmath15 reaction @xcite .
these data show that the extrapolation of the gedr to low energies fails to describe the experimental values of the e1 strength function and indicates a finite value of @xmath16 in the limit @xmath17 . as a result
, a model for the e1 strength function was developed by kadmenski , markushev and furman ( kmf ) @xcite which takes into account the energy and temperature dependence of the gedr width .
today , this model and its empirical modifications @xcite are frequently used in the description of experimental data , but at the same time , it needs additional experimental verification .
the e1 strength function is not solely governing the @xmath0-ray emission for lower @xmath0-ray energies .
other multipolarities , and especially the m1 strength function , play important roles as well .
the experimental information on the @xmath0-ray strength of m1 transitions is more scarce .
it is commonly assumed that the m1 strength is well described by the weisskopf model @xcite , where the dipole @xmath0-ray strength function is independent of the @xmath0 energy .
but some experiments indicate the existence of an m1 giant resonance originating from spin - flip excitations in the nucleus @xcite .
also , the analysis of @xmath0-ray spectra from ( n,@xmath6 reactions @xcite indicates that the use of the m1 giant dipole resonance model gives a better fit to the experimental data than the weisskopf model . special attention
@xcite has been devoted to the anomalous bump found in the @xmath0-ray spectra of the ( n,@xmath6 and ( d , p@xmath6 reactions at low energies .
the same bump has probably also been observed in the ( @xmath3he,@xmath4 ) reaction @xcite .
a previous work @xcite shows that the energy of the @xmath0-ray bump increases with neutron number in the @xmath18 82126 region .
the bump is called the pygmy resonance due to the considerably lower strength compared to the gedr .
the pygmy resonance has first been explained by the enhancement of the e1 strength function @xcite .
however , one can not rule out a possible m1 character connected to orbital m1 strength ( scissors mode ) in nuclei , which was first observed in electron scattering experiments @xcite .
the @xmath0-ray strength function is difficult to measure in the @xmath0-decay between highly excited states , since the decay rate also depends on the number of accessible levels .
the analysis of @xmath0-ray strength functions from the spectra of ( n,@xmath6 reactions @xcite shows that the results depend crucially on the level density model employed .
therefore , conclusions based on a certain level density formula should be considered as preliminary and thus , need confirmation .
recently @xcite , a new experimental technique has been developed , based on a set of primary @xmath0-ray spectra measured at consecutive excitation energies @xmath19 in light - ion reactions with one charged ejectile .
the technique allows to disentangle the @xmath0-ray spectra into a @xmath0-energy dependent function @xmath20 ( which , as will be shown below , can be uniquely connected to the @xmath0-ray strength function ) and the level density @xmath21 .
it makes it possible to study the @xmath0-ray strength function and level density independently from each other , in contrast to what can be done by using radiative neutron capture techniques . in previous works
@xcite the extracted level densities @xmath22 have been utilized to deduce thermodynamical properties for several rare earth nuclei . in this paper , however , we will focus on the @xmath20 function .
the experimental method is described in sect .
ii . in sect .
iii we give a short outline of the models used to describe the experimental data , and in sect .
iv we compare various predictions to the experimental data .
our results are also compared to data from the ( n,@xmath6 reaction performed by others .
conclusions are given in sect .
the experiments were carried out with 45 mev @xmath3he - projectiles at the oslo cyclotron laboratory ( ocl ) .
the particle-@xmath0 coincidences are measured with the cactus multidetector array @xcite using the ( @xmath3he,@xmath4 ) reaction on @xmath23dy and @xmath24yb self - supporting targets .
the charged ejectiles were detected with eight particle telescopes placed at an angle of 45@xmath25 relative to the beam direction .
an array of 28 nai @xmath0-ray detectors with a total efficiency of @xmath2615% of @xmath27 surrounded the target and particle detectors . the experimental extraction procedure and
the assumptions made are described in refs .
@xcite and references therein . from the @xmath28-@xmath0-coincidences ,
spectra of the total @xmath0-ray cascade can be sorted out according to the initial excitation energy @xmath19 .
these spectra are the basis for making the first generation ( or primary ) @xmath0-ray matrix @xmath29 , which is factorized according to the brink - axel hypothesis @xcite @xmath30 here , @xmath31 and @xmath22 are the @xmath0-ray energy dependent factor and the level density , respectively .
it is now possible to determine @xmath31 and @xmath22 by an iterative procedure . the first trial function for @xmath22 is simply taken as a straight line and the corresponding @xmath31 is determined by eq .
( [ eq : matrix ] ) .
then , a @xmath32 minimum is calculated for each data point of @xmath31 and @xmath22 , keeping the others fixed .
this procedure is repeated about 50 times , until a global least square fit to the @xmath261400 data points of the @xmath29 matrix is achieved .
it has been shown @xcite that if one solution for @xmath31 and @xmath22 has been found , functions of the form @xmath33)\\ \tilde{f}(e_\gamma)&=&b\,f(e_\gamma)\,\exp(\alpha\,e_\gamma ) \label{eq : solut}\end{aligned}\ ] ] give exactly the same fit to the @xmath29 matrix .
the values of @xmath34 , @xmath35 and @xmath28 can be determined by additional conditions .
the @xmath34 and @xmath28 parameters are used for absolute normalization of the level density @xmath22 : they are adjusted to reproduce ( _ i _ ) the number of levels observed in the vicinity of the ground state and ( _ ii _ ) the neutron resonance spacing at the neutron binding energy @xmath36 .
further details on the extraction procedure and the simulation of errors are given in ref .
@xcite . in the following
we will concentrate on the @xmath0-ray energy dependent function @xmath20 and its normalization .
we assume that the main contributions to the derived @xmath31 function are from e1 and m1 @xmath0-transitions and that the accessible levels of positive and negative parity are equal in number for any energy and spin i.e. @xmath37 thus , the observed @xmath31 is expressed by a sum of the e1 and m1 @xmath0-ray strength functions only @xmath38 e_{\gamma}^3 , \label{eq : prim1}\ ] ] where @xmath35 is the unknown normalization constant .
our experiment does not provide the possibility to derive the absolute normalization of @xmath20 ( see eq .
( [ eq : solut ] ) ) , therefore , the normalization constant has to be determined from other experimental data .
the experimental , average total radiative width of neutron resonances @xmath39 at the neutron binding energy @xmath36 can e.g. be written in terms of @xmath31 . to show this , we start with eq .
( 3.1 ) of ref .
@xcite @xmath40 where @xmath41 is the average total radiative width of levels with energy @xmath19 , spin @xmath42 and parity @xmath43 .
the summations and integration are going over all final levels with spin @xmath44 and parity @xmath45 which are accessible by @xmath0 radiation with energy @xmath10 , multipolarity @xmath12 and electromagnetic character x. if we , again , assume that only dipole radiation contributes significantly to the sum and that the number of accessible levels with positive and negative parity are equal , we obtain , by combining eqs .
( [ eq : prim1 ] ) and ( [ eq : ggdef ] ) , the average total radiative width of neutron s - wave capture resonances with spins @xmath46 expressed in terms of the @xmath47 function @xmath48 where @xmath49 and @xmath50 are the spin and parity of the target nucleus in the ( n,@xmath0 ) reaction and @xmath22 is the experimental level density .
furthermore , we have expressed @xmath22 as the product of the total level density , summed over all spins and the spin distribution @xmath51 .
the spin distribution of the level density is given by @xcite @xmath52,\ ] ] where @xmath53 is the excitation - energy dependent spin cut - off parameter .
the spin distribution is normalized to @xmath54 .
the experimental value of the average total radiative width of neutron resonances @xmath39 is then the weighted sum of contributions with @xmath46 according to eq .
( [ eq : norm ] ) .
because of methodical difficulties , the functions @xmath20 and @xmath21 can not be determined experimentally in the interval @xmath55 mev and @xmath56 mev , respectively .
in addition , the data at the highest @xmath0-ray energies , @xmath57 mev , suffer from poor statistics .
therefore , extrapolations of @xmath31 and @xmath22 were necessary in order to calculate the integral in eq .
( [ eq : norm ] ) .
the contribution from the extrapolation to the total radiative width in eq .
( [ eq : norm ] ) does not exceed 15% , thus the errors due to a possibly poor extrapolation are expected to be of minor importance .
there have been developed several models for the @xmath0-ray strength functions @xmath58 .
the theories behind the models are complicated , and will not be outlined here .
however , the resulting strength functions can be written in simple analytical forms . in this work ,
we have tested various e1 and m1 models . for e1 @xmath0-transitions
these are : * the standard giant electric dipole resonance ( gedr ) model based on the brink - axel approach @xcite @xmath59 where @xmath60 , @xmath61 and @xmath62 are the giant electric dipole resonance parameters derived from photoabsorption experiments .
equals @xmath63 mb@xmath64mev@xmath65 . ] * the model of kadmenski , markushev and furman ( kmf ) @xcite @xmath66 where @xmath67 is the temperature of the nucleus which is usually determined as @xmath68 with @xmath69 being the shifted excitation energy and @xmath70 the level density parameter .
the energy and temperature dependent width of the gedr in this model is expressed by @xmath71 these expressions are developed in the framework of a collisional damping model for @xmath72 and although they should hold for @xmath73 mev , the absence of thermal shape fluctuations in the model limits their validity to @xmath74 mev . for deformed nuclei ,
the giant dipole resonance is split into two components , hence the sum of two strength functions with different gedr parameters has been employed . for m1 @xmath0-transitions
we are testing : * the adjusted single - particle model of weisskopf @xcite where @xmath75 is independent of @xmath10 and the absolute value of @xmath76 has been taken from @xmath77 systematics close to the neutron binding energy @xcite . *
a lorentzian based on the existence of a giant magnetic dipole resonance ( gmdr ) which is assumed to be related to the spin - flip transition between single - particle states @xcite .
the @xmath0-ray strength function in this case is determined by @xmath78 in the following , we will compare these models to the experimental findings .
figure [ fig : level ] shows the experimentally extracted @xmath0-ray energy dependent factor @xmath31 and the level density @xmath22 for @xmath1dy and @xmath2yb .
these data are the very same as in ref .
@xcite , except that the data of @xmath79yb have been retuned by adjusting the parameters @xmath34 and @xmath28 to fit the level density based on known discrete levels at low excitation energy .
we recognize that the shape of the unnormalized @xmath31 functions are rather equal for neighboring isotopes indicating that @xmath31 is a slowly varying function of mass number . in the extraction procedure
, we have used @xmath0-ray spectra from excitation energy bins between 4 and 8 mev . this span in excitation
energy corresponds to a temperature region of 0.5 to 0.7 mev for the initial states , and 0.4 to 0.6 mev for the final states .
the spin window in the pick - up reaction is @xmath80 , which is assumed to be approximately equal for initial and final states .
hence , the discussion below concerns average properties of nuclei for @xmath81 mev and @xmath82 with the assumption of equal density of positive and negative parity states .
the normalized experimental @xmath0-ray strength functions @xmath83 for @xmath1dy and @xmath2yb are presented in fig .
[ fig : norm ] .
the experimental values of the average total radiative width @xmath84
@xcite used to determine the normalization constant @xmath35 are listed in table [ tab : param ] .
the figure shows that each experimental curve consists of two components .
the first one is a smooth function of the @xmath0-ray energy and the second one is connected to a local enhancement of the @xmath0-ray strength function at low @xmath0-ray energies ( @xmath263 mev ) .
the latter component is due to the pygmy resonance , which was first observed in ( n,@xmath6 reactions @xcite .
theoretical curves calculated with the models of sect .
iii are shown as dashed curves in fig .
[ fig : norm ] .
the parameters adopted in the description of the gedr and gmdr are presented in table [ tab : param ] .
the gedr parameters have been determined from the interpolation of systematics over neighboring isotopes @xcite .
for the gmdr parameters , there are no rich experimental systematics available .
our parameters have been taken from ref .
@xcite , namely @xmath85 mev and @xmath86 mev .
the value of @xmath87 has been derived from @xmath88 systematics at @xmath0-ray energies close to the neutron binding energy @xcite .
figure [ fig : norm ] also shows that the lorentzian e1 model [ label a ) , eq .
( [ eq : lorentz ] ) ] gives an acceptable description for @xmath0-ray energies near @xmath36 , in accordance with the systematics of kopecky and uhl for deformed nuclei @xcite .
but for lower energies this gedr model overestimates the experimental data .
the combination of the kmf e1 model and the weisskopf m1 model [ label b ) ] fails to describe the data due to the strong m1 component especially in the region of low @xmath0-ray energies .
the kmf model plus the lorentzian m1 model [ label c ) , eqs .
( [ eq : kad ] , [ eq : m1 ] ) ] is seen to give the best description of the general slope of the @xmath0-ray strength function .
however , since the m1 strength - function model is generally only @xmath89% of the gedr model for the investigated nuclei @xcite , no further conclusion concerning m1 models could be drawn in this work . in detail , the agreement of the last curve with the data for @xmath1dy is satisfactory , excluding the low energy region where the pygmy resonance is observed . for @xmath2yb ,
the slopes of the calculated curves differ somewhat from the experimental ones . in oder to obtain a good parameterization of the @xmath0-ray strength function which can fit the experimental data , the sum of the kmf e1 and the lorentzian m1 models has been selected for further modification .
in contrast to the common use , where the nuclear temperature is defined as @xmath68 , we will keep the temperature fixed ( as first proposed by grudzevich @xcite ) according to a constant temperature model of the nuclear level density , which is supported by recent findings @xcite .
the constant temperature model may also be regarded as to mimic the generalized superfluid model of the nuclear level density @xcite in this excitation energy region .
the mean value of the temperature in the excitation energy region under study is @xmath81 mev for the final levels as has been mentioned above .
we should point out that the temperature dependence of the gedr width used in the e1 model is a much disputed topic .
experimental data on damping of the gedr at low temperatures ( @xmath74 mev ) are absent . at higher temperatures , the damping of the gedr
is intensively studied with inelastic scattering of light particles ( e.g.@xmath28 particles @xcite ) but different theoretical approaches give ambiguous results .
for example , in the adiabatic coupling model @xcite the increasing width is explained in terms of thermally induced shape fluctuations , yielding in general a @xmath90 dependence .
these shape fluctuations become important for @xmath91 mev . in the collisional damping model
@xcite the width of the gedr is due to collisional damping of nucleons , giving a @xmath92 law .
experiments on @xmath93pb show that the data set can be fitted by both of these parameterizations @xcite , while new calculations on the collisional damping model using realistic in - medium cross - sections @xcite show that the width is in general underestimated within this model .
also , a recent calculation @xcite on @xmath94sn within the phonon damping model shows good agreement with experiment . at temperatures appropriate for the present study ( @xmath81 mev ) ,
pairing correlations @xcite and shell effects @xcite have to be taken into account .
most experimental data on the strength function at the low - energetic tale of the gedr are obtained from ( @xmath95 ) reactions , where the quadratic temperature - dependence of the gedr width @xcite is a popular parameterization @xcite .
we therefore use this parameterization , knowing that the model behind can not account properly for the damping mechanism of the gedr @xcite . in order to obtain a good fit of
the chosen @xmath0-ray strength - function models to the data we use the temperature as a free parameter because of the uncertain temperature dependence of the gedr width in our temperature region .
also , a common normalization constant @xmath96 was introduced as a free parameter .
additionally , in order to fit the experimental data in the low - energy region , a lorentzian @xmath97 with three free parameters @xmath98 , @xmath99 and @xmath100 has been used for describing the pygmy resonance . hence , the fitting function consists of the sum @xmath101 where @xmath16 and @xmath102 are given by eqs .
( [ eq : kad ] ) and ( [ eq : m1 ] ) with parameters from systematics ( see table [ tab : param ] ) . the values of the fitting parameters for the pygmy resonance , the normalization constant @xmath96 and the temperature @xmath67 are quoted in table [ tab : fit ] .
the resulting curves are shown as solid lines in fig .
[ fig : norm ] .
the agreement with experiment is excellent .
it can of course be debated that other models for the @xmath0-ray strength functions can be fitted to the experimental data by letting a sufficient number of parameters free for fitting .
we have chosen the sum of the kmf e1 and the lorentzian m1 models with one fit parameter ( @xmath67 ) and one free overall normalization constant ( @xmath96 ) in order to obtain a simple parameterization of the experimental data for further applications .
in addition this parameterization is convenient to extract the parameters of the pygmy resonance experimentally . to judge the relevance of the adopted model of eq .
( [ eq : fitfunc ] ) , it is important to discuss the fitting values obtained .
table [ tab : fit ] shows that the coefficient @xmath96 is close to 1.0 for all four nuclei .
this means that the adopted model reproduces the absolute values of the @xmath0-ray strength functions .
the small deviation of @xmath96 from 1.0 can be explained e.g. by uncertain values of the gedr parameters for the investigated nuclei , or they can be ascribed uncertainties in the experimental normalization of the @xmath0-ray strength functions .
a close inspection of the fitting of eq . ( [ eq : fitfunc ] ) to the data shows that the parameter @xmath67 is mainly determined by the experimental value of the @xmath0-ray strength function in a small @xmath0-energy region around 1.5 mev .
the fitted values for the nuclear temperature are found around @xmath103 mev , a value which is consistent for all four nuclei within the fitting errors .
however , the value is lower than expected from other studies @xcite , giving @xmath81 mev .
since the temperature dependence of the gedr is only roughly determined by theory , we therefore interpret @xmath67 as a fit parameter not necessarily equal to the nuclear temperature .
it is interesting to compare the pygmy resonance parameters to those obtained from ( n,@xmath6 experiments .
unfortunately , the available systematics on pygmy resonances is very scarce @xcite and it is difficult to make definite conclusions .
nevertheless , fig .
[ fig : compar ] shows that our fitted pygmy resonance parameters are in good agreement with ref .
both the resonance energy and the width of the pygmy resonances fit into the systematics .
our strength parameter @xmath98 is in between the values @xmath98 and @xmath104 found in the @xmath105 experiment @xcite , see fig .
[ fig : compar ] .
the comparison supports that the pygmy resonance observed in the two reactions has the same physical origin and manifests itself as a common property of the @xmath0-ray strength function . as a last test
, we have calculated a @xmath0-ray spectrum using the extracted @xmath0-ray strength function @xmath106 and level density @xmath22 for @xmath107dy .
the spectrum includes @xmath0-ray cascades from an excitation energy of @xmath108 and down to the ground state .
the spectrum is calculated using a monte carlo simulation , where @xmath106 and @xmath22 determine the decay pattern . in fig .
[ fig : igashira ] this spectrum is compared to the @xmath0-ray spectrum from the @xmath109dy(n,@xmath110dy reaction ( data points with error bars ) measured at the neutron energy @xmath111 kev @xcite . since the ( @xmath3he,@xmath4 )
data are based on a broader spin window , details in our spectra are expected to be more smeared out at final excitation energies having low level density .
this concerns the fine structures seen below 2 mev and above 5 mev of @xmath0-ray energy .
the overall agreement between the two spectra is gratifying and supports our results .
the @xmath0-ray energy dependent factor and the level density for @xmath1dy and @xmath2yb have been measured using the ( @xmath3he,@xmath4 ) reaction . for the first time , the normalized @xmath0-ray strength function @xmath112 could be extracted from such data .
various models are tested against the observed @xmath0-ray strength function and the best description is found for the e1 model of kmf with a fixed temperature plus lorentzian models for the gmdr and the pygmy resonance .
the pygmy resonance parameters for @xmath1dy and @xmath2yb fit into the available systematics obtained from ( n,@xmath6 experiments .
hence , the adopted approach gives consistent @xmath0-ray strength functions for the investigated nuclei .
a few tentative explanations exist for the pygmy resonance .
still , the question remains open whether the pygmy resonance is of e1 or m1 character .
measurements of the electromagnetic character of the pygmy resonance is therefore important in order to pin down the true nature of this peculiar phenomenon . stimulating discussions with p.f .
bortignon are greatly appreciated .
we also thank m. igashira and s. mizuno for sending us their data on @xmath107dy .
the authors are grateful to e.a .
olsen and j. wikne for providing the excellent experimental conditions .
this work is supported by the norwegian research council ( nfr ) .
blatt and v.f .
weisskopf , _ theoretical nuclear physics ( john wiley & sons , new york , 1952 ) . s.s .
dietrich and b.l .
berman , at .
data nucl .
data tables * 38 , 199 ( 1988 ) .
p. popov , in neutron induced reactions , proceedings of the europhysics topical conference , june 21 - 25 , 1982 smolenice , institute of physics eprc , slovak academy of sciences , bratislava 121 ( 1982 ) .
kadmenski , v.p .
markushev , and v.i .
furman , yad.fiz . * 37 , 277 ( 1983 ) [ sov .
j. nucl .
phys . * 37 , 165 ( 1983 ) ] .
j. kopecky and m. uhl , phys .
c * 41 , 1941 ( 1990 ) .
a. bohr and b.r .
mottelson , _ nuclear structure ( w.a .
benjamin , inc . , reading , massachusetts , 1975 ) vol .
ii , p. 636 .
j. kopecky , m. uhl , and r.e .
chrien , phys .
c * 47 , 312 ( 1993 ) .
m. igashira , h. kitazawa , m. shimizu , h. komano , and n. yamamuro , nucl . phys .
* a457 , 301 ( 1986 ) .
i. bergqvist and n. starfelt , nucl .
phys . * 39 , 353 ( 1962 ) .
bartholomew , i. bergqvist , e.d .
earle , and a.j .
ferguson , can .
j. phys .
* 48 , 687 ( 1970 ) .
barrett , k.h .
bray , b.j .
allen , and m.j .
kenny , nucl.phys .
* a278 , 204 ( 1977 ) .
s. joly , d.m .
drake , and l. nilsson , phys .
c * 20 , 2072 ( 1979 ) .
m. guttormsen , j. rekstad , a. henriquez , f. ingebretsen , and t.f .
thorsteinsen , phys .
* 52 , 102 ( 1984 ) .
d. bohle , a. richter , w. steffen , a.e.l .
dieperink , n. lo iudice , f. palumbo , and o. scholten , phys .
b * 137 , 27 ( 1984 ) .
l. henden , l. bergholt , m. guttormsen , j. rekstad , and t.s .
tveter , nucl .
* a589 , 249 ( 1995 ) . t.s .
tveter , l. bergholt , m. guttormsen , e. melby , and j. rekstad , phys .
* 77 , 2404 ( 1996 ) .
a. schiller , l. bergholt , m. guttormsen , e. melby , j. rekstad , and s. siem , nucl .
instrum .
methods phys .
a * 447 , 498 ( 2000 ) .
e. melby , l. bergholt , m. guttormsen , m. hjorth - jensen , f. ingebretsen , s. messelt , j. rekstad , a. schiller , s. siem , and s.w .
degrd , phys .
lett . * 83 , 3150 ( 1999 ) .
a. schiller , a. bjerve , m. guttormsen , m. hjorth - jensen , f. ingebretsen , e. melby , s. messelt , j. rekstad , s. siem , and s.w .
degrd , nucl - ex/9909011 .
m. guttormsen , a. bjerve , m. hjorth - jensen , e. melby , j. rekstad , a. schiller , s. siem , and a. beli , phys .
c * 62 , 024306 ( 2000 ) .
m. guttormsen , a. atac , g. lvhiden , s. messelt , t. ramsy , j. rekstad , t.f .
thorsteinsen , t.s .
tveter , and z. zelazny , phys.scr .
* t32 , 54 ( 1990 ) .
brink , ph.d .
thesis , oxford university , 1955 .
p. axel , phys .
rev . * 126 , 671 ( 1962 ) .
a. gilbert and a.g.w .
cameron , can .
j. phys .
* 43 , 1446 ( 1965 ) .
j. kopecky and m. uhl , in _ proceedings of a specialists meeting on measurement , calculation and evaluation of photon production data , bologna , italy 1994 ( nea / nsc / doc(95)1 ) p. 119 .
mughabghab , _ neutron cross sections ( academic press , new york , 1984 ) , vol . 1 , part b. g.m .
gurevich , l.e .
lazareva , v.m .
mazur , s.yu .
merkulov , g.v .
solodukhov , and v.a .
tyutin , nucl .
* a351 , 257 ( 1981 ) .
grudzevich , phys . atom .
nucl . * 62 , 192 ( 1999 ) .
grudzevich , phys . atom .
* 63 , 414 ( 2000 ) .
m. guttormsen , m. hjorth - jensen , e. melby , j. rekstad , a. schiller , and s. siem , phys .
c * 61 , 067302 ( 2000 ) .
ignatyuk , k.k .
istekov , and g.n .
smirenkin , yad .
* 29 , 875 ( 1979 ) [ sov .
j. nucl .
* 29 , 450 ( 1979 ) ] .
ignatyuk , j.l .
weil , s. raman , and s. kahane , phys .
c * 47 , 1504 ( 1993 ) .
e. ramakrishnan _
b * 383 , 252 ( 2000 ) .
ormand , p.f .
bortignon , and r.a .
broglia , phys .
rev.lett . *
77 , 607 ( 1996 ) .
ormand , p.f .
bortignon , r.a .
broglia , and a. bracco , nucl.phys .
* a614 , 217 ( 1997 ) .
osman yilmaz , ahmet gokalp , serbulent yildirim , and sakir ayik , phys .
b * 472 , 258 ( 2000 ) . n. dinh dang , k. eisenman , j. seitz , and m. thoennessen , phys.rec .
c * 61 , 027302 ( 2000 ) .
f. bev , p. cejnar , j. honztko , k. konen , i. tomandl , and r.e .
chrien , phys .
c * 52 , 1278 ( 1995 ) .
satoshi mizuno , masayuki igashira , and koji masuda , j. nucl.sci .
technol .
* 36 , 493 ( 1999 ) . * * * * * * * _ * * * * * * * _ _ * * * * * * * * * * * * * * * * _ * * * * _ @xmath109dy&12.13&210&2.6&15.8&250&5.05&7.66&1.60&4.0&108 + @xmath107dy&12.13&210&2.6&15.8&250&5.05&7.65&1.49&4.0&113 + @xmath79yb&12.25&239&2.6&15.5&302&4.80&7.50&1.50&4.0&63 + @xmath122yb&12.25&239&2.6&15.5&302&4.80&7.50&1.76&4.0&75 + @xmath109dy&2.69(4)&0.49(5)&1.37(22)&0.29(11)&1.34(11 ) + @xmath107dy&2.73(5)&0.42(4)&1.35(25)&0.34(10)&1.08(8 ) + @xmath79yb&3.35(6)&0.65(7)&0.97(16)&0.34(3 ) & 1.22(10 ) + @xmath122yb&3.48(7)&0.45(5)&1.30(23)&0.32(2 ) & 1.24(6 ) + -ray strength functions ( data points with error bars ) for @xmath1dy and @xmath2yb . the dashed curves are calculations where a ) denotes the lorentzian gedr model [ eq . ( ) ] , b ) the kmf model [ eq .
( ) ] plus a weisskopf estimate for m1 transitions , and c ) the kmf model [ eq .
( ) ] plus a lorentzian gmdr model [ eq .
( ) ] . for b ) and c ) , the temperature is given by @xmath68 .
the solid curves are the kmf model with constant temperature [ eq .
( ) ] plus a lorentzian gmdr model [ eq . ( ) ] plus a lorentzian pygmy resonance model [ eq .
( ) ] ( see text ) . ]
he,@xmath4 ) reaction ( filled circles ) compared to those from the ( n,@xmath0 ) reaction [ 8 ] ( open circles ) as function of neutron number @xmath123 . in the upper panel ,
the resonance energy @xmath100 is displayed as data points and the width @xmath99 is given by the length of the lines through the data points .
the cross - sections with error bars are shown in the lower panel .
for the ( @xmath3he,@xmath4 ) reaction , the quantity @xmath98 is plotted and assigned an additional systematic error of 20% from the normalization in eq .
( ) . for the ( n,@xmath0 ) reaction , the quantities @xmath98 ( open circles ) and @xmath104 ( open squares ) [ 8 ] are plotted . ]
-ray spectrum for @xmath107dy .
the data points with error bars are taken from the @xmath109dy(n,@xmath110dy reaction [ 39]@xmath0-ray strength function and the level density extracted from the present @xmath124dy@xmath125he,@xmath126dy data .
the calculation is performed by averaging over 100 kev intervals . ] | the @xmath0-ray strength function for @xmath0 energies in the 17 mev region has been measured for @xmath1dy and @xmath2yb using the ( @xmath3he,@xmath4 ) reaction .
various models are tested against the observed @xmath0-ray strength functions .
the best description is based on the kadmenski , markushev and furman e1 model with constant temperature and the lorentzian m1 model .
a @xmath0-ray bump observed at @xmath5 mev is interpreted as the so - called pygmy resonance , which has also been observed previously in ( n,@xmath6 experiments .
the parameters for this resonance have been determined and compared to the available systematics . 2 |
in a recent publication @xcite we have presented a study of event shapes observables and determinations of @xmath9 using data of @xmath12 annihilations at @xmath13 to @xmath4 gev recorded with the former jade detector @xcite at the petra collider .
this study provided valuable information which was not available before from @xmath14-annihilations in the petra energy range .
the results on @xmath9 , obtained in a similar manner as those from the experiments at lep , demonstrated that the energy dependence of @xmath15 is in good agreement with the prediction of quantum chromodynamics ( qcd ) . evolved to the @xmath16 mass scale , the results are in good agreement with those obtained at lep , and are of similar precision .
in addition , power corrections , applied to analytic qcd calculations of the mean values of event shape distributions , were found to qualitatively and quantitatively match the effects of hadronisation .
thus qcd could be tested without the need of phenomenological hadronisation models .
meanwhile the perturbative calculations for the jet broadening variables were improved by including a proper treatment of quark recoil @xcite .
furthermore for the @xmath2-parameter a resummation of leading and next - to - leading logarithm terms to all orders of @xmath9 ( nlla ) became available @xcite . beside these advances in the perturbative description of event shape observables progress was made in the understanding of non - perturbative power corrections to the event shape observables and their mean values .
in particular two - loop calculations of such corrections were performed @xcite which modify the one - loop result of the power correction to the event shapes by a factor ( milan factor ) . in this paper
we complement our previous publication by a new @xmath17+nlla determination of @xmath9 from @xmath2-parameter and update the determination from jet broadening at @xmath18 and @xmath4 gev in the sections [ sec - procedure ] and [ sec - alphas ] .
we also applied power corrections to the @xmath2-parameter distributions and re - investigated those for the mean values of the thrust , heavy jet mass and both jet broadening observables in section [ sec - meanvalues ] .
we start in section [ sec - data ] with a brief summary of the data samples used and draw conclusions from our results in section [ sec - conclusions ] .
we analysed data recorded with the jade detector in 1984 to 1986 at centre - of - mass energies of @xmath19-@xmath20 gev and around @xmath21 gev .
the jade detector was operated from 1979 until 1986 at the petra electron - positron collider at centre - of - mass energies of @xmath22 to @xmath20 gev .
a detailed description of the jade detector can be found in @xcite .
the main components of the detector were the central jet chamber to measure charged particle tracks and the lead glass calorimeter to measure energy depositions of electromagnetic showers , which both covered almost the whole solid angle of @xmath23 .
multihadronic events were selected by the standard jade selection cuts @xcite .
all charged particle tracks , assumed to be pions , with a total momentum of @xmath24 mev@xmath25 were considered in the analysis .
energy clusters in the electromagnetic calorimeter , assumed to be photons , were considered if their energies exceeded @xmath26 mev after correction for energy deposited by associated tracks .
background from two - photon processes and @xmath27-pair events and from events with hard initial state photon radiation were removed by cuts on the visible energy @xmath28 , the total missing momentum @xmath29 ( @xmath30 and @xmath31 are the 3-momentum and the energy of the tracks and clusters ) , the longitudinal balance relative to the @xmath12 beam axis of momenta @xmath32 and the polar angle of the thrust axis , @xmath33 : * @xmath34 ; * @xmath35 ; * @xmath36 ; * @xmath37 .
thus the backgrounds from @xmath38 and @xmath27-pair events were reduced to less than @xmath39 and @xmath40 , respectively @xcite .
the numbers of events which were retained after these cuts for this analysis are listed in table [ tab - eventnumbers ] .
.[tab - eventnumbers ] number of events in data and in monte carlo detector simulation retained after application of the multihadron selection cuts described in the text . [ cols="^,^,^,^",options="header " , ] | e@xmath0e@xmath1 annihilation data recorded by the jade detector at petra were used to measure the @xmath2-parameter for the first time at @xmath3 and @xmath4 gev .
the distributions were compared to a resummed qcd calculation which recently became available for this observable .
in addition , we applied extended resummed calculations to the heavy and wide jet broadening variables , @xmath5 and @xmath6 , which now include a proper treatment of the quark recoil against multi - gluon emission with single - logarithmic accuracy .
we further investigated power corrections to the mean values of the observables mentioned above . in this study , we considered all available e@xmath0e@xmath1 data between @xmath7 and @xmath8 gev . to * measurement of @xmath2-parameter + and determination of @xmath9 from +
@xmath2-parameter and jet broadening + at petra energies * p.a .
movilla fernndez@xmath10 , o. biebel@xmath10 , s. bethke@xmath10 + and the jade collaboration@xmath11 @xmath10 \iii .
institut der rwth aachen , d-52056 aachen , germany + contact e - mail : [email protected] + @xmath11 for a full list of members of the jade collaboration see reference @xcite |
flavour physics has usually been treated as a low - scale effective theory , ignoring high - scale theories , except perhaps for motivating insights from short - distance physics .
it has long been thought by some that high - scale theories with gravity - mediated supersymmetry ( susy ) breaking would typically imply too large flavour - changing neutral current ( fcnc ) contributions .
some people have argued for heavy scalars to suppress the fcncs ( e.g. , @xcite ) , but these are done in the context of the low - scale effective theory , depend on a number of detailed assumptions , and lack deeper motivation .
obtaining detailed flavour predictions from high - scale string theories is difficult because it requires extensive knowledge about the superpotential and the khler potential that is not yet available @xcite . in recent years
there has been progress in constructing string / m - theories compactified to 4d , with broken susy and moduli stabilized ( as is necessary for any theory to be a candidate to describe our world @xcite ) .
generically such theories have moduli to describe the sizes , shapes , and other properties of the curled up dimensions .
the moduli quanta are unstable and decay via gravitational coupling to all matter .
if they decayed too late , the successes of big bang nucleosynthesis would be spoiled , and/or they would carry too much energy density .
their lifetime depends on their mass , so consistency with cosmology generically requires that they have masses heavier than about 20 tev @xcite . the importance of this for flavour physics arises because in generic string / m - theories one can show that the gravitino mass , which measures the effects of susy breaking , and to which the soft - breaking lagrangian is proportional , must then itself be heavier than about 20 tev .
then the supergravity theory implies that the scalar superpartner ( squark and slepton ) masses , heavy higgs masses and also the trilinear couplings must all be larger than about 20 tev as well .
that in turn has major effects on flavour physics because the heavy particles and trilinears mainly decouple , though care is needed when the off - diagonal flavour structure is included .
studying these implications is the main goal of this paper .
the results just stated being the generic properties of string / m - theories , if our world is indeed described by a compactified string / m - theory with susy breaking and stabilized moduli ( which is crucial to define the coupling and masses needed for predictions ) , it would be rather likely that the world is described by a theory with heavy scalars and trilinears . if we specialize to m - theory compactified on a manifold with @xmath1 symmetry , some stronger results hold that may or may not be valid for all string theories . in particular in the m - theory case it has been possible to show that the soft cp - violating phases are zero @xcite , so there is no weak cp problem , and also that the strong cp problem can be solved @xcite . for the @xmath1 compactification ,
once the requirement of a de sitter vacuum and the small cosmological constant are imposed , the gaugino masses are suppressed @xcite . with this perspective , we investigate limits from flavour and cp violation on the trilinear couplings and soft - squared scalar masses in supersymmetric models with light gluinos and heavy scalars .
some analyses of this kind have been considered previously @xcite , focusing on very particular examples , perhaps with only two heavy families , and ad - hoc assumptions for yukawa and trilinear couplings .
one of the reasons for doing this is that a general analysis represents a formidable task without a priori definite information about the form of the yukawa couplings and the supersymmetric spectrum .
however , for a particular set - up consistent with requirements of compactified string theories , we can make some general statements and obtain precise bounds , even though we do not yet know the precise form of yukawa couplings in the @xmath1-mssm models .
our analysis is organized as follows . 1 .
while we concentrate on the @xmath1-mssm , we analyze three different cases : 1 .
the case where trilinear terms are proportional to yukawa couplings at the unification scale @xmath7 .
such a proportionality is often used in the literature for simplicity .
it arises if we have a trivial khler potential , which respects the @xmath1 holonomy , and yukawa couplings that do not depend on hidden - sector fields . due to the running of the parameters , yukawa and trilinear couplings
are not proportional to each other at low energies .
hence , non - zero off - diagonal elements in the trilinear matrices remain even after diagonalizing the yukawa matrices at the electroweak scale .
the case where trilinear terms are not proportional to yukawa couplings at @xmath7 , but such that the non - proportionality is determined by real factors .
thus , complex phases at high energy enter only via the yukawa couplings .
this should generically happen in the context of the @xmath1-mssm @xcite .
this case can be reproduced with a non - trivial khler potential .
we generate a series of random numbers determining the non - proportionality .
the case where trilinear terms are not proportional to yukawa couplings at @xmath7 and where new phases appear at the high scale .
this will represent a scenario beyond the @xmath1-mssm .
we explore this scenario as a contrast to the @xmath1-mssm case so that we can determine whether or not there could be an important impact of the phases . again
, we use random numbers determining the non - proportionality , which are now complex .
+ we assess the impact of the trilinear terms on flavour and cp violation for each case .
2 . at @xmath7
the boundary conditions are as follows : the yukawa matrices of both up- and down - type quarks are non - diagonal complex @xmath8 matrices .
their diagonalizing matrices are similar to the ckm matrix @xmath9 , so their off - diagonal elements are small , except for the right - handed diagonalizing matrix of down quarks , @xmath10 , which has sizable off - diagonal elements . for concreteness ,
we use yukawa matrices constructed in a grand unified model with a family symmetry @xcite .
it is important to mention that we have taken this as a definite example but this yukawa pattern can be embedded in other contexts .
the trilinear terms are also non - diagonal complex @xmath8 matrices , either ( a ) proportional or ( b ) not proportional to the yukawa matrices without new phases or ( c ) not proportional to the yukawa matrices with new cp phases .
the soft - squared mass matrices are proportional to the unit matrix at @xmath7 .
recall that whenever we have a trivial khler metric , the soft - squared masses at that scale will be proportional to the unit matrix , because the same matrices diagonalizing the khler metric will diagonalize the soft - squared matrices .
non - trivial khler metrics could also reproduce diagonal soft - squared masses but lift the universality condition . as long as @xmath11 at the gut scale
, the result from this analysis will be valid .
we focus our studies on supersymmetric mass spectra featuring heavy scalars ( @xmath12 ) and light gauginos .
in particular , the light gluino ( @xmath13 ) , due to its strong interactions , can potentially play a significant role for low - energy observables even if the scalars are heavy . for the specification of such susy mass spectra
, we use the @xmath1-mssm @xcite as a concrete uv - complete model , which helps us to clarify the potential effects of the high - energy physics on the flavour physics phenomena at the electroweak scale .
the model is based on the effective field theories arising from a class of @xmath14 fluxless compactifications of m - theory on a @xmath1 manifold . for concreteness
, we choose a set of benchmark @xmath1-mssm spectra that has been analyzed in @xcite .
4 . in addition to bounds coming mainly from the kaon sector ,
we also consider constraints from the stability of the scalar potential , which are relevant for heavy spectra since they are independent of the mass scale of the supersymmetric particles . in section [ g2results ] for completeness of this work , we summarize the defining features of @xmath1-mssm models
one goal of this analysis is to set bounds on the trilinear and soft - squared masses .
the general relation in supergravity theories @xcite between trilinear and yukawa couplings is @xmath15 \notag \\ & & { } -\braket { { \mathcal{f}}^m } \left [ \braket{\tilde{k}^{\delta\bar\rho } \ , ( \partial_m \tilde k_{\bar\rho{\alpha}})}y_{\delta { \beta } { \gamma } } + ( { \alpha } \leftrightarrow { \beta } ) + ( { \alpha } \leftrightarrow { \gamma } ) \right ] , \label{eq : acontino}\end{aligned}\ ] ] where @xmath16 with @xmath17 , that is the greek indices help to differentiate among the different chiral superfields of the theory .
greek indices with bar are related to operations on the antichiral superfields , e.g. @xmath18 . here
@xmath19 denotes the elements of the inverse matrix . besides
, @xmath20 are hidden - sector fields whose @xmath21-term vacuum expectation values break susy , @xmath22 is the part of the khler potential that depends only on these fields , @xmath23 and @xmath24 . after taking the flat limit
, the visible - sector superpotential has to be rescaled as @xmath25 where @xmath26 is the superpotential of the hidden sector and @xmath27 is the reduced planck mass .
the primed quantities enter into @xmath28 and the unprimed ones into @xmath29 . for simplicity , we assume a trivial matter khler metric @xmath30 . in this case
the soft - squared scalar masses are proportional to the unit matrix , and the second line in eq . vanishes .
however , we allow the yukawa couplings to depend non - trivially on @xmath20 . consequently , the second term in eq .
gives a contribution to the trilinears that is not proportional to the yukawa matrix .
in other words , what we explore here is @xmath31 where @xmath32 are family indices , @xmath33 , and @xmath34 are unknown numbers determining the non - proportionality . it has been realized in @xcite that only for @xmath35 and @xmath36 at @xmath7 or at the scale where the boundary conditions of the set - up are given , we can realize at low energies , near the electroweak scale @xmath37 , the minimal flavour violation ( mfv ) condition @xcite .
however , even with large @xmath38 , this does not imply that fcncs can not be under control .
in fact , even in models with a light supersymmetric spectrum , family symmetries are a nice way to control dangerous fcncs olive:2008vv , abel:2001cv , ross:2002mr , ross:2004qn , antusch:2007re , antusch:2008jf , calibbi:2009ja , calibbi:2010rf , kadota:2010cz , antusch:2011sq . for heavy scalar masses , one may expect that supersymmetric effects will mostly decouple , hence ameliorating the susy flavour problem .
for the concrete examples to be discussed in
[ examplesection ] , for instance , fcncs and cp violation will be suppressed because of the hierarchy between the gaugino and the scalar masses .
however , given the precision of observations especially in the kaon sector , even suppressed susy contributions can be relevant .
the most important indirect tests that most scenarios for physics beyond the standard model ( sm ) have to face are the electroweak precision observables , the anomalous magnetic moment of the muon , fcncs , and cp violation . for the @xmath1-mssm examples we shall discuss in [ examplesection ] , for instance ,
the electroweak parameters are worked out in such a way that contributions due to the large values of higgs masses involved in the theory are avoided .
the higgs sector behaves as an effective single doublet , with one light scalar and the other mass eigenstates heavy . in the fcnc sector
the @xmath39 observables @xmath40 and @xmath41 can indeed give us a hint of ways to restrict boundary conditions of soft terms at @xmath7 . in this section
we discuss the computation of these parameters .
recall that qcd corrections are important for these observables and therefore the different scales involved in the determination of @xmath40 and @xmath41 play an important role . in
[ g2results ] , where we consider specific examples , we mention other processes as well , for example , @xmath42 , @xmath43 , and @xmath44-@xmath45 mixing , which are not constraining .
the cp - violating parameter in neutral kaon mixing is defined as @xmath47 with @xmath48 , where @xmath49 is the effective hamiltonian describing @xmath50 transitions in the @xmath51-@xmath52 system .
the sm prediction and the experimental value of @xmath40 are @xcite @xmath53 respectively .
it is well - known that gluino interactions typically give the most relevant susy contributions to @xmath40 for general soft parameters .
how important these are when the scalars are heavy while the gluino remains light is an interesting question on its own .
we know that the sm and gluino - sdown contributions to @xmath54 are proportional to @xmath55 + \frac{\alpha_s^2}{4 m_{\tilde g}^2 } k_{\tilde g \tilde d } g_{\tilde g}(x_{\tilde g})$ ] , where @xmath56 , @xmath57 and @xmath58 .
are the well - known inami - lim functions @xcite entering in the sm box contributions ; @xmath59 is listed also in appendix [ sec : loop_fcts ] .
@xmath60 is the loop function of the box diagram involving internal gluinos and squarks and is defined in appendix [ sec : loop_fcts ] . ] in figure [ fig : ds_loop_gl_sm_1coups ] we have plotted @xmath61 against @xmath62 $ ] , where @xmath63 and @xmath64 for two different values of the down squark mass , @xmath65 , from top to bottom of the graph . in the models that we study here , we can never have scalar masses as low as @xmath66 .
we show the @xmath66 curve to illustrate the order of magnitude of the effective coupling entering into the cp - violating parameter @xmath40 that such a scalar would produce and to compare with the effect of a heavy scalar with mass above @xmath67 . from the figure
we see that even if @xmath68 , a coupling three to four orders of magnitude bigger than the sm coupling would make the supersymmetric contribution comparable to that of the sm
. such an enhancement factor is possible due to the strong suppression of the sm contribution .
for instance , @xmath69 , where @xmath70 , @xmath71 and @xmath72 are the well - known wolfenstein parameters @xcite . $ ] for @xmath73 .
we can see that for @xmath74 we need a coupling a bit more than three orders of magnitude bigger than in the sm in order to make the susy contribution comparable to the sm one.,width=340 ] for the subject of this work , we are going to see , once we choose a definite flavour structure at @xmath7 encompassing non - diagonal yukawa and trilinear couplings , even if large mixing is present , flavour - violating effects arising from the running of soft parameters ( from @xmath7 down to @xmath75 and then to the kaon scale ) are much smaller than the order of magnitude of the sm contribution and therefore flavour violation is under control in these models .
the details of the boundary conditions at @xmath7 and the running down to the decoupling scale of scalars are given in appendix app : runn .
we follow @xcite to compute the wilson coefficients from the @xmath50 susy processes at the decoupling scale of the heavy scalars .
we have checked explicitly that in our scenario the gluino indeed gives the largest supersymmetric contribution . from the scale
at which scalars decouple , @xmath76 , we then follow the effective hamiltonian approach @xcite .
there are two steps in the evolution from @xmath76 down to the kaon scale .
the first step is to evolve the effective hamiltonian to the scale where the gluinos decouple @xmath77 and the second step is to evolve from there down to the kaon scale .
the wilson coefficients involved in @xmath78 at @xmath76 are @xmath79 , \notag \\ c_2 & = & -\frac{\alpha_s^2}{216 m^2_{\tilde{q } } } { ( \delta^d_\text{rl})}^2_{12 } \,204 x f_6(x ) , \notag \\ c_3 & = & \frac{\alpha_s^2}{216 m^2_{\tilde{q } } } { ( \delta^d_\text{rl})}^2_{12 } \ , 36 x f_6(x ) , \notag \\ c_4 & = & -\frac{\alpha_s^2}{216 m^2_{\tilde{q } } } \left [ { ( \delta^d_\text{ll})}_{12 } { ( \delta^d_\text{rr})}_{12 } \left[504 x f_6(x)- 72 \tilde f_6(x)\right ] - { ( \delta^d_\text{lr})}_{12 } { ( \delta^d_\text{rl})}_{12 } \ , 132 \tilde f_6(x ) \right ] , \notag \\ c_5 & = & -\frac{\alpha_s^2}{216 m^2_{\tilde{q } } } \left [ { ( \delta^d_\text{ll})}_{12 } { ( \delta^d_\text{rr})}_{12 } \left [ 24 x f_6(x ) + 120 \tilde f_6(x ) \right ] - { ( \delta^d_\text{lr})}_{12 } { ( \delta^d_\text{rl})}_{12 } \ , 180 \tilde f_6(x ) \right],\end{aligned}\ ] ] where the operators are given in appendix [ ssc : wilsoncoeff ] . the coefficients @xmath80 and operators @xmath81
are obtained from @xmath82 and @xmath83 , respectively , by interchanging @xmath84 .
the functions @xmath85 and @xmath86 are defined in appendix [ sec : loop_fcts ] .
the mass - insertion parameters are defined as usual , @xmath87 where x , y @xmath88l , r@xmath89 and where a hat denotes a matrix in the super - ckm ( sckm ) basis @xcite , where yukawa couplings are diagonal .
and consequently trilinear terms are rotated as @xmath90 and soft mass squared matrices as @xmath91 and @xmath92 . ]
we take the results of @xcite as a first approximation for the effective hamiltonian at the kaon scale .
in particular , we use the values for the low - energy wilson coefficients given in that work .
this neglects the fact that in the scenario of @xcite one sfermion family is significantly lighter than the two heavy ones , while the models of our interest , to be discussed with concrete examples in examplesection , contain scalar masses of the same order of magnitude . in order to estimate the impact of this difference
, we have calculated the running of the strong gauge coupling due to two - loop qcd corrections with and without the contributions of the first squark family from the scale where the heaviest families decouple to the gluino mass scale .
the difference between the values of @xmath93 in the two cases is only about 4% , which gives us a reason to expect the change in the running of the wilson coefficients not to be dramatic either .
we consider here the most important contributions to @xmath95 , being the parameter measuring the direct cp violation in the decay amplitude of @xmath96 , @xmath97(\text{re } [ a_0 ] \text{im } [ a_2]/\text{re}[a_2]-\text{im } [ a_0])/(\sqrt{2 } \text{re}[a_0])$ ] , @xmath98 , @xmath99 .
] namely the gluino contributions whose significance in light of the heavy scalars is the subject of this work .
they come from the chromomagnetic penguin operators @xmath100 where @xmath101 is the gluon field strength .
the corresponding wilson coefficients @xmath102 and @xmath103 are defined as in @xcite .
the direct cp violation from these operators can be estimated as @xcite @xmath104^{-1 } \ !
\left(c_8(x ) - \tilde c_8(x)\right ) \right],\ ] ] where @xmath105 ( @xmath106 represents the amplitude for @xmath107 ) , @xmath108 is the pion decay constant , @xmath109 represents the uncertainty in the hadronic matrix element calculation for the magnetic operator between @xmath51 and the 2 pion state , @xmath71 represents the running effect from @xmath110 to @xmath111 , @xmath112 the contributions from @xmath102 and @xmath113 coefficients can be decomposed into the chirality changing and conserving contributions as @xmath114^{-1 } \left(c_8(x ) - \tilde c_8(x)\right)=\lambda_\text{llrr}(x ) + \lambda_g(x)$ ] with @xmath115 x \left[-\frac{1}{3}m_1(x)-3 m_2(x ) \right ] , \notag \\
\lambda_\text{llrr}(x ) & = & \left[(\delta^d_\text{ll})_{12}-(\delta^d_\text{rr})_{12}\right ] \frac{m_s}{m_{\tilde g}}\ x \left[-\frac{1}{3}m_3(x)-3 m_4(x ) \right],\end{aligned}\ ] ] where the functions @xmath116 are defined in @xcite and @xmath117 . the chirality - changing terms ,
for the models under consideration in this letter , show up in the down sector and we shall here consider the significant gluino contributions due to the off - diagonal @xmath118-terms which can arise from the non - proportionality between yukawa and trilinear couplings after diagonalizing yukawa couplings @xmath119 @xmath120 being the average of the two diagonal elements as in eq . which on the other hand can keep electric dipole moments ( edms ) sufficiently small @xcite .
the contributions from @xmath121 and @xmath122 can also be relevant if they are much bigger than @xmath123 and could even overcome the enhancement factor @xmath124 that multiplies this last contribution @xcite .
those chirality - conserving mass insertion parameters however turn out to be more stringently constrained from @xmath125 and @xmath40 gabrielli:1995bd , murayama:1999yn , and they can not make significant contributions to @xmath126 under those constraints from those indirect cp violations .
we hence , in the following , discuss the effects of @xmath127 on @xmath128 , which can constrain the potential new physics effects on the flavour - changing interactions that may stem from the non - proportionality of trilinear and yukawa couplings .
before performing the numerical analysis for the flavour violation observables , let us briefly discuss the vacuum stability bounds which constrain the flavour - violating trilinear soft terms by requiring the absence of charge or color breaking ( ccb ) minima and directions unbounded from below ( ufb ) in the scalar potential @xcite .
ccb and ufb constraints can become particularly important or even more stringent than those from fcncs for large soft susy breaking terms , because the former are related to the ratio of scalar masses and trilinear couplings while the latter tend to decrease as the scale of susy breaking increases .
an undesirable deep ccb minimum appears unless the trilinear scalar couplings satisfy @xmath129 , \\
\bigl [ ( \hat m^2_{\tilde{d}\text{ll}})^{}_{ii } + ( \hat m^2_{\tilde{d}\text{rr}})^{}_{jj } + m^2_{h_d}+|\mu|^2 \bigr ] , \label{eq : ccbconsts } \\ \bigl [ ( \hat m^2_{\tilde{u}\text{ll}})^{}_{ii } + ( \hat m^2_{\tilde{u}\text{rr}})^{}_{jj } + m^2_{h_u}+|\mu|^2 \bigr]\end{aligned}\ ] ] in the sckm basis .
analogously to the ccb bounds , the ufb bounds for off - diagonal trilinear scalar couplings read - flat direction @xmath130 @xmath131 in the limit @xmath132/[(\hat y_{ii}^e)^2 + ( \hat y_{jj}^e)^2]$ ] with @xmath133 @xcite . ]
@xmath134 , \label{eq : ufblepton } \\ \bigl [ ( \hat m^2_{\tilde{d}\text{ll}})^{}_{ii } + ( \hat m^2_{\tilde{d}\text{rr}})^{}_{jj } + ( \hat m^2_{\tilde\nu})^{}_{mm } \bigr ] , \label{eq : ufbconsts } \\ \bigl [ ( \hat m^2_{\tilde{u}\text{ll}})^{}_{ii } + ( \hat m^2_{\tilde{u}\text{rr}})^{}_{jj } + ( \hat m^2_{\tilde{e}\text{ll}})^{}_{pp } + ( \hat m^2_{\tilde{e}\text{rr}})^{}_{qq } \bigr],\end{aligned}\ ] ] where @xmath135 and @xmath136 . while one can not give general predictions for the values of trilinear parameters without specifying the dependence of the khler potential and yukawa couplings on the hidden - sector fields as pointed out in eq .
( [ eq : acontino ] ) , we shall restrict the range of the off - diagonal terms @xmath137 by these ccb / ufb bounds when we perform the numerical studies in [ g2results ] .
let us briefly overview the basic properties of the @xmath1-mssm and their origin before discussing the flavour issues .
the starting point is a compactified m - theory , which is assumed to have the mssm embedded in the @xmath1 manifold , with no extra matter , following the work of witten @xcite .
the gauge group is not extended from the sm one .
supersymmetry breaking arises from the gaugino condensation mechanism , which is generic in this theory , and leads to a non - vanishing gravitino mass .
the supergravity theory then allows calculating all the soft - breaking parameters in terms of the gravitino mass ( detailed calculations in @xcite )
. then the scalar ( squark and slepton and higgs sector ) masses are equal to the gravitino mass with small corrections and the trilinear factors are close to the scalar masses . the moduli khler and super - potentials of @xmath0-mssm models are partially determined @xcite @xmath0-holonomy khler potentials but the matter khler potentials are not @xcite . in m - theory
the moduli are stabilized generically because all moduli occur on an equal footing in the gauge kinetic function , and it occurs in the superpotential , so the moduli have some interactions and therefore a potential with a minimum .
their vacuum expectation values and masses can be calculated . in the @xmath1-mssm ,
both moduli khler and super - potentials are basic ingredients used for the stabilization of moduli . however
, matter khler and super - potentials do not play a role in the stabilization .
although these must also respect @xmath0-holonomy , the many possibilities can be reduced by studying their low - energy phenomenology . in this respect
the khler metric , of the @xmath1-mssm considered so far , is assumed to be diagonal since the families arise at singularities on the manifold that are unlikely to overlap .
however , non - trivial corrections to the off - diagonal elements of the khler metric may appear through higher corrections in terms of hidden sector fields .
studying effects of non - diagonal and non - universal diagonal terms phenomenologically is beyond the scope of this work , however we explore here some indirect effects by allowing deviations from the consequences of assuming a trivial khler metric , that is by studying effects of the non - proportionality of trilinear and yukawa couplings . in any case , the phenomenology of these models is characterized by a suppression of gaugino masses relative to the gravitino and the moduli masses . that scalars ( squarks , sleptons , etc . ) should be heavier than about @xmath138 is more general than the @xmath1-mssm , depending only on the generic derivation that the moduli masses are connected to the gravitino mass , the moduli masses have a lower bound of order @xmath138 from robust cosmological arguments , and supergravity implies the scalar masses are closely equal to the gravitino mass .
we are now in a position to illustrate our aforementioned analysis using examples with a concrete uv - completion .
we consider for this purpose the @xmath1-mssm spectra shown in table [ tab2 ] , which are characterized by heavy scalar masses of order the gravitino mass ( @xmath139 ) and a light gluino ( @xmath140 ) . ' '' '' ' '' '' & point 1 & point 2 & point 3 & point 4 & point 5 & point 6 & point 7 + @xmath141 ' '' '' & 20000 & 20000 & 20000 & 20000 & 30000 & 50000 & 30000 + @xmath142 ' '' '' & 3 & 2.65 & 2.65 & 3 & 3 & 2.5 & 3 + @xmath143 & -11943 & -13377 & -13537 & -10969 & -10490 & -34019 & + 17486 + @xmath144 & & & & & & & + @xmath61 & 401 & 449 & 622 & 492 & 1784 & 1001 & 596.8 + @xmath145 & 145.1 & 155.6 & 189 & 170 & 473 & 373.4 & 271 + @xmath146 & 153 & 159 & 214.3 & 181.5 & 702.4 & 397 & 334.2 + @xmath147 & 145.2 & 155.8 & 214.5 & 181.7 & 702.6 & 373.6 & 334.2 + @xmath148,@xmath149 & 19799 & 19803 & 19809 & 18785 & 21052 & 49524 & 29727 + @xmath150 & 15342 & 15250 & 15224 & 14635 & 16783 & 38473 & 23236 + @xmath151 & 9130 & 8779 & 8662 & 8928 & 11151 & 22887 & 14264 + @xmath152 & 19848 & 19851 & 19845 & 18832 & 21096 & 49694 & 29794 + @xmath153 & 19849 & 19851 & 19856 & 18832 & 21096 & 49695 & 29767 + @xmath154 & 15342 & 15251 & 15224 & 14635 & 16783 & 38470 & 23235 + @xmath155 & 116.4 & 114.3 & 114.6 & 116.0 & 115.9 & 115.1 & 114.6 + @xmath156,@xmath157,@xmath158 & 24614 & 25846 & 25943 & 23158 & 25029 & 65690 & 36623 + the stabilization of moduli requires setting up the gravitino mass as @xmath159 and as a result gives a definitive hierarchy of masses .
1 . heavy particles : the susy higgses , the superpartners of the fermions and the higgsinos are heavy , since their masses are related to the gravitino mass as @xmath160 2 . light particles : gauginos and sm particles .
the gauginos become light because they are suppressed when the constraints are imposed that require a de sitter vacuum and a small cosmological constant @xcite .
3 . trilinear and yukawa couplings : the overall scale of trilinear terms is @xmath161 at @xmath7 @xcite .
so far only particular cases of matter khler potentials have been studied .
these studies have considered a proportionality between the yukawa couplings and the trilinear terms .
the running of the @xmath0-mssm parameters from the scale where @xmath1-holonomy moduli are stabilized to @xmath162 has been performed by some authors @xcite .
these works did not take into account the running of @xmath8 yukawa and trilinear matrices but only the running of the third - family parameters . as these effects and the moderate deviations from a proportionality between trilinear and yukawa matrices we consider can not have a significant influence on the masses of the superparticles , we use the results of @xcite for the mass spectra of seven @xmath1-mssm benchmark points , as shown in table [ tab2 ] .
they were calculated numerically using softsusy @xcite , thus taking into account the two - loop running and ensuring correct electroweak symmetry breaking as well as the absence of tachyons . in order to calculate the low - energy mass - insertion parameters , we employ a one - loop leading - log approximation of the running of the complex yukawa , trilinear and soft - squared mass matrices .
numerical checks with spheno 3.1.5 @xcite and softsusy 3.2.3 @xcite indicate that our approximation is rough but yields the correct order of magnitude .
we will see that this accuracy is sufficient for the scenario studied in this work .
we combine the @xmath1-mssm spectra with the yukawa couplings as given by the case of fit 4 of @xcite where we have updated the values of the yukawa coefficients at the gut scale @xmath7 , @xmath163 \notag \\ & & y^u=\frac{\sqrt{2 } m_t}{v \sin\beta } 0.53\left [ \begin{array}{ccc } -1.58 \times 10^{-6}-0.000017 i & -0.000076 + 0.000032 i & \ 0.0020 + 0.0020 i \\ -0.00034 + 0.00024 i & 0.0020 + 0.0002 i & 0.011 + 0.011 i \\
-0.0057 - 0.0024 i & 0.0044 + 0.0115 i & 0.70 + 0.71 i\end{array } \right ] \notag \\ & & y^e= \frac{\sqrt{2 } m_{\tau}}{v \cos\beta } \left [ \begin{array}{ccc } 0.0014 - 0.0007 i & 0.0005 - 0.0056 i & 0.13 - 0.13 i \\ 0.0082 & 0.023 - 0.059 i & 0.18 - 0.1 i \\
0.0018 + 0.0009 i & 0.035 - 0.029 i & -0.99 - 0.09 i\end{array } \right ] .
\label{eq : yuk_hier_at_mg}\end{aligned}\ ] ] as boundary conditions for the trilinear couplings , we use the relation and @xmath164 with the exception of @xmath165 , which is fixed to be @xmath166 in order to preserve the aforementioned prediction for the overall scale of the trilinear couplings .
all relations are valid at @xmath7 .
the maximum absolute value of @xmath167 is chosen to ensure that the running does not create off - diagonal elements in the soft - squared mass matrices that are larger than the diagonal elements , as explained in appendix [ app : runn ] .
for the case ( c ) above , in table [ tbl : c_f_random ] we show the values of the coefficients @xmath168 that have produced the maximum values of the flavour - violating parameters @xmath169 , which are listed in table [ tbl : deltas12xy ] . for completeness
we also show the values of @xmath170 . we have chosen the matrix of coefficients @xmath171 . for all sm parameters we use the values of @xcite .
[ [ section ] ] in the @xmath1-mssm cases the susy contribution to @xmath172 is really small , therefore we can express @xmath173 with @xmath174 .
at the scale of @xmath67 , @xmath175 is the leading contribution , while the other flavour - violating parameters are at least one order of magnitude smaller .
the flavour - violating parameter @xmath123 involves a yukawa coupling due to the chirality flip and is therefore suppressed for very heavy scalars .
the values of @xmath122 in table [ tbl : deltas12xy ] yield a contribution @xmath176 , safe enough in comparison to the sm contribution and the experimental limit of order @xmath177 .
the parameters @xmath178 , which dominate the chargino contribution to the kaon observables @xcite , are @xmath179 .
however , we have checked that this chargino contribution is still smaller than that mediated by the gluino .
it is important to mention that in the down sector only @xmath180 , while @xmath181 contains large mixings .
together with the relatively large difference between the masses of @xmath182 and the squarks of the first and second generation , this is responsible for the difference of the off - diagonal elements in @xmath183 and @xmath184 in this case .
we have checked the validity of the mass - insertion approximation via the vertex mixing method , analogously to the discussion for the case of @xmath185 in section iv of @xcite . in figure
[ fig : epsk_allg2_np_y ] we plot the predictions for @xmath186 for the set - up , where trilinear terms are not proportional to yukawa couplings but no new phases are involved . as @xmath187 , where @xmath188 is the total susy contribution and the individual terms refer to the charged higgs , the chargino , the neutralino , the neutralino - gluino , and the gluino contribution , respectively .
] considering the uncertainty of @xmath189 , eq . , it is possible to be in agreement with the experimental value @xmath190 for all the benchmark points .
in fact , the order of magnitude of the supersymmetric contribution , @xmath188 , is the same , that is @xmath191 , as in the case with trilinear terms proportional to yukawa couplings . in both cases ,
no points violate the ccb / ufb bounds . for the case that trilinear terms are not proportional to yukawa couplings but there are no new phases involved .
if we add these values to the sm value together with the corresponding uncertainty , we can be in agreement with the experimental value , since the order of magnitude of @xmath188 is @xmath191.,width=434 ] finally in figure [ fig : epsk_allg2 ] we plot the maximum ranges of values of @xmath188 for the case where trilinear terms are not proportional to yukawa couplings and where we have used eq . with the complex random coefficients @xmath192 as defined in eq .
we have considered only such values for the coefficients that are allowed by the ccb / ufb constraints .
this excludes about 10% of the points in the random scan .
figure [ fig : epsk_allg2 ] shows that for each @xmath1 point all the values are in agreement with the experimental value @xmath190 since all contributions to @xmath188 are still at most of @xmath193 .
the benchmark points @xmath194@xmath195 yield significantly smaller susy contributions due to the larger gravitino mass and consequently heavier scalars compared to points @xmath166@xmath196 .
, except that in this case trilinear terms are not proportional to yukawa couplings and new phases appear , i.e. , we have used eq . with complex coefficients .
for each @xmath1-mssm point we obtain values in agreement with the experimental value @xmath197 since all contributions to @xmath198 are of @xmath193.,width=434 ] [ [ textrelargeprotectepsilonprime - protectepsilon ] ] @xmath199 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + all kinds of mass insertions contribute to @xmath41 @xcite , however those potentially large are the ones multiplied by the factor @xmath124 , which are @xmath200 and @xmath201 , contained in the sum of the terms @xmath202 . due to the hierarchy of mass insertions we have found in this example , @xmath203 @xmath204
@xmath205 @xmath206 , we have checked if contributions from @xmath121 and @xmath122 could play an important role .
the current experimental average of @xmath207 from ktev and na48 is @xcite @xmath208 with a conservative theoretical uncertainty , the sm contribution is @xmath209 @xcite . for the case of trilinear terms proportional to yukawa couplings , eq .
, the susy contribution to @xmath210 is of the order @xmath211 for all @xmath1-mssm points , as expected because the off - diagonal trilinear terms generated after the running are too small . for trilinear terms not proportional to yukawa couplings , figure [ fig : epefig_ynp ] shows the values of @xmath212 in the case where no phases are involved , eq . , while the results with new phases , eq . ,
are plotted in figure [ fig : epefig_yphases ] . in all cases
the susy contribution is smaller than @xmath213 and thus negligible .
[ fig : epefig_ynp ] for the case where trilinear terms are not proportional to yukawa couplings but there are no new phases involved , eq . .
it is far smaller than the observed value @xmath214 .
the points correspond to different random choices of the parameters determining the relation between trilinear and yukawa couplings .
, title="fig:",width=434 ] [ fig : epefig_yphases ] for the case where trilinear terms are not proportional to yukawa couplings and where the parameters determining the relation between trilinear and yukawa couplings are complex , eq . .
the points correspond to different random choices of these parameters .
the susy contribution is far smaller than the observed value @xmath214 .
some of the scanned points ( larger circles , in red / light shade ) violate the ccb / ufb constraints .
this happens for large off - diagonal trilinear couplings , as can be seen in eqs .
( [ eq : ccbconsts],[eq : ufbconsts ] ) . however , there are always nearby sets of parameters that give safe potentials .
, title="fig:",width=434 ] [ [ electric - dipole - moments . ] ] electric dipole moments .
+ + + + + + + + + + + + + + + + + + + + + + + + we have discussed the effects of the off - diagonal trilinear couplings , but there are constraints on the diagonal terms as well .
for instance , the experimental upper limit on the mercury edm constrains the imaginary part of @xmath215 .
according to table 3 of @xcite , @xmath216 if we use @xmath217 and the smallest value @xmath218 considered in @xcite .
for the smaller @xmath219 we encounter in the @xmath1-mssm , the bound may be relaxed by about an order of magnitude @xcite . in all the cases analyzed
, we have found that @xmath220 is at most @xmath221 , while @xmath222 is at most @xmath191 .
an analogous constraint for @xmath223 can be estimated from the electron edm whose approximate contribution reads @xcite @xmath224 where @xmath225 and the loop function @xmath226 is given in appendix [ sec : loop_fcts ] .
we can use this rough formula to compare with the experimental upper bound of @xmath227 @xcite . in our case
, this limit requires that @xmath228 .
for all the cases analyzed here , we have found that @xmath229 is at most @xmath191 .
[ [ mathbfg-2 . ] ] @xmath230 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + the main contributions from susy to the anomalous magnetic moment of the muon , @xmath231 , come from the smuon - neutralino and the sneutrino - chargino couplings @xcite . the observed value of @xmath232 is larger than the sm prediction by @xmath233 @xcite , so if susy was relevant for @xmath232 , the muon - neutralino and the sneutrino - chargino contributions would have to be of this order . in the case of @xmath1-mssm spectra with two light gauginos and a light chargino ,
only the diagrams involving these particles are relevant . in this case
@xmath234 the couplings @xmath235 are of the order @xmath236 , that is @xmath237 , while @xmath238 are at most @xmath239 .
both are too small to produce a significant contribution to @xmath240 , and thus can not explain the deviation from the sm expectation that has been experimentally observed @xcite .
we stress , however , that the sm computations are not final and the predicted sm value could change @xcite . [
[ b - decays ] ] b decays + + + + + + + + constrain the flavour - violating parameters @xmath241 from their gluino - sdown , neutralino - sdown contributions and @xmath242 from chargino - sup and charged higgs - sup contributions .
we know that these processes are sensitive to the squark mass scale and thus expected to be quite small for the models at hand .
thus , in order to have an idea of the order of magnitude of the decay width , @xmath243 , we can estimate the contribution to the squared amplitudes .
indeed for the cases analyzed here , the gluino - sdown contribution is the largest .
this is because of the kind of yukawa matrices we have chosen . at this scale and at leading order @xmath244 where the numerical value range corresponds to the range necessary to saturate the experimental 2@xmath245 region @xcite and the leading - order expressions for the wilson coefficients correspond to those of @xcite . in the expression above
, @xmath246 is the decay scale .
remember that we have to make the comparison at that scale because the gluino contribution follows a different qcd correction from @xmath247 down to @xmath248 @xcite .
for the analysis we follow @xcite .
for all the points analyzed , we have found at most @xmath249 . with a light chargino , @xmath250 , and a light gluino , one may wonder if the chargino - stop and gluino - sdown loops could ever compete significantly , in cases where the contribution could be of a concern
. in @xmath251 decays the leading terms in the amplitudes for these two diagrams are proportional to @xmath252 and @xmath253 .
the mixing in the @xmath254 and @xmath255 sectors is parameterized by @xmath256 and @xmath257 , respectively .
the loop functions @xmath258 and @xmath259 are of course different but they are of similar size whenever the ratios @xmath260 and @xmath261 are comparable
. then as long as these mass ratios are similar , a cancellation could occur or not , depending on the correlation of the mixing in the @xmath254 and @xmath255 sectors . in our case
this does not occur because the mixing in these sectors is quite different .
[ [ mathbfd0-barmathbfdmathbf0-mixing ] ] @xmath262-@xmath263 mixing + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + is known for setting strong requirements on @xmath264 . for a light spectrum with @xmath265 and
@xmath61 around @xmath266 , the upper limits lie between @xmath177 and @xmath267 @xcite .
they are sensitive to the susy mass scale and become weaker for larger @xmath265 . for the models discussed here ,
we find @xmath268 , so the susy contribution to @xmath44-@xmath269 mixing is negligible .
we have mainly focused on the effects of light gluinos on flavour- and cp - violating processes . in the considered scenario with heavy scalars , the susy contributions to flavour and cp observables are two to three orders of magnitude smaller than the sm contributions .
so even if the rather crude approximations we used underestimated the susy contributions by an order of magnitude , the conclusion that they are negligible would still hold . therefore , an order - of - magnitude estimate is sufficient for the present study .
we leave for future work a more general study with improved accuracy and a set - up where we vary the mass scale of gluinos versus that of down squarks and sleptons .
it was beyond the scope of this letter to give a general detailed study of lepton flavour violation , partly due to the uncertainties of lepton mixing @xcite .
however , we have checked that the flavour - violating parameters of the @xmath1-mssm models that we have analyzed are pretty small and safe .
the orders of magnitude for all cases analyzed in [ g2results ] are presented in table [ tbl : deltas - eijxy ] . in this paper , motivated by the recognition that generically compactified string and m - theories predict heavy sfermion masses and trilinear couplings ( @xmath270 ) , we have studied in detail whether or not the decoupling effects could leave any remaining places where phenomenological issues could arise .
while no concerns emerge , it is important to understand the effects on the relations between yukawa couplings and trilinear terms , together with the improvement of hadronic uncertainties .
this will help to limit the sizes and phases associated to the trilinear couplings .
most interestingly , some sets of parameters could lead to ccb or ufb potentials , but nearby sets of parameters always exist which give safe potentials .
we also note that , contrary to what has been experimentally observed @xcite , the value we obtain for @xmath240 is too small to change the sm one and hence can not provide an agreement with the experimental value at the @xmath271 c.l
. however , the predicted sm value could change @xcite due to the many uncertainties in its calculation . due to the very heavy squarks and sleptons characteristic for the studied scenario unacceptably large flavour- or cp - violating effects can be avoided .
therefore , in the models suggested by the compactified string / m - theories that predict heavy scalars and trilinears and assuming off - diagonal elements of yukawas and trilinears that are not unusually large , gravity mediation of susy breaking does not have serious flavour and cp problems .
let us here briefly comment on the issue of a possible tachyonic stop .
while most part of flavour constraints can be relaxed by heavy first and second generation scalars , it has been pointed out that such heavy scalars could drive the squared mass of the stop , @xmath272 , negative via the renormalization group evolution unless @xmath273 @xcite . in the models of our interest , however , the scalars typically have a common mass scale of order the gravitino mass , a few tens of tev , at the gut scale , while the gaugino masses are typically a couple of hundred gev at the gut scale .
such a constraint avoids tachyonic scalars and hence is not a concern for us in this letter .
we particularly thank bobby acharya for encouragement and participating in the initial stages of the analysis , and we also thank francesca borzumati for clarifications on issues regarding @xmath274 .
we thank luca silvestrini for clarifications on the loop functions entering into the wilson coefficients of the @xmath50 hamiltonian .
we thank brent nelson and werner porod for useful discussions .
j.k . would like to thank cinvestav in mexico city for hospitality during stages of this work .
this work was supported by the michigan center for theoretical physics , and by the german science foundation ( dfg ) via the junior research group `` susy phenomenology '' within the collaborative research centre 676 `` particles , strings and the early universe '' .
[ tbl : deltas12xy ] .maximum values of @xmath275 for the scanned values of the trilinear parameters for the case of eq . .
[ cols="^,^,^,^,^",options="header " , ]
we follow various references @xcite for the extraction of the effective hamiltonian . the @xmath276 operators involved in eq .
are @xmath277 where @xmath278 and @xmath279 are the left- and right - handed projection operators , respectively , @xmath280 , and @xmath281 .
we collect in this appendix the loop functions that we have used in our analysis . @xmath282
we take the scalar soft squared mass matrices to be proportional to the unit matrix at @xmath7 .
their running to @xmath284 , the scale at which the scalars decouple , will produce off - diagonal entries .
we require these off - diagonal elements to be significantly smaller than the diagonal elements , since otherwise the mass - insertion approximation would not be justified . to be concrete and conservative , let us consider @xmath285 and demand @xmath286 at @xmath76 .
as we consider small values of @xmath142 , we can neglect the contributions of @xmath287 and @xmath288 to the running .
furthermore , we consider ckm - like matrices diagonalizing @xmath289 , which implies @xmath290 .
consequently , @xmath289 does not affect the running of the off - diagonal elements of @xmath291 .
of course , the same is true of the terms in the renormalization group equation ( rge ) of @xmath291 that involve gauge couplings .
thus , the only relevant terms in the rge are those proportional to @xmath292 and @xmath293 . approximating the right - hand side of the rge by a constant value ( leading - log approximation )
, we then obtain at @xmath76 @xmath294 , [ eq : c_randomnum ] ) , @xmath295 as well as @xmath296 , where @xmath297 is the sine of the cabibbo angle , and assuming no accidental cancellations , we obtain @xmath298 where @xmath299 is the maximum value of @xmath300 .
thus , the strongest constraint stems from @xmath301 and @xmath302 in eq . , @xmath303 with @xmath304 , this yields @xmath305 . to be conservative ,
we have chosen @xmath306 for our numerical analysis .
[ [ trilinear - terms . ] ] trilinear terms .
+ + + + + + + + + + + + + + + + the term mfv @xcite refers to scenarios where _ all higher - dimensional operators , constructed from sm and fields with yukawa interactions , are invariant under cp and under the flavour group @xmath307 .
_ here @xmath308 , and the yukawa couplings are formally regarded as auxiliary fields that transform under @xmath307 . as a consequence ,
_ mfv requires that the dynamics of flavour violation is completely determined by the structure of the ordinary yukawa couplings and in particular , all cp violation originates from the ckm phase_. because of the running of all couplings of a theory , this scenario can only be realized at one particular scale , usually a low energy scale . starting with parameters defined at @xmath7 , mfv
can only be a good approximation at @xmath309 , if @xmath310 where @xmath311 is a universal mass parameter for all families and kind of fermions that is small in comparison with other soft masses of the theory _ and _ if yukawa couplings are small .
this can be analyzed by studying the dependence of the rges of yukawa and trilinear couplings on @xmath312 @xcite .
is diagonal and @xmath289 is not . recall that it is not possible to work in a basis where both are diagonal precisely due to the ckm matrix . ]
if just the third family yukawa couplings are evolved , of course the size of @xmath311 does not matter because no off - diagonal terms are produced . with a full rg evolution of complex @xmath313 yukawa and trilinear matrices with small off - diagonal values , mfv can be emulated , albeit never reproduced , for sufficiently small values of @xmath314 @xcite .
the one - loop running of the soft - squared parameters @xmath315 in the sckm basis is governed by the @xmath316 functions @xmath317 where @xmath318 and the functions contain flavour - diagonal contributions to the running involving gauge couplings and gaugino masses .
note that at an arbitrary scale @xmath319 , the terms which contain @xmath320 are not diagonal because of the different running of the diagonal elements in .
therefore , off - diagonal terms will necessarily be induced .
recall that even if we consider only the running of the yukawa couplings of the third family , this will produce a split in the masses of @xmath291 .
we can always choose to go to the basis where one of the yukawa couplings is diagonal at @xmath7 , but this does not guarantee diagonal soft mass - squared matrices in the sckm basis because the fact that @xmath321 necessarily implies that not all of the matrices @xmath322 and @xmath323 are diagonal .
as it is known @xmath40 is very sensitive to this @xcite .
if the coupling of the particles beyond the sm was of the same order as that of the sm particles , this would push the limit on the scale of new physics entering into the @xmath50 processes up to @xmath324 in the lepton sector , we assume heavy right - handed neutrinos that decouple close to the gut scale .
therefore only the superpartners of right- and left - handed charged leptons as well as left - handed neutrinos can induce flavour violation .
considering the structure of fermion masses we are using , see section [ sbs : hieryukcoups ] and references therein , the yukawa coupling matrix for neutrinos is the same as that for the up - quark sector at @xmath7 , therefore the flavour violation induced in this scenario is relatively small .
table [ tbl : deltas - eijxy ] shows the mi parameters relevant for the observables @xmath325 . in the @xmath1-mssm case , where all the examples that are known @xcite correspond to the case that trilinear couplings are proportional to yukawa couplings
, we can have a theory , depending on the choice of yukawa couplings , for which at low energy , all flavour violation present is below the experimental bounds .
it is only in this sense that we can say that we have an _ ultraviolet version of mfv _ but not in the sense in which mfv is defined . for the case of the relation with @xmath192 as in eq .
, that is @xmath326 real random numbers between @xmath327 and @xmath328 , at low energies cp violating phases in addition to the ckm phase appear but also flavour violation is below experimental bounds . with @xmath192 as in eq . , i.e. , with random numbers between @xmath327 and @xmath328 and explicit cp phases at @xmath7 , flavour violation is more difficult to neglect but still below the experimental bounds .
this means that even with more cp phases than in the sm and large mixing present due to the choice of yukawa couplings , after the running to low energy we obtain a theory which satisfies flavour violation constraints .
not surprisingly , the reason are the large scalar masses .
what is not a trivial result of the analysis is that still bounds on the size of trilinear terms can be obtained .
b. s. acharya , g. kane , and e. kuflik , `` string theories with moduli stabilization imply non - thermal cosmological history , and particular dark matter '' , http://arxiv.org/abs/1006.3272[arxiv:1006.3272 [ hep - ph ] ] .
a. g. cohen , d. kaplan , and a. nelson , `` the more minimal supersymmetric standard model '' , http://dx.doi.org/10.1016/s0370-2693(96)01183-5[_phys .
lett . _ * b388 * ( 1996 ) 588598 ] , http://arxiv.org/abs/hep-ph/9607394 [ arxiv : hep - ph/9607394 ] . m. k. gaillard and b. d. nelson , `` on quadratic divergences in supergravity , vacuum energy and the supersymmetric flavor problem '' , http://dx.doi.org/10.1016/j.nuclphysb.2006.05.035[_nucl
. phys . _ * b751 * ( 2006 ) 75107 ] , http://arxiv.org/abs/hep-ph/0511234 [ arxiv : hep - ph/0511234 ] .
g. kane , p. kumar , and j. shao , `` cp - violating phases in m theory and implications for electric dipole moments '' , http://dx.doi.org/10.1103/physrevd.82.055005[_phys .
_ * d82 * ( 2010 ) 055005 ] , http://arxiv.org/abs/0905.2986[arxiv:0905.2986 [ hep - ph ] ] .
b. s. acharya , k. bobkov , and p. kumar , `` an m theory solution to the strong cp problem and constraints on the axiverse '' , http://dx.doi.org/10.1007/jhep11(2010)105[_jhep_ * 1011 * ( 2010 ) 105 ] , http://arxiv.org/abs/1004.5138[arxiv:1004.5138 [ hep - th ] ] .
b. s. acharya , k. bobkov , g. l. kane , p. kumar , and j. shao , `` explaining the electroweak scale and stabilizing moduli in m theory '' , http://dx.doi.org/10.1103/physrevd.76.126010[_phys .
* d76 * ( 2007 ) 126010 ] , http://arxiv.org/abs/hep-th/0701034 [ arxiv : hep - th/0701034 ] . n. arkani - hamed and h. murayama , `` can the supersymmetric flavor problem decouple ? '' , http://dx.doi.org/10.1103/physrevd.56.r6733[_phys .
* d56 * ( 1997 ) 67336737 ] , http://arxiv.org/abs/hep-ph/9703259[arxiv:hep-ph/9703259 ] .
g. f. giudice , m. nardecchia , and a. romanino , `` hierarchical soft terms and flavor physics '' , http://dx.doi.org/10.1016/j.nuclphysb.2008.12.030[_nucl .
_ * b813 * ( 2009 ) 156173 ] , http://arxiv.org/abs/0812.3610 [ arxiv:0812.3610 [ hep - ph ] ] .
n. arkani - hamed , s. dimopoulos , g. giudice , and a. romanino , `` aspects of split supersymmetry '' , http://dx.doi.org/10.1016/j.nuclphysb.2004.12.026[_nucl .
* b709 * ( 2005 ) 346 ] , http://arxiv.org/abs/hep-ph/0409232 [ arxiv : hep - ph/0409232 ] .
g. kane , s. king , i. peddie , and l. velasco - sevilla , `` study of theory and phenomenology of some classes of family symmetry and unification models '' , http://dx.doi.org/10.1088/1126-6708/2005/08/083[_jhep_ * 0508 * ( 2005 ) 083 ] , http://arxiv.org/abs/hep-ph/0504038 [ arxiv : hep - ph/0504038 ] .
b. s. acharya , k. bobkov , g. l. kane , j. shao , and p. kumar , `` @xmath1-mssm : an @xmath329 theory motivated model of particle physics '' , http://dx.doi.org/10.1103/physrevd.78.065038[_phys .
rev . _ * d78 * ( 2008 ) 065038 ] , http://arxiv.org/abs/0801.0478 [ arxiv:0801.0478 [ hep - ph ] ] .
b. s. acharya and k. bobkov , `` kahler independence of the @xmath1-mssm '' , http://dx.doi.org/10.1007/jhep09(2010)001[_jhep_ * 09 * ( 2010 ) 001 ] , http://arxiv.org/abs/0810.3285[arxiv:0810.3285 [ hep - th ] ] .
a. brignole , l. e. ibaez , and c. muoz , `` soft supersymmetry - breaking terms from supergravity and superstring models '' , http://arxiv.org/abs/hep-ph/9707209[arxiv:hep-ph/9707209 ] .
k. a. olive and l. velasco - sevilla , `` constraints on supersymmetric flavour models from @xmath330 '' , http://dx.doi.org/10.1088/1126-6708/2008/05/052[_jhep_ * 05 * ( 2008 ) 052 ] , http://arxiv.org/abs/0801.0428[arxiv:0801.0428 [ hep - ph ] ] .
g. dambrosio , g. giudice , g. isidori , and a. strumia , `` minimal flavor violation : an effective field theory approach '' , http://dx.doi.org/10.1016/s0550-3213(02)00836-2[_nucl .
_ * b645 * ( 2002 ) 155187 ] , http://arxiv.org/abs/hep-ph/0207036 [ arxiv : hep - ph/0207036 ] .
s. abel , s. khalil , and o. lebedev , `` additional stringy sources for electric dipole moments '' , http://dx.doi.org/10.1103/physrevlett.89.121601[_phys .
_ * 89 * ( 2002 ) 121601 ] , http://arxiv.org/abs/hep-ph/0112260[arxiv:hep-ph/0112260 ] .
g. g. ross and o. vives , `` yukawa structure , flavor changing , and cp violation in supergravity '' , http://dx.doi.org/10.1103/physrevd.67.095013 [ _ phys .
_ * d67 * ( 2003 ) 095013 ] , http://arxiv.org/abs/hep-ph/0211279[arxiv:hep-ph/0211279 ] .
g. g. ross , l. velasco - sevilla , and o. vives , `` spontaneous cp violation and non - abelian family symmetry in susy '' , http://dx.doi.org/10.1016/j.nuclphysb.2004.05.020[_nucl .
phys . _ * b692 * ( 2004 ) 5082 ] , http://arxiv.org/abs/hep-ph/0401064[arxiv:hep-ph/0401064 ] .
s. antusch , s. f. king , and m. malinsk , `` solving the susy flavour and cp problems with su(3 ) family symmetry '' , http://dx.doi.org/10.1088/1126-6708/2008/06/068[_jhep_ * 06 * ( 2008 ) 068 ] , http://arxiv.org/abs/0708.1282[arxiv:0708.1282 [ hep - ph ] ] .
s. antusch , s. f. king , m. malinsk , and g. g. ross , `` solving the susy flavour and cp problems with non - abelian family symmetry and supergravity '' , http://dx.doi.org/10.1016/j.physletb.2008.11.020[_phys .
* b670 * ( 2009 ) 383389 ] , http://arxiv.org/abs/0807.5047 [ arxiv:0807.5047 [ hep - ph ] ] .
l. calibbi , j. jones - perez , a. masiero , j .- h . park , w. porod , and o. vives , `` fcnc and cp violation observables in a su(3)-flavoured mssm '' , http://dx.doi.org/10.1016/j.nuclphysb.2009.12.029[_nucl .
* b831 * ( 2010 ) 2671 ] , http://arxiv.org/abs/0907.4069 [ arxiv:0907.4069 [ hep - ph ] ] .
l. calibbi , e. j. chun , and l. velasco - sevilla , `` bridging flavour violation and leptogenesis in su(3 ) family models '' , http://dx.doi.org/10.1007/jhep11(2010)090[_jhep_ * 1011 * ( 2010 ) 090 ] , http://arxiv.org/abs/1005.5563[arxiv:1005.5563 [ hep - ph ] ] .
k. kadota , j. kersten , and l. velasco - sevilla , `` supersymmetric musings on the predictivity of family symmetries '' , http://dx.doi.org/10.1103/physrevd.82.085022[_phys .
rev . _ * d82 * ( 2010 ) 085022 ] , http://arxiv.org/abs/1007.1532 [ arxiv:1007.1532 [ hep - ph ] ] .
w. altmannshofer , a. j. buras , s. gori , p. paradisi , and d. m. straub , `` anatomy and phenomenology of fcnc and cpv effects in susy theories '' , http://dx.doi.org/10.1016/j.nuclphysb.2009.12.019[_nucl .
* b830 * ( 2010 ) 1794 ] , http://arxiv.org/abs/0909.1333[arxiv:0909.1333 [ hep - ph ] ] .
particle data group , k. nakamura _ et al .
_ , `` review of particle physics '' , http://dx.doi.org/10.1088/0954-3899/37/7a/075021[_j .
phys . _ * g37 * ( 2010 ) 075021 ] . t. inami and c. lim , ``
effects of superheavy quarks and leptons in low - energy weak processes @xmath331 , @xmath332 and @xmath333 '' , http://dx.doi.org/10.1143/ptp.65.297[_prog .
phys . _ * 65 * ( 1981 ) 297 ] .
s. bertolini , f. borzumati , a. masiero , and g. ridolfi , `` effects of supergravity induced electroweak breaking on rare @xmath334 decays and mixings '' , http://dx.doi.org/10.1016/0550-3213(91)90320-w[_nucl .
_ * b353 * ( 1991 ) 591649 ] .
r. contino and i. scimemi , `` the supersymmetric flavor problem for heavy first - two generation scalars at next - to - leading order '' , http://dx.doi.org/10.1007/s100520050597[_eur .
phys . j. _
* c10 * ( 1999 ) 347356 ] , http://arxiv.org/abs/hep-ph/9809437[arxiv:hep-ph/9809437 ] .
l. j. hall , v. a. kostelecky , and s. raby , `` new flavor violations in supergravity models '' , http://dx.doi.org/10.1016/0550-3213(86)90397-4[_nucl .
* b267 * ( 1986 ) 415 ] .
f. gabbiani , e. gabrielli , a. masiero , and l. silvestrini , `` a complete analysis of fcnc and cp constraints in general susy extensions of the standard model '' , http://dx.doi.org/10.1016/0550-3213(96)00390-2 [ _ nucl .
* b477 * ( 1996 ) 321352 ] , http://arxiv.org/abs/hep-ph/9604387[arxiv:hep-ph/9604387 ] .
j. s. hagelin , s. kelley , and t. tanaka , `` supersymmetric flavor changing neutral currents : exact amplitudes and phenomenological analysis '' , http://dx.doi.org/10.1016/0550-3213(94)90113-9[_nucl .
phys . _ * b415 * ( 1994 ) 293331 ] .
g. eyal , a. masiero , y. nir , and l. silvestrini , `` probing supersymmetric flavor models with @xmath207 '' , http://dx.doi.org/10.1088/1126-6708/1999/11/032[_jhep_ * 9911 * ( 1999 ) 032 ] , http://arxiv.org/abs/hep-ph/9908382 [ arxiv : hep - ph/9908382 ] .
r. barbieri , r. contino , and a. strumia , `` @xmath41 from supersymmetry with nonuniversal @xmath70 terms ? '' , http://dx.doi.org/10.1016/s0550-3213(99)00747-6[_nucl .
phys . _ * b578 * ( 2000 ) 153162 ] , http://arxiv.org/abs/hep-ph/9908255 [ arxiv : hep - ph/9908255 ] .
e. gabrielli , a. masiero , and l. silvestrini , `` flavor changing neutral currents and cp violating processes in generalized supersymmetric theories '' , http://dx.doi.org/10.1016/0370-2693(96)00158-x[_phys . lett . _ * b374 * ( 1996 ) 8086 ] , http://arxiv.org/abs/hep-ph/9509379[arxiv:hep-ph/9509379 ] . s. abel and j. frere , `` could the mssm have no cp violation in the ckm matrix ? '' , http://dx.doi.org/10.1103/physrevd.55.1623[_phys .
_ * d55 * ( 1997 ) 16231629 ] , http://arxiv.org/abs/hep-ph/9608251[arxiv:hep-ph/9608251 ] .
a. masiero and h. murayama , `` can @xmath207 be supersymmetric ? '' , http://dx.doi.org/10.1103/physrevlett.83.907[_phys .
* 83 * ( 1999 ) 907910 ] , http://arxiv.org/abs/hep-ph/9903363 [ arxiv : hep - ph/9903363 ] .
s. khalil , t. kobayashi , and o. vives , `` edm - free supersymmetric cp violation with non - universal soft terms '' , http://dx.doi.org/10.1016/s0550-3213(00)00239-x[_nucl .
_ * b580 * ( 2000 ) 275288 ] , http://arxiv.org/abs/hep-ph/0003086 [ arxiv : hep - ph/0003086 ] .
j. casas and s. dimopoulos , `` stability bounds on flavor violating trilinear soft terms in the mssm '' , http://dx.doi.org/10.1016/0370-2693(96)01000-3[_phys .
lett . _ * b387 * ( 1996 ) 107112 ] , http://arxiv.org/abs/hep-ph/9606237 [ arxiv : hep - ph/9606237 ] . b. s. acharya , f. denef , and r. valandro , `` statistics of m theory vacua '' , http://dx.doi.org/10.1088/1126-6708/2005/06/056[_jhep_ * 0506 * ( 2005 ) 056 ] , http://arxiv.org/abs/hep-th/0502060 [ arxiv : hep - th/0502060 ] . b. c. allanach , `` softsusy : a program for calculating supersymmetric spectra '' , http://dx.doi.org/10.1016/s0010-4655(01)00460-x [ _ comput
* 143 * ( 2002 ) 305331 ] , http://arxiv.org/abs/hep-ph/0104145[arxiv:hep-ph/0104145 ] .
w. porod , `` spheno , a program for calculating supersymmetric spectra , susy particle decays and susy particle production at @xmath335 colliders '' , http://dx.doi.org/10.1016/s0010-4655(03)00222-4[_comput .
* 153 * ( 2003 ) 275315 ] , http://arxiv.org/abs/hep-ph/0301101[arxiv:hep-ph/0301101 ] .
w. porod and f. staub , `` spheno 3.1 : extensions including flavour , cp - phases and models beyond the mssm '' , http://arxiv.org/abs/1104.1573[arxiv:1104.1573 [ hep - ph ] ] . s. khalil and o. lebedev , `` chargino contributions to @xmath40 and @xmath41 '' , http://dx.doi.org/10.1016/s0370-2693(01)00880-2 [ _ phys . lett . _
* b515 * ( 2001 ) 387394 ] , http://arxiv.org/abs/hep-ph/0106023[arxiv:hep-ph/0106023 ] .
y. g. kim , p. ko , and j. s. lee , `` possible new - physics signals in @xmath336 and @xmath337 '' , http://dx.doi.org/10.1016/s0550-3213(99)00048-6[_nucl .
* b544 * ( 1999 ) 6488 ] , http://arxiv.org/abs/hep-ph/9810336 [ arxiv : hep - ph/9810336 ] .
l. conti , c. allton , a. donini , v. gimenez , l. giusti , _ et al .
_ , `` @xmath334-parameters for @xmath276 supersymmetric operators '' , http://dx.doi.org/10.1016/s0920-5632(99)85059-2[_nucl .
suppl . _ * 73 * ( 1999 ) 315317 ] , http://arxiv.org/abs/hep-lat/9809162[arxiv:hep-lat/9809162 ] .
o. gedalia , g. isidori , and g. perez , `` combining direct and indirect kaon cp violation to constrain the warped kk scale '' , http://dx.doi.org/10.1016/j.physletb.2009.10.097[_phys .
lett . _ * b682 * ( 2009 ) 200206 ] , http://arxiv.org/abs/0905.3264 [ arxiv:0905.3264 [ hep - ph ] ] .
s. abel , s. khalil , and o. lebedev , `` edm constraints in supersymmetric theories '' , http://dx.doi.org/10.1016/s0550-3213(01)00233-4 [ _ nucl .
phys . _ * b606 * ( 2001 ) 151182 ] , http://arxiv.org/abs/hep-ph/0103320[arxiv:hep-ph/0103320 ] .
i. masina and c. a. savoy , `` sleptonarium : constraints on the cp and flavor pattern of scalar lepton masses '' , http://dx.doi.org/10.1016/s0550-3213(03)00294-3[_nucl .
_ * b661 * ( 2003 ) 365393 ] , http://arxiv.org/abs/hep-ph/0211283 [ arxiv : hep - ph/0211283 ] .
t. moroi , `` the muon anomalous magnetic dipole moment in the minimal supersymmetric standard model '' , http://dx.doi.org/10.1103/physrevd.53.6565 , 10.1103/physrevd.56.4424[_phys .
rev . _ * d53 * ( 1996 ) 65656575 ] , http://arxiv.org/abs/hep-ph/9512396[arxiv:hep-ph/9512396 ] . muon @xmath339 collaboration , g. bennett _ et al . _ ,
`` final report of the muon e821 anomalous magnetic moment measurement at bnl '' , http://dx.doi.org/10.1103/physrevd.73.072003[_phys . rev . _ * d73 * ( 2006 ) 072003 ] , http://arxiv.org/abs/hep-ex/0602035 [ arxiv : hep - ex/0602035 ] .
f. borzumati , c. greub , t. hurth , and d. wyler , `` gluino contribution to radiative b decays : organization of qcd corrections and leading order results '' , http://dx.doi.org/10.1103/physrevd.62.075005[_phys .
rev . _ * d62 * ( 2000 ) 075005 ] , http://arxiv.org/abs/hep-ph/9911245[arxiv:hep-ph/9911245 ] .
d. wyler and f. borzumati , `` gluino contribution to radiative b decays : new operators , organization of qcd corrections and leading order results '' , http://arxiv.org/abs/hep-ph/0104046[arxiv:hep-ph/0104046 ] . p. gambino and m. misiak , `` quark mass effects in @xmath340 '' , http://dx.doi.org/10.1016/s0550-3213(01)00347-9[_nucl .
phys . _ * b611 * ( 2001 ) 338366 ] , http://arxiv.org/abs/hep-ph/0104034 [ arxiv : hep - ph/0104034 ] . t. hurth , e. lunghi , and w. porod , `` untagged @xmath341 cp asymmetry as a probe for new physics '' , http://dx.doi.org/10.1016/j.nuclphysb.2004.10.024[_nucl
* b704 * ( 2005 ) 5674 ] , http://arxiv.org/abs/hep-ph/0312260 [ arxiv : hep - ph/0312260 ] .
m. ciuchini , e. franco , d. guadagnoli , v. lubicz , m. pierini , _ et al .
_ , `` @xmath342 - @xmath343 mixing and new physics : general considerations and constraints on the mssm '' , http://dx.doi.org/10.1016/j.physletb.2007.08.055[_phys .
_ * b655 * ( 2007 ) 162166 ] , http://arxiv.org/abs/hep-ph/0703204 [ arxiv : hep - ph/0703204 ] .
p. paradisi , `` constraints on susy lepton flavor violation by rare processes '' , http://dx.doi.org/10.1088/1126-6708/2005/10/006 [ _ jhep _ * 0510 * ( 2005 ) 006 ] , http://arxiv.org/abs/hep-ph/0505046[arxiv:hep-ph/0505046 ] .
k. agashe and m. graesser , `` supersymmetry breaking and the supersymmetric flavor problem : an analysis of decoupling the first two generation scalars '' , http://dx.doi.org/10.1103/physrevd.59.015007[_phys .
rev . _ * d59 * ( 1999 ) 015007 ] , http://arxiv.org/abs/hep-ph/9801446[arxiv:hep-ph/9801446 ] . j. a. bagger , k. t. matchev , and r .- j .
zhang , `` qcd corrections to flavor - changing neutral currents in the supersymmetric standard model '' , http://dx.doi.org/10.1016/s0370-2693(97)00920-9[_phys .
lett . _ * b412 * ( 1997 ) 7785 ] , http://arxiv.org/abs/hep-ph/9707225[arxiv:hep-ph/9707225 ] .
j. ellis , r. n. hodgkinson , j. s. lee , and a. pilaftsis , `` flavour geometry and effective yukawa couplings in the mssm '' , http://dx.doi.org/10.1007/jhep02(2010)016[_jhep_ * 1002 * ( 2010 ) 016 ] , http://arxiv.org/abs/0911.3611[arxiv:0911.3611 [ hep - ph ] ] .
k. blum , y. grossman , y. nir , and g. perez , `` combining @xmath51 - @xmath52 mixing and @xmath44 - @xmath45 mixing to constrain the flavor structure of new physics '' , http://dx.doi.org/10.1103/physrevlett.102.211802 [ _ phys .
* 102 * ( 2009 ) 211802 ] , http://arxiv.org/abs/0903.2118[arxiv:0903.2118 [ hep - ph ] ] . | in recent years it has been learned that scalar superpartner masses and trilinear couplings should _ both _ generically be larger than about 20 tev at the short distance string scale if our world is described by a compactified string or m - theory with supersymmetry breaking and stabilized moduli @xcite .
here we study implications of this , somewhat generally and also in detail for a particular realization ( compactification of m - theory on a @xmath0 manifold ) where there is significant knowledge of the superpotential and gauge kinetic function , and a light gluino . in a certain sense this yields an ultraviolet completion of minimal flavour violation .
flavour violation stems from off - diagonal and non - universal diagonal elements of scalar mass matrices and trilinear couplings , and from renormalization group running .
we also examine stability bounds on the scalar potential .
while heavy scalars alone do not guarantee the absence of flavour problems , our studies show that models with heavy scalars and light gluinos can be free from such problems . * flavour issues for string - motivated heavy scalar spectra with a low gluino mass : the @xmath1-mssm case + * * k. kadota@xmath2 , g. kane @xmath3 , j. kersten@xmath4 , and l. velasco - sevilla@xmath5 * + _ @xmath3 michigan center for theoretical physics , university of michigan , + ann arbor , mi , 48104 , usa + @xmath6 department of physics , nagoya university , nagoya 464 - 8602 , japan + @xmath4 university of hamburg , ii .
institute for theoretical physics , + luruper chaussee 149 , 22761 hamburg , germany + @xmath5 cinvestav - ipn , apdo .
postal 14 - 740 , 07000 , mxico d.f .
, mxico . _ |
to derive the form of the free energy given in eqs .
( 2 ) and ( 4 ) in the main text , we perform a hubbard - stratonovich ( hs ) decoupling thereby obtaining the electron - mediated interactions between the magnetic order parameters .
formally the hs decoupling relies on inserting unity in the partition function , where unity , in the present case , is given by @xmath101 \nonumber \\ & & \exp\bigg[-\int_{q}\big({\mathbf{m}}_{\mu\nu}^{x}(q)\left(u^{-1}\right)_{\rho\lambda}^{\mu\nu}{\mathbf{m}}_{\rho\lambda}^{x}(-q ) \nonumber \\ & & \qquad \qquad \ + { \mathbf{m}}_{\mu\nu}^{y}(q)\left(u^{-1}\right)_{\rho\lambda}^{\mu\nu}{\mathbf{m}}_{\rho\lambda}^{y}(-q)\big)\bigg]\,,\end{aligned}\ ] ] and @xmath102 $ ] is chosen such that the path - integral evaluates to unity and @xmath43 ( @xmath103 being a bosonic matsubara frequency ) .
the electrons are then integrated out resulting in an effective action for the magnetic order @xmath104 & = & \sum_{i}\int_{q}{\mathbf{m}}_{\mu\nu}^{i}(q)\left(u^{-1}\right)_{\rho\lambda}^{\mu\nu}{\mathbf{m}}_{\rho\lambda}^{i}(-q ) \nonumber \\ & - & \text{tr}\ln\left[\mathbb{g}_{\mu\nu}^{0}(k)^{-1}-\mathcal{v}_{\mu\nu}(q)\right]\,,\end{aligned}\ ] ] where @xmath35 , @xmath105 and @xmath106 are orbital indices , @xmath107 , @xmath108 is the fermionic matsubara frequency , and the trace is over all external indices ( the spin indices have been suppressed , the green s function is diagonal in spin ) .
@xmath109 is the matrix green s function , obtained from the first term in eq .
( 1 ) of the main text , and @xmath110 originates from the coupling between the magnetic order parameters and the electrons , the second term . in the basis @xmath111
these are given by the matrices @xmath112 where each element of the matrices should be understood as an @xmath113 matrix in orbital space , with the green function being @xmath114 where @xmath115 refers to band basis and @xmath116 refer to orbital basis .
expanding the trace - log to fourth order in the magnetic order parameters and applying the pauli matrix identity @xmath117 yields the magnetic free energy as written in eqs .
( 2 ) and ( 4 ) of the main text , with the fourth order coefficients @xmath118 where repeated orbital indices are not summed . here
all the green functions are implicit functions of @xmath119 and @xmath120 and @xmath121 .
preparing for an additional hs - decoupling we introduce two bosonic fields @xmath122 and @xmath123 with the partition function @xmath124\,,\end{aligned}\ ] ] with integration measures chosen appropriately such that @xmath125 . by performing the shifts @xmath126 the terms quartic in @xmath127
cancel accordingly .
following the standard procedure we introduce a field ( @xmath128 ) conjugate to @xmath129 and define @xmath130 .
the resulting action is then @xmath131 & = & \sum_{q , i = x , y } \left(r^{\mu\nu}_{i}(q ) + \psi_{\mu\nu}\right ) { \mathbf{m}}^{i}_{\mu}\cdot{\mathbf{m}}^{i}_{\nu } \nonumber \\ & - & \frac{1}{2}\left ( u^{\rho\nu\eta\mu } \right)^{-1 } \psi_{\rho\nu}\psi_{\eta\mu } \nonumber \\ & + & \frac{1}{2 } \left ( g^{\rho\nu\eta\mu } \right)^{-1 } \left ( \widetilde{\phi}_{\rho\nu } - h_{\eta\mu } \right ) \left ( \widetilde{\phi}_{\eta\mu } - h_{\eta\mu } \right ) \nonumber \\ & - & \widetilde{\phi}_{\rho\nu}\left ( { \mathbf{m}}^{x}_{\rho}\cdot { \mathbf{m}}^{x}_{\nu } - { \mathbf{m}}^{y}_{\rho}\cdot{\mathbf{m}}^{y}_{\nu } \right)\,.\end{aligned}\ ] ] here @xmath132 and @xmath133 .
it is now straightforward to compute the nematic susceptibility : @xmath134 where we used the fact that @xmath135 as we are above the nematic instability . to continue
we note that @xmath136 where the free energy is @xmath137 obtained by integrating out the magnetic degrees of freedom and taking the large @xmath138 limit .
we find the effective action @xmath139 & = & \frac{1}{2}\left ( g^{\rho\nu\eta\mu } \right)^{-1}\phi_{\rho\nu}\phi_{\eta\mu } \nonumber \\ & + & \frac{1}{2 } \text{tr } \ \ln \big [ \chi^{-1}_{\iota\kappa , y}\chi^{-1}_{\kappa\lambda , x } - \phi_{\iota\kappa}\phi_{\kappa\lambda } \nonumber \\ & + & \chi^{-1}_{\iota\kappa , y}\phi_{\kappa\lambda } - \phi_{\iota\kappa}\chi^{-1}_{\kappa\lambda , x } \big]\,,\end{aligned}\ ] ] where we have ignored the gaussian fluctuations @xmath122 and @xmath140 .
finally @xmath141 and after some manipulations we arrive at the expression given in the text for the nematic susceptibility .
in this section we illustrate the frequency dependence of the magnetic propagator at various temperatures for representative filling factors of the @xmath84 phase diagram ( fig .
4(a ) of the main text ) . because the magnetic propagator peaks at @xmath142 , we focus on @xmath143 . for @xmath144 as we approach the instability ( at @xmath145 mev ) , the frequency dependence of the propagator @xmath146 has the form shown in fig . [
fig : frequency_dependence_n=00003d00003d590 ] , where the bosonic matsubara frequency is given by @xmath147 .
the gray area denotes the region included in the cut - off @xmath148 , and the dotted line indicates @xmath149 . the plots in fig .
[ fig : frequency_dependence_n=00003d00003d590 ] justify the statement made in the main text that near a finite - temperature magnetic transition , one can safely neglect the higher frequency contributions .
frequency dependence of the magnetic propagator @xmath146 for @xmath144 at different temperatures .
the parameters used are quoted in the main text .
the magnetic instability takes place at @xmath150 mev . from ( a )
we see that the contribution to the bare nematic susceptibility comes mostly from the zero frequency part of the magnetic susceptibility . ] to illustrate the importance of including high frequency contributions in the case where magnetic order is absent , in fig .
[ fig : frequency_dependence_n=00003d00003d604 ] we also plot the frequency dependence of the magnetic propagator for @xmath151 .
it is clear that the peak is broadened , implying that it is no longer justified to ignore the contributions originating from finite frequencies .
frequency dependence of the magnetic propagator @xmath146 for @xmath151 at different temperatures .
the parameters used are quoted in the main text . as is evident in ( a )
, the peak broadens as zero temperature is approached . however , even at higher temperatures , shown in ( b ) and ( c ) , finite matsubara frequencies provide considerable contributions to the bare nematic susceptibility . ]
ferro - orbital order is characterized by the breaking of the degeneracy between the @xmath11 and @xmath12 orbitals . in the itinerant framework
this is seen by an inequivalent occupation of the two orbitals , i.e. @xmath152 . defining @xmath153 as in the main text , the ferro - orbital susceptibility is given by @xmath154 . using the definition of @xmath155
, we find that this is nothing but a linear combination of specific components of the charge susceptibility , @xmath156 . in the standard rpa approach ,
the full expression is @xcite @xmath157 where the rpa charge susceptibility is given by the usual expression @xcite @xmath158^{-1}\right)_{\rho\gamma}^{\mu\delta}(\chi_{0})_{\gamma\lambda}^{\delta\nu}\,,\label{eq : charge_rpa}\end{aligned}\ ] ] where @xmath159 is the standard particle - hole bubble @xmath160 and @xmath161 is the interaction matrix in the charge channel .
the latter differs from the interaction in the sdw channel and is given by ( @xmath162 ) @xmath163 we note that , due to the implicit summation over repeated indices in eq .
( [ eq : charge_rpa ] ) , all orbitals contribute to the rpa orbital order susceptibility .
the static part of eq .
( [ eq : chi_oo ] ) at @xmath30 is the quantity plotted in fig .
4(b ) in the main text .
m. yi , d. lu , j .- h .
chu , j. g. analytis , a. p. sorini , a. f. kemper , b. moritz , s .- k .
mo , r. g. moore , m. hashimoto , w. s. lee , z. hussain , t. p. devereaux , i. r. fisher , z .- x .
shen , proc .
2011 * 108 * , 6878 ( 2011 ) .
k. ishida , y. nakai and h. hosono , j. phys .
japan * 78 * , 062001 ( 2009 ) ; d. c. johnston , adv . phys .
* 59 * , 803 ( 2010 ) ; j. paglione and r. l. greene , nature phys .
* 6 * , 645 ( 2010 ) ; p. c. canfield and s. l. budko , annu .
* 1 * , 27 ( 2010 ) ; h. h. wen and s. li , annu .
cond . mat .
* 2 * , 121 ( 2011 ) . | nematic order resulting from the partial melting of density - waves has been proposed as the mechanism to explain nematicity in iron - based superconductors . an outstanding question , however , is whether the microscopic electronic model for these systems the multi - orbital hubbard model displays such an ordered state as its leading instability .
in contrast to usual electronic instabilities , such as magnetic and charge order , this fluctuation - driven phenomenon can not be captured by the standard rpa method . here ,
by including fluctuations beyond rpa in the multi - orbital hubbard model , we derive its nematic susceptibility and contrast it with its ferro - orbital order susceptibility , showing that its leading instability is the spin - driven nematic phase .
our results also demonstrate the primary role played by the @xmath0 orbital in driving the nematic transition , and reveal that high - energy magnetic fluctuations are essential to stabilize nematic order in the absence of magnetic order .
the elucidation of electronic ising - nematic order @xcite the state in which electronic degrees of freedom spontaneously lower the point - group symmetry of the system has become an important problem in unconventional superconductors @xcite . in both pnictides
@xcite and cuprates @xcite , the experimentally observed nematic order has been proposed to arise from the partial melting of an underlying spin density - wave ( sdw ) @xcite or charge density - wave ( cdw ) @xcite stripe - order .
this mechanism is based on robust symmetry considerations .
consider for concreteness the stripe sdw case : the ground state has an @xmath1 degeneracy , with @xmath2 denoting the direction of the magnetic order parameter in spin space , and @xmath3 denoting the selection of the sdw ordering vector @xmath4 or @xmath5 ( in the cdw case , the system has an @xmath6 degeneracy ) .
fluctuations in layered systems suppress the continuous ( @xmath2 or @xmath7 ) and the discrete ( @xmath3 ) symmetries differently , favoring an intermediate regime in which only the @xmath3 symmetry is broken @xcite . because the @xmath3 symmetry distinguishes between two ordering vectors related by a @xmath8 rotation , its breaking implies a tetragonal - to - orthorhombic transition , and therefore nematic order .
although this mechanism for spin - driven ( or charge - driven ) nematic order has been established in simplified low - energy models for pnictides @xcite and cuprates @xcite , it remains hotly debated whether more realistic microscopic models display nematic order as the leading electronic instability .
the cuprates , a sensible microscopic model is the single - band hubbard model , whose phase diagram has been reported to display nematic correlations in the strong - coupling regime @xcite .
for the pnictides , due to the @xmath9 configuration of fe and to the small crystal field splittings , a five - orbital hubbard model , including hund s rule interactions , is a more appropriate starting point @xcite .
furthermore , because many pnictides display metallic behavior , a weak - coupling analysis of this intricate model can reveal important information about the underlying physics of these materials .
indeed , conventional rpa approaches have been employed to study the onset of sdw , cdw , and ferromagnetism . however , in contrast to these usual electronic instabilities , the standard rpa approach does not capture the nematic instability even qualitatively , as we show below , making it difficult to assess whether the realistic multi - orbital hubbard model has a tendency towards nematic order . in this letter
, we extend the standard rpa approach and derive the nematic susceptibility of an arbitrary multi - orbital hubbard model .
the fluctuations included in this formalism arise solely from the non - interacting part of the hamiltonian , such that interactions are treated at the same order as in the typical rpa method .
we apply this formalism to the case of sdw - driven nematicity in iron pnictides , and establish that the leading instability of the five - orbital interacting model is a spin - driven nematic phase for a wide range of parameters . in general , we find that nematic order exists in a narrow @xmath10 range above the magnetic transition line , in agreement with experiments in the pnictides @xcite .
however , magnetic fluctuations at higher energies can induce a sizable splitting between the two transitions , particularly in the regime where the sdw transition is suppressed to zero .
we propose that this effect may be relevant to understanding the unusual nematic phase of fese @xcite .
previously , the investigation of the multi - orbital hubbard model in ref .
@xcite revealed the importance of the orbital content of the fermi surface in the low - energy spin - nematic model of the pnictides . here
, we find from the orbitally - resolved nematic susceptibility that whereas the @xmath11 , @xmath12 , and @xmath0 orbitals contribute almost equally to the sdw instability , the @xmath0 orbital plays a stronger role in driving the nematic instability .
finally , we compare the nematic susceptibility with the rpa - derived ferro - orbital order susceptibility .
work provides a promising route to search for nematicity in different compounds , as it is compatible with _ ab initio _ approaches and also with methods that include the effects of moderate interactions , such as lda+dmft @xcite .
our starting point is the multi - orbital hubbard model with onsite interactions @xcite .
the non - interacting part is given by @xmath13 , where @xmath14 creates an electron with momentum @xmath15 and spin @xmath16 at orbital @xmath17 and the hopping parameters @xmath18 are determined from tight - binding fits to _ ab initio _ calculations ( sums over spin and momentum indices are left implicit ) .
the four onsite interaction terms correspond to the intra - orbital hubbard term , @xmath19 , the inter - orbital hubbard term , @xmath20 , the hund s rule coupling , @xmath21 , and the pair - hopping term @xmath22 .
these coefficients are related by @xmath23 and @xmath24 .
previous approaches considered the nematic susceptibility of a spin - fermion model @xcite ; here , we will focus on the hubbard model within rpa .
the mechanism in which nematic order arises from the partial melting of an sdw or a cdw requires fluctuations at two momenta related by @xmath8 , in general @xmath25 and @xmath26 , with integer @xmath27 .
although our formalism can be extended in a straightforward way to arbitrary @xmath27 , hereafter we focus on @xmath28 . to make contact with the pnictides
, we consider the sdw channel .
performing a hartree - fock decoupling of @xmath29 in both the @xmath30 charge channel and the @xmath31 sdw channel : @xmath32 where @xmath33 incorporates the changes in the mean - field densities and @xmath34 are the sdw order parameters with @xmath35 .
the interaction matrix @xmath36 is @xmath37 and @xmath38 .
we consider only intra - orbital magnetism , since previous hartree - fock calculations revealed that in the ground state the intra - orbital sdw order parameters are the dominant ones @xcite . in the standard rpa approach for the sdw instability , the electronic degrees of freedom are integrated out , yielding the quadratic magnetic free energy : @xmath39 & = & \sum_{q , i = x , y}\left[\chi_{i}^{\mu\nu}(q)\right]^{-1}{\mathbf{m}}_{q,\mu}^{i}\cdot{\mathbf{m}}_{-q,\nu}^{i}\,,\label{eq : second_order_free_energy}\end{aligned}\ ] ] with the magnetic propagator @xmath40 @xmath41^{-1}\,,\label{eq : mag_susc}\end{aligned}\ ] ] where @xmath42 is the green s function in orbital basis , @xmath43 , @xmath44 , and @xmath45 is the matsubara frequency .
the rpa magnetic susceptibility @xmath46 is proportional to and diverges at the same temperature as the magnetic propagator @xmath40 . note that the tetragonal symmetry of the system implies that a peak of @xmath40 at @xmath4 will be accompanied by an equal peak at @xmath5 .
therefore , at this order in perturbation theory , the system does not distinguish the case in which either @xmath47 or @xmath48 is selected ( single-*q * order ) from the case in which both are selected ( double-*q * order ) , i.e. the standard rpa approach is blind to nematicity . to remedy this problem
, we go beyond the second - order expansion of the free energy and calculate the quartic - order terms : @xmath49 & = & \frac{1}{2}u^{\rho\nu\eta\mu}\left({\mathbf{m}}_{\rho}^{x}\cdot{\mathbf{m}}_{\nu}^{x}+{\mathbf{m}}_{\rho}^{y}\cdot{\mathbf{m}}_{\nu}^{y}\right)\nonumber \\ & & \qquad\qquad\times\left({\mathbf{m}}_{\eta}^{x}\cdot{\mathbf{m}}_{\mu}^{x}+{\mathbf{m}}_{\eta}^{y}\cdot{\mathbf{m}}_{\mu}^{y}\right)\nonumber \\ & - & \frac{1}{2}g^{\rho\nu\eta\mu}\left({\mathbf{m}}_{\rho}^{x}\cdot{\mathbf{m}}_{\nu}^{x}-{\mathbf{m}}_{\rho}^{y}\cdot{\mathbf{m}}_{\nu}^{y}\right)\nonumber \\ & & \qquad\qquad\times\left({\mathbf{m}}_{\eta}^{x}\cdot{\mathbf{m}}_{\mu}^{x}-{\mathbf{m}}_{\eta}^{y}\cdot{\mathbf{m}}_{\mu}^{y}\right)\nonumber \\ & + & 2w^{\rho\nu\eta\mu}\left({\mathbf{m}}_{\rho}^{x}\cdot{\mathbf{m}}_{\nu}^{y}\right)\left({\mathbf{m}}_{\eta}^{x}\cdot{\mathbf{m}}_{\mu}^{y}\right)\,,\label{eq : fourth_order_free_energy}\end{aligned}\ ] ] the quartic coefficients , whose expressions are shown explicitly in appendix [ app : fourth_order_coeff ] , depend only on the non - interacting green s functions . although interactions can also contribute to them , as shown in refs .
@xcite , within the rpa approach these contributions are sub - leading and can be neglected . the most relevant coefficient for the nematic instability is @xmath50 , whose term distinguishes between single-*q * and double-*q * order .
specifically , a hubbard - stratonovich transformation of this term reveals the nematic order parameter @xmath51 , a rank-2 tensor in orbital space that breaks the tetragonal symmetry of the system by making @xmath52 .
the term with coefficient @xmath53 is related to gaussian magnetic fluctuations in both sdw channels , while the term with coefficient @xmath54 mainly distinguishes between the two types of double-*q * order @xcite .
eq .
( [ eq : fourth_order_free_energy ] ) is the multi - orbital generalization of the magnetic free energy previously obtained in effective low - energy models in the band basis , where the coefficient @xmath55 becomes a scalar @xcite .
( color online ) normal - state fermi surface based on the parameters of ikeda _ et al_. @xcite .
the colors indicate the dominant orbital contribution . ]
it is now possible to compute the static nematic susceptibility @xmath56 in the paramagnetic phase ( see appendix [ app : nem_susc ] for details of the derivation ) : @xmath57 the orbitally - resolved nematic susceptibility @xmath58 is a rank-4 tensor that generalizes the scalar nematic susceptibility derived previously for effective low - energy models @xcite .
the impact of the magnetic fluctuations encoded in the coefficient @xmath59 is clear : if this term were absent , then the ( bare ) nematic susceptibility would be merely a higher - order convolution of the magnetic propagator , @xmath60 , and therefore diverge at the same @xmath10 as the sdw susceptibility . to establish whether the nematic susceptibility diverges already in the paramagnetic phase , one needs to compute its leading eigenvalue @xmath61 from @xmath62 , with @xmath63 .
the structure of the corresponding eigen - matrix @xmath64 reveals which orbitals promote the nematic instability , and which orbitals favor a double-*q * structure with no underlying nematicity .
we note that in principle the gaussian fluctuations associated with the term with coefficient @xmath53 can also renormalize the magnetic propagator @xmath65 .
however , because this effect merely renormalizes the sdw transition temperature , we do not include it hereafter .
( color online ) the largest eigenvalues @xmath66 of * ( a ) * the bare nematic susceptibility @xmath60 , the @xmath67 magnetic propagator @xmath68 , and * ( b ) * the full nematic susceptibility @xmath58 as a function of @xmath10 for the case @xmath69 .
the inset in * ( a ) * shows the upturn of the magnetic susceptibility as it diverges . ]
equation ( [ eq : chi_nem ] ) is the rpa - generalized nematic susceptibility , which can be compared on equal - footing with other rpa instabilities of a weakly - interacting system described by a multi - orbital hubbard model .
we apply this formalism to a five - orbital model for the iron - based superconductors and contrast the nematic susceptibility to the ferro - orbital rpa susceptibility .
the hopping parameters are those from ref .
@xcite , whereas the interactions are set to @xmath70 ev and @xmath71 @xcite .
small changes in these parameters do not alter our main results .
the fermi surface for the occupation number @xmath69 is presented in fig .
[ fig : fermi_surface ] , consisting of three hole pockets at the center and the corner of the brillouin zone , and two electron pockets at the borders of the brillouin zone . that the @xmath0 hole pocket at @xmath72 is not present in all materials , as it depends on the fe - as distance @xcite
.
we evaluate eqs .
( [ eq : mag_susc ] ) and ( [ eq : chi_nem ] ) numerically as functions of @xmath10 for various values of the occupation number @xmath27 . consider first @xmath69 : in fig .
[ fig : nematic_and_magnetic_instabilities](a ) , we plot the @xmath10 dependence of the largest eigenvalue of the static magnetic propagator @xmath73 as well as the largest eigenvalue of the _ bare _ nematic susceptibility @xmath60 . despite having different @xmath10 dependencies ,
both eigenvalues diverge at the same temperature @xmath74 , confirming our assertion that the standard rpa is blind to the nematic instability . in fig .
[ fig : nematic_and_magnetic_instabilities](b ) , we plot the largest eigenvalue of the _ full _ nematic susceptibility @xmath58 , as given by eq .
( [ eq : chi_nem ] ) . clearly , the eigenvalue diverges at @xmath75 : this is exactly the nematic transition temperature @xmath76 . interestingly , our results reveal a relatively small splitting between @xmath76 and @xmath74 , with @xmath77 , which resembles the small @xmath10-range in which a nematic - paramagnetic phase is observed experimentally in the iron pnictides @xcite .
we caution , however , that this value should be understood as an upper boundary for the splitting between the nematic and the actual magnetic transition , since @xmath78 calculated inside the nematic state is generally larger than @xmath74 calculated in the tetragonal state .
furthermore , the value for @xmath74 obtained via rpa overestimates the actual transition temperature due to the absence of gaussian fluctuations , as discussed above .
( color online ) color plot of the normalized elements of the eigen - matrix @xmath64 corresponding to the leading eigenvalue of the bare ( left ) and of the full ( right ) nematic susceptibilities .
the dominant contributions arise from the @xmath11 , @xmath12 , and @xmath0 orbital , with the @xmath0 being the most important for nematicity . ]
while the largest eigenvalue @xmath61 determines @xmath76 , the structure of the corresponding @xmath79 eigen - matrix @xmath64 reveals the orbital - resolved nematic order parameter driving the transition , since @xmath80 . in fig .
[ fig : nematic_and_magnetic_orbital_resolution ] we plot the normalized elements of the leading eigen - matrix @xmath64 for both the full and the bare nematic susceptibility which , as shown above , contains information only about the magnetic instability . in both cases ,
the dominant processes involve the @xmath11 , @xmath12 , and @xmath0 orbitals .
there is however one important difference : the relative weight of the @xmath0 orbital is larger for @xmath58 than for @xmath60 , i.e. while the three orbitals seem to contribute equally to drive the magnetic instability , the @xmath0 orbital plays a more important role in driving the nematic instability .
we interpret this in terms of the nesting properties of the orbital content of the fermi surface in fig .
[ fig : fermi_surface ] : while the @xmath0 hole - pocket at @xmath72 can form a single-*q * sdw by combining with either the @xmath81 or @xmath82 electron - pockets , since both have @xmath0 spectral weight , the two @xmath11/@xmath12 hole - pockets at @xmath83 can form a double-*q * sdw by combining with both the @xmath81 and @xmath82 pockets , since they have @xmath12 and @xmath11 spectral weight , respectively . having analyzed the @xmath69 case , we present in fig . [
fig : phase_diagram](a ) the complete @xmath84 phase diagram for the magnetic and nematic transitions .
we restrict our analysis to @xmath85 , since below this value we find incommensurate magnetic order .
accounting for the nematic transition in this regime requires changes in the formalism beyond the scope of this work .
note that , in contrast to experiments , @xmath74 is not peaked at @xmath69 .
this is likely due to the absence of disorder effects introduced by doping , which are known to suppress @xmath74 @xcite .
most importantly , across the entire phase diagram the nematic transition line tracks closely the magnetic transition line , in agreement with the phase diagrams of the iron pnictides .
( color online ) * ( a ) * occupation number - temperature @xmath84 phase diagram for the bare magnetic and nematic phase transitions , evidencing the narrow region displaying nematic - paramagnetic order .
the solid @xmath76 line takes into account only the contribution from low - energy ( @xmath86 ) magnetic fluctuations , whereas the dashed line includes contributions from higher energies ( @xmath87 ev ) . for @xmath88
, an incommensurate magnetic order appears . *
( b ) * ferro - orbital order susceptibility @xmath89 a function of @xmath10 for various values of the occupation number @xmath27 .
in contrast to the nematic susceptibility shown in fig .
[ fig : nematic_and_magnetic_instabilities ] , @xmath89 is nearly featureless and @xmath10-independent at low energies . ]
an important issue in obtaining this phase diagram is that , as shown in eq .
( [ eq : chi_bare_nem ] ) , the computation of the nematic susceptibility requires summing the magnetic fluctuations not only over the entire brillouin zone , but also over energy ( i.e. over matsubara frequencies ) .
although the propagator @xmath90 is strongly peaked at @xmath86 ( see appendix [ app : freq_dep_mag_susc ] ) , within rpa it saturates to a finite value for large energies [ see eq .
( [ eq : mag_susc ] ) ] , requiring a frequency cutoff @xmath91 . near a finite-@xmath10 magnetic transition , due to the very sharp peak in @xmath92
, it is reasonable to take only the @xmath86 contribution the low - energy magnetic fluctuations resulting in the solid line of fig .
[ fig : phase_diagram ] . however , near the region where @xmath93 , ignoring the high - energy magnetic fluctuations ( @xmath94 ) is not justified . to address this problem
, we introduce a cutoff @xmath95 ev , at which the propagator reaches values close to its saturation value , as shown in appendix [ app : freq_dep_mag_susc ] .
the corresponding nematic transition line is shown as a dashed line in fig .
[ fig : phase_diagram ] . near the regime where the magnetic transition takes place at finite @xmath10 ,
the only effect of the cutoff is to increase the nematic transition temperature , as expected .
however , near the regime where @xmath93 , the nematic transition is stabilized even in the absence of long - range magnetic order .
although the precise value of @xmath76 depends on the cutoff value , the main result is that higher - energy magnetic fluctuations are essential to promote nematic order without magnetic order . in this regard ,
it is interesting to note that , in fese , the only parent material in which nematic order is observed in the absence of magnetic order , nmr measurements find no evidence for low - energy magnetic fluctuations @xcite , whereas neutron scattering reports sizable fluctuations at modest energy values @xcite .
a remaining question is whether or not the spin - driven nematic instability is the leading instability of the system . in particular ,
an ongoing debate @xcite concerning iron - based materials is whether ferro - orbital order , signaled by an unequal occupation of the @xmath11 and @xmath12 orbitals , @xmath96 , could drive the nematic transition , instead of the spin - driven mechanism explored above . to investigate this issue
, we calculate the @xmath30 static component of the rpa orbital order susceptibility , @xmath97 for the multi - orbital hubbard model @xcite , of which a brief derivation is included in appendix [ app : ferro_orb_order_susc ] . as shown in fig .
[ fig : phase_diagram](b ) , our results reveal a nearly @xmath10-independent @xmath89 for the doping range and interactions investigated .
this is not unexpected , since for reasonable values of @xmath98 and @xmath99 , there is no attraction in the rpa charge channel .
therefore , within rpa , ferro - orbital order is unable to drive the nematic instability .
of course , once the coupling to magnetic fluctuations is included , which requires going beyond rpa , @xmath89 will diverge at the same @xmath10 as @xmath100 @xcite . in this regard , by effectively decoupling these two channels , rpa provides an interesting route to investigate which instability is the leading one at least for weak or moderate interactions .
in summary , we developed an appropriate extension of the rpa approach to obtain the orbital - resolved spin - driven nematic susceptibility of an arbitrary multi - orbital hubbard model .
application to the case of iron - based superconductors reveals that the leading instability of the system is an interaction - driven nematic phase .
the @xmath0 orbital plays a leading role in promoting the nematic instability , and higher - energy magnetic fluctuations are essential to stabilize nematic order in the absence of long - range magnetic order .
comparison with other rpa susceptibilities reveals that the nematic and magnetic transitions follow each other closely , and that the ferro - orbital susceptibility does not diverge on its own .
more generally , our formalism can also be combined with first - principle approaches to search for other materials that may display electronic nematicity .
furthermore , because interactions appear only in the determination of the magnetic propagator , eq .
( [ eq : mag_susc ] ) , this formalism can be combined with other approaches that specifically include moderate electronic interactions , such as dft+u or lda+dmft @xcite .
the authors are grateful to a. v. chubukov , l. fanfarillo , m. schtt , p. orth , and b. valenzuela for discussions .
mhc and bma acknowledge financial support from a lunbeckfond fellowship ( grant a9318 ) .
rmf and jk are supported by the u.s .
department of energy , office of science , basic energy sciences , under award number de - sc0012336 . |
a present day main motivation for the extra - terrestrial neutrino astronomy is to obtain first evidence of tau neutrinos from the cosmos around us above the relatively well known atmospheric neutrino background.@xcite the tau neutrinos are an unavoidable consequence of neutrino flavor mixing as suggested by the high statistics super - kamiokande detector ( skk).@xcite a recent skk analysis of the atmospheric neutrino data imply the following range of neutrino mixing parameters @xmath1 this is a @xmath2 range with the best fit values approximately given by @xmath3 and @xmath4 respectively .
this range of neutrino mixing parameters results in purely two flavor oscillation explanation of the zenith angle dependence of the atmospheric mu neutrino deficit , along with another indication .
the tau neutrinos as a result of these @xmath5 oscillations are so far however identified on statistical basis only ( rather than on event by event basis).@xcite on the other hand , the total number of observed non tau neutrinos are by now greater than @xmath6 from various detectors ranging in energy between @xmath7 gev to @xmath8 gev.@xcite thus , it is of some interest to estimate the tau neutrino flux from the earth atmosphere as well as from the nearby astrophysical sites such as our galactic plane to provide a more complete basis for the hypothesis of @xmath5 oscillations .
the neutrino oscillation probability in the two neutrino flavor approximation is @xmath9 here @xmath10 can be estimated using @xmath11 the @xmath10 is the distance between the detector and the height at which the atmospheric mu neutrinos are produced .
the @xmath12 km is the earth radius , and @xmath13 km is the mean altitude at which the atmospheric mu neutrinos are produced . in general
, @xmath14 is not only a function of the zenith angle @xmath15 , the neutrino flavor but also the neutrino energy.@xcite also @xmath16 in usual notation .
this paper is organized as follows . in section 2 ,
the mu and tau neutrino flux originating from the earth atmosphere and the galactic plane is briefly discussed . in section 3 ,
the neutrino oscillation effects are studied for both . in section 4 ,
the limited future prospects for possible observations of galactic tau neutrinos are mentioned , whereas in section 4 , conclusions are presented .
briefly , the incoming cosmic rays interact with the air nuclei @xmath17 , in the earth atmosphere and give rise to mu neutrino flux . for @xmath18 , the @xmath19 , @xmath20 production and their direct and indirect decays
are the main sources of mu neutrinos , both being in region of conventional mu neutrino production.@xcite the absolute normalization of the conventional atmospheric neutrino flux is presently known to be no better than ( 20@xmath2125)%.@xcite for present estimates , the mu neutrino flux is taken from ref . .
these are neutrino flux calculations in one dimension without geomagnetic field effects .
the up down mu neutrino flux is taken to be the same , as the present discussion is independent of any specific detector . at higher energy ,
the prompt mu neutrino production from @xmath22 s dominates over the conventional one.@xcite the atmospheric tau neutrino flux arises mainly from @xmath23 and is calculated in ref . .
the quark gluon string model ( qgsm ) is used in ref . to model the @xmath24 interactions .
the low energy atmospheric tau neutrino flux is essentially isotropic.@xcite for @xmath25 gev , the atmospheric tau neutrino flux is obtained by re scaling w.r.t new cosmic ray flux spectrum , taking it to be dominantly the protons.@xcite the galactic mu neutrino flux for @xmath26 gev is calculated in ref . , whereas the galactic plane tau neutrino flux is calculated in ref . .
these calculations consider @xmath27 interactions inside the galaxy with target proton number density @xmath28 1/@xmath29 along the galactic plane .
the tau neutrino production is rather suppressed in the galactic plane relative to mu neutrino production .
the galactic plane mu and tau neutrino flux for @xmath25 gev is obtained by re scaling w.r.t new cosmic ray flux spectrum .
the distance inside the galactic plane is taken to be @xmath30 kpc where 1 pc @xmath31 km .
the mu neutrino flux is larger than the tau neutrino flux for @xmath25 gev from the two sites .
a detailed study that explicitly estimates the tau neutrino flux from the two sites for low energy indicates that the simple re scaling adopted here is a good approximation for @xmath32 gev.@xcite
in the two flavor approximation , the total tau neutrino flux is @xmath33 where @xmath34 is given by eq .
( [ osc - prob ] ) and @xmath35 .
the @xmath36 is in units of @xmath37s@xmath38sr@xmath38 and is taken according to discussion in section 2 .
three general directions in the earth atmosphere are considered as representative examples to compare the atmospheric tau neutrino flux with the galactic one in the presence of neutrino oscillations .
these are downward , horizontal and upward .
[ figone ] depicts the @xmath39 given by eq .
( [ osc - len ] ) for the range of @xmath40 given by eq .
( [ range ] ) with maximal mixing .
the three distances are taken from eq .
( [ len ] ) , with , for instance , the downward distance is obtained by setting @xmath41 .
the horizontal distance is obtained by setting @xmath42 .
( [ tot ] ) , the total downward going atmospheric tau neutrino flux is estimated .
it is then compared with the total galactic plane tau neutrino flux in fig .
[ figtwo ] for the whole range of @xmath40 with maximal mixing .
the distance @xmath10 for galactic plane neutrinos is taken as @xmath28 5 kpc . since @xmath43 , the galactic plane mu neutrinos oscillate before reaching the earth .
also , note that this flux is averaged out for the whole range of @xmath40 in the entire considered energy range .
the effect of different @xmath40 values diminishes for @xmath44 gev for total atmospheric tau neutrino flux . from the figure
, it can be seen that the galactic plane tau neutrino flux starts dominating over the downward going atmospheric tau neutrino flux even for @xmath45 as low as 10 gev in the presence of neutrino oscillations .
this is a very specific feature of tau neutrinos , and is absent for mu neutrinos .
this specific behavior has to do with the neutrino oscillations .
the galactic plane tau neutrino flux for @xmath46 in the presence of neutrino oscillations can be parameterized as @xmath47^{2.74 } } , \label{parameterize}\ ] ] where @xmath48 is in units of @xmath37s@xmath38sr@xmath38 and on r.h.s .
@xmath45 is in units of gev . in fig .
[ figthree ] , the galactic plane tau neutrino flux is compared with the atmospheric tau neutrino flux , using eq .
( [ tot ] ) for the three general directions for the atmospheric tau neutrino flux reaching the detector . here
the best fit values of the neutrino mixing parameters are used .
the oscillatory nature of the upward going tau neutrino flux can be seen from eq .
( [ osc - prob ] ) .
the cross over for the galactic tau neutrinos relative to the horizontal atmospheric tau neutrinos occurs at @xmath28 50 gev , whereas the same occurs for the upward direction at @xmath28 400 gev .
the total atmospheric tau neutrino flux is maximum in the upward direction ( see fig .
[ figthree ] ) .
it is minimum in downward direction , relative to the galactic plane tau neutrino flux in the presence of neutrino oscillations , owing to the behavior of @xmath49 ratio as a function of neutrino energy .
[ figthree ] indicates that zenith angle dependence of the total tau neutrino flux can at least in principle help to distinguish between atmospheric and non atmospheric tau neutrino flux .
the galactic tau neutrino flux transverse to the galactic plane is three to four orders of magnitude smaller than the galactic plane one.@xcite fig .
[ figfour ] gives a comparison of the downward going atmospheric and the galactic plane mu neutrino flux in the presence of neutrino oscillations .
for this comparison , mu neutrino flux is taken from ref . without re scaling for @xmath50 gev .
this mu neutrino flux includes contribution from the @xmath22 s for @xmath51 gev .
the total mu neutrino flux is estimated according to eq .
( [ tot ] ) with appropriate modifications for the best fit values of the two neutrino mixing parameters .
in contrast to the possibility of seeing the galactic plane with multi gev tau neutrinos , note here that with mu neutrinos , it can occur only for @xmath52 gev .
a relevant remark is that for the best fit values of the neutrino mixing parameters , the @xmath34 is relatively large along the horizontal and upward directions in the earth atmosphere [ see eq . ( [ osc - prob ] ) and fig . [ figone ] ] for @xmath53 .
so essentially the atmospheric mu neutrino flux in the absence of neutrino oscillations alone determine the total atmospheric tau neutrino flux in comparison with the total galactic tau neutrino flux .
for @xmath54 , a signature for the tau neutrinos is to measure the energy spectrum of the tau lepton induced electromagnetic and hadronic showers produced in tau neutrino nucleon interactions occurring inside a densely instrumented cherenkov radiation detector.@xcite though , it is a challenging task to distinguish between tau and non tau neutrinos for the present generation of detectors in the above energy range@xcite , however certain shower signatures remain distinctive for tau neutrinos.@xcite the galactic tau neutrino induced shower production rate can be approximately estimated by convolving the total galactic tau neutrino flux in the presence of neutrino oscillations , given by eq .
( [ parameterize ] ) with the @xmath55 , where @xmath56 for @xmath54 . for recent evaluations of @xmath55 ,
see ref . .
the possible tau lepton polarization effects@xcite are not taken into account in the event rate estimates presented here .
table [ tableone ] gives the galactic tau neutrino induced shower event rate for a 1 mega ton detector , in units of @xmath57 in @xmath58 steradians of upper hemi - sphere in four wide logarithmically equally spaced energy bins .
the table indicates that with a 3 to 5 year data collection time for a one mega ton detector , the galactic tau neutrino induced shower event rate can be in the range of ( 1@xmath2110 ) for @xmath32 gev .
this detector faces only the downward going atmospheric tau neutrino flux as background to the dominant galactic plane tau neutrino flux in the presence of neutrino oscillations .
the effects of neutrino oscillations on low energy ( @xmath50 gev ) tau neutrino flux produced in the earth atmosphere and in our galactic plane are presented in two neutrino flavor approximation .
the galactic plane should be observable with tau neutrinos with energy @xmath59 gev , depending on the orientation of the concerned detector w.r.t .
galactic center / plane at the time of observation .
the observation of galactic plane with multi gev tau neutrinos is in sharp contrast to the case of mu neutrinos with which the galactic plane is observable only with energy @xmath60 gev for the same orientation of the detector .
this observation may also have some relevance for the long baseline experiments searching for the tau neutrinos in @xmath5 oscillations.@xcite
0 h. athar , _ nucl . phys . proc . suppl .
_ * 122 * , 305 ( 2003 ) . for a recent brief review article , see , _
ibid_. , arxiv : hep - ph/0308188 [ to appear in _ chin . j. phys . _ ] . s. fukuda _ et al . _
[ super - kamiokande collaboration ] , _ phys .
lett . _ * 85 * , 3999 ( 2000 ) .
see , for instance , plenary talk given by y. suzuki at _
28th intl .
cosmic ray conf . _
( icrc 2003 ) , tokyo , japan , 31 july @xmath21 07 august , 2003 ( http://www-rccn.icrr.u-tokyo.ac.jp/icrc2003/talks/plenary-suzuki.pdf , to appear in its proceedings ) . t. kajita and y. totsuka , _ rev . mod .
_ * 73 * , 85 ( 2001 ) . for a recent discussion of up coming astrophysical neutrino telescopes ,
see , a. b. mcdonald _ et al_. , arxiv : astro - ph/0311343 . t. k. gaisser and t. stanev , _ phys .
_ * d57 * , 1977 ( 1998 ) ; _ ibid . , _ _ nucl .
suppl . _ * 70 * , 335 ( 1999 ) ; m. honda , t. kajita , k. kasahara and s. midorikawa , _ phys .
rev . _ * d64 * , 053011 ( 2001 ) .
g. t. zatsepin and v. a. kuzmin , _ sov .
* 14 * , 1294 ( 1962 ) ; l. v. volkova , _ sov .
j. nucl .
phys . _ * 31 * , 784 ( 1980 ) ; t. k. gaisser and todor stanev , s. a. bludman and h. lee , _ phys .
lett . _ * 51 * , 223 ( 1983 ) ; a. dar , _ phys .
lett . _ * 51 * , 227 ( 1983 ) ; k. mitsui , h. komori and y. minorikawa , _
nuovo cim . _ * c9 * , 995 ( 1986 ) ; e. v. bugaev and v. a. naumov , _ sov .
j. nucl .
phys . _ * 45 * , 857 ( 1987 ) ; t. k. gaisser , t. stanev and g. barr , _ phys . rev .
_ * d38 * , 85 ( 1988 ) ; a. v. butkevich , l. g. dedenko and i. m. zheleznykh , _ sov .
j. nucl .
phys . _ * 50 * , 90 ( 1989 ) ; e. v. bugaev and v. a. naumov , _ phys .
lett . _ * b232 * , 391 ( 1989 ) ; g. barr , t. k. gaisser and t. stanev , _ phys .
_ * d39 * , 3532 ( 1989 ) ; h. lee and y. koh , _ nouvo cim . _
* b105 * , 884 ( 1990 ) ; m. honda , k. kasahara , k. hidaka and s. midorikawa , _ phys .
_ * b248 * , 193 ( 1990 ) ; p. lipari , _ astropart .
phys . _ * 1 * , 195 ( 1993 ) ; d. h. perkins , _ astropart .
* 2 * , 249 ( 1994 ) ; t. k. gaisser , _
et al_. , _ phys .
rev . _ * d54 * , 5578 ( 1996 ) ; m. honda , t. kajita , k. kasahara and s. midorikawa , _ prog .
_ * 123 * , 483 ( 1996 ) ; p. lipari , t. stanev and t. k. gaisser , _ phys . rev . _
* d58 * , 073003 ( 1998 ) .
see , for instance , g. battistoni _ et al_. , _ astropart .
* 12 * , 315 ( 2000 ) . for a recent semi - analytic treatment , see , t. k. gaisser , _ astropart .
* 16 * , 285 ( 2002 )
. m. honda , t. kajita , k. kasahara and s. midorikawa , _ phys .
rev . _ * d52 * , 4985 ( 1995 ) ; v. agrawal , t. k. gaisser , p. lipari and t. stanev , _ phys . rev . _ * d53 * , 1314 ( 1996 ) . h. inazawa and k. kobayakawa , _ prog .
phys . _ * 69 * , 1195 ( 1983 ) ; l. v. volkova and g. t. zatsepin , _ sov .
j. nucl .
phys . _ * 37 * , 212 ( 1983 ) ; c. castagnoli _ et al .
_ , _ nuovo cim .
_ * a82 * , 78 ( 1984 ) ; h. inazawa , k. kobayakawa and t. kitamura , _ j. phys .
_ * g12 * , 59 ( 1986 ) ; l. v. volkova , w. fulgione , p. galeotti and o. saavedra , _
nuovo cim . _ * c10 * , 465 ( 1987 ) ; e. zas , f. halzen and r. a. vazquez , _ astropart .
phys . _ * 1 * , 297 ( 1993 ) ; g. battistoni _ et al_. , _ astropart .
phys . _ * 4 * , 351 ( 1996 ) ; v. a. naumov , t. s. sinegovskaya and s. i. sinegovsky , _
nuovo cim .
_ * a111 * , 129 ( 1998 ) ; e. v. bugaev _ et al_. , _ phys .
rev . _ * d58 * , 054001 ( 1998 ) ; l. pasquali , m. h. reno and i. sarcevic , _ phys . rev . _ * d59 * , 034020 ( 1999 ) ; g. gelmini , p. gondolo and g. varieschi , _ phys .
rev . _ * d61 * , 036005 ( 2000 ) ; l. v. volkova and g. t. zatsepin , _ phys . atom .
nucl . _ * 64 * , 266 ( 2001 ) ; c. g. s. costa , f. halzen and c. salles , _ phys .
rev . _ * d66 * , 113002 ( 2002 ) ; a. m. stasto , arxiv : astro - ph/0310636 ; h. athar , k. cheung , g. l. lin and j. j. tseng , arxiv : astro - ph/0311586 . l. pasquali and m. h. reno , _ phys .
rev . _ * d59 * , 093003 ( 1999 ) .
h. athar , k. cheung , g. l. lin and j. j. tseng , _ astropart .
phys . _ * 18 * , 581 ( 2003 ) .
t. k. gaisser and m. honda , _ ann .
sci . _ * 52 * , 153 ( 2002 ) .
t. k. gaisser , m. honda , p. lipari , and t. stanev , talk given at _
27th intl .
cosmic ray conf . _
( icrc 2001 ) , hamburg , germany , 07@xmath2115 august , 2001 , pp . 1643 - 1646 . for a recent calculation ,
see , g. ingelman and m. thunman , arxiv : hep - ph/9604286 .
h. athar , fei - fain lee , guey - lin lin , to be submitted .
m. thunman , g. ingelman and p. gondolo , _ astropart .
phys . _ * 5 * , 309 ( 1996 ) .
t. stanev , _ phys .
* 83 * , 5427 ( 1999 ) .
l. j. hall and h. murayama , _ phys .
_ * b463 * , 241 ( 1999 ) .
see also , f. vissani and a. yu . smirnov , _ phys .
lett . _ * b432 * , 376 ( 1998 ) .
e. a. paschos and j. y. yu , _ phys .
rev . _ * d65 * , 033002 ( 2002 ) ; s. kretzer and m. h. reno , _ phys .
* d66 * , 113007 ( 2002 ) ; _
ibid_. , arxiv : hep - ph/0306307 . k. hagiwara , k. mawatari and h. yokoya , _ nucl . phys . _ * b668 * , 364 ( 2003 ) .
see , for instance , http://www.hep.anl.gov/ndk/longbnews/ for details of the long base line neutrino experiments . | neutrinos with energy greater than gev are copiously produced in the @xmath0 interactions occurring in the earth atmosphere and in our galactic plane .
a comparison of the tau and mu neutrino flux in the presence of neutrino oscillations from these two astrophysical sites is presented .
it is pointed out that the galactic plane tau neutrino flux dominates over the downward going atmospheric tau neutrino flux at much lower energy value than that for the dominance of the mu neutrino flux from these two sites .
future prospects for possible observations of galactic tau neutrino flux are also briefly mentioned . |
let s consider 2d ising model on the simplest regular lattice with the partition function @xmath0\right\ } , \nonumber \\[8pt ] m=1,\,\ldots,\,m , \quad n=1,\,\ldots,\,n . \label{z}\end{aligned}\ ] ] a pair of numbers @xmath1 enumerate the vertex of the lattice which is at the intersection of @xmath2-th column and @xmath3-th line of the lattice .
first the partition function ( [ z ] ) was calculated by l. onsager @xcite , and then by several another authors and methods ( see , for example , @xcite-@xcite , and others references can be found in @xcite ) .
taking into account that @xmath4 , rewrite the partition function as @xmath5 .
\label{z1}\end{aligned}\ ] ] the right - hand side of the last equation is a polynomial in the variables @xmath6 , each variable @xmath7 being in the degree not higher than 4 . as a result of summation in ( [ z1 ] )
all summands with odd degrees of each of these variables reduce to zero .
thus one should take into account only the summands which are proportional to @xmath8 for all @xmath2 and @xmath3 .
it follows from ( [ z1 ] ) that the sum is equal to the sum of all closed contours ( loops ) on the lattice , generally with intersections and self - intersections .
let s call the elementary parts of loops connecting the nearest vertexes ( say , @xmath1 and @xmath9 or @xmath1 and @xmath10 ) by connections .
thus the length of the loop is equal to the number of its connections .
number of connections coming to each vertex can be equal to 0 , 2 or 4 .
thus the partition function ( [ z1 ] ) can be expressed as @xmath11 where @xmath12 is the total number of connections of the loop . here
an another method of calculation of partition function ( [ z ] ) is proposed .
let s consider the clifford algebra with @xmath13 generatives : @xmath14 matrices @xmath15 are hermitian and i assume that their dimensions to be @xmath16 , that is the minimal possible one at given number of matrices . and
@xmath17 are even . ]
algebra ( [ ac ] ) implies that the trace of any odd product of @xmath18-matrices is equal to zero , and @xmath19 and so on . by virtue of algebra ( [ ac ] ) any even product of @xmath18-matrices
is reduced to the @xmath20 ( versus the number of their permutations ) or to the product of two by two different @xmath18-matrices . according to ( [ ac1 ] )
the trace of the product is equal to @xmath21 in the first case and to zero in the last case .
the theory of @xmath18-matrices in the spaces of large dimensions as well as in hilbert space can be found in @xcite .
further we shall consider that the indices @xmath22 and so on enumerate the nodes of the lattice and thus they are integer vectors in the plane : @xmath23 ; it is assumed that the index @xmath2 increases on the right while the index @xmath3 increases up .
thus , the matrix @xmath24 is related to each node @xmath25 .
we define also the elementary basic vectors @xmath26 and @xmath27 , so that @xmath28 .
the following statement is valid : the partition function ( [ z ] ) or ( [ z1 ] ) can be rewritten also as @xmath29 \ldots\times\big[(\lambda+\mu\gamma_{{\bf x}}\gamma_{{\bf x}+{\bf e}_2})(\lambda+\mu\gamma_{{\bf x } } \gamma_{{\bf x}+{\bf e}_1})\big ] \big[(\lambda+\mu\gamma_{{\bf x}+{\bf e}_1}\gamma_{{\bf x}+{\bf e}_1+{\bf e}_2 } ) ( \lambda+\mu\gamma_{{\bf x}+{\bf e}_1}\gamma_{{\bf x}+2{\bf e}_1})\big]\times\cdots \nonumber \\[10pt ] \cdots\times\big[(\lambda+\mu\gamma_{{\bf x}+{\bf e}_2}\gamma_{{\bf x}+2{\bf e}_2})(\lambda+ \mu\gamma_{{\bf x}+{\bf e}_2}\gamma_{{\bf x}+{\bf e}_1+{\bf e}_2})\big]\times \nonumber \\[6pt ] \times\big[(\lambda+\mu\gamma_{{\bf x}+{\bf e}_1+{\bf e}_2 } \gamma_{{\bf x}+{\bf e}_1 + 2{\bf e}_2})(\lambda+\mu\gamma_{{\bf x}+{\bf e}_1+{\bf e}_2 } \gamma_{{\bf x}+2{\bf e}_1+{\bf e}_2 } ) \big]\times\cdots\big\}. \label{z3}\end{gathered}\ ] ] the expression in braces in ( [ z3 ] ) is some polynomial in @xmath18-matrices .
the statement is true if the traces of all monomials of @xmath18-matrices in ( [ z3 ] ) are non - negative ( the positivity condition which is proved further ) and @xmath30 note that the last expression ( [ z3 ] ) is obtained from ( [ z1 ] ) by substitutions @xmath31 and @xmath32 .
but in contrast to ( [ z1 ] ) the sequence of multipliers arrangement as well as their form are crucial in the case ( [ z3 ] ) .
draw attention to the fact that in the product under the sign of trace in ( [ z3 ] ) at first the brackets are multiplied in series along the lines , and then the results of multiplications along the lines are multiplied in series along the column . to establish the coincidence of the right - hand sides of eqs .
( [ z3 ] ) and ( [ z2 ] ) , it is sufficient to prove the positivity condition and to take into account the fact that under the sign of trace in ( [ z3 ] ) only that polynomials in @xmath18-matrixes `` survive '' which contain each of @xmath18-matrix in degrees 0 , 2 or 4 ( as well as in the case of variables @xmath33 ) and use designations ( [ r ] ) .
let s prove positivity condition with the help of mathematical induction method . during the process of calculations
the numerical factors are ignored since only the question about the sign of the loop is significant here .
we begin the calculation with verification of the fact that the matrix factor of elementary cell , corresponding to elementary loop binding elementary cell with vertexes @xmath34 is equal to unit . indeed , according to ( [ z3 ] ) this matrix factor
is equal to @xmath35 the further calculation is based on the possibility of presentation of loop matrix factor as a product of matrix factor of smaller loop , binding smaller number of elementary cells as compared with initial loop , and the matrix factors of elementary cells supplying smaller loop up to the initial one .
each of the equalities in figs . @xmath36 and @xmath37 represents graphically equality between loop matrix factor in the left - hand side and smaller loop matrix factor times elementary cells matrix factors .
all loop matrix factors are calculated according to order in ( [ z3 ] ) . in fig .
1 solid lines describe loops while dotted lines divide loop interiors into elementary cells .
equations in fig .
1 are established easily by direct testing .
for example , for figs .
@xmath38 and @xmath39 we have : fig . 1_a _ fig .
1_b _ @xmath40= \nonumber \\[8pt ] = \big[\big(\gamma_1\gamma_4\big)\big(\gamma_1\gamma_2\big)\big(\gamma_2\gamma_5\big)\big(\gamma_4\gamma_5\big)\big ] \big[\big(\gamma_2\gamma_5\big)\big(\gamma_2\gamma_3\big)\big(\gamma_3\gamma_6\big)\big(\gamma_5\gamma_6\big)\big ] , \nonumber \\[8pt ] \big[\big(\gamma_7\gamma_9\big)\big(\gamma_7\gamma_8\big)\big(\gamma_8\gamma_{10}\big ) \big(\gamma_9\gamma_{11}\big)\big(\gamma_{10}\gamma_{12}\big)\big(\gamma_{11}\gamma_{12}\big)\big]= \nonumber \\[8pt ] = \big[\big(\gamma_7\gamma_9\big)\big(\gamma_7\gamma_8\big ) \big(\gamma_8\gamma_{10}\big)\big(\gamma_9\gamma_{10}\big)\big ] \big[\big(\gamma_9\gamma_{11}\big)\big(\gamma_9\gamma_{10}\big ) \big(\gamma_{10}\gamma_{12}\big)\big(\gamma_{11}\gamma_{12}\big)\big ] , \label{ind3}\end{gathered}\ ] ] and so on .
developing loops and their matrix factors following to the schemes in fig . 1 , one can realize matrix factors of any loops as the product of elementary cells matrix factors which are equal to unit according to ( [ ind2 ] ) .
thus the equality of any loops matrix factors to unit is proved inductively .
hence the positivity condition is established .
it is evident that each multiplier @xmath41 in the braces in ( [ z3 ] ) is the matrix of orthogonal rotation by the angle @xmath42 in spinor representation in @xmath13-dimensional euclidean space in the plane marked by the pair of @xmath18-matrices in the multiplier .
therefore , all product in the braces in ( [ z3 ] ) is the matrix of an orthogonal rotation in spinor representation in @xmath13-dimensional euclidean space .
it is known @xcite that this matrix can be expressed in the form @xmath43 and @xmath44 the trace of the braces in ( [ z3 ] ) , i.e. the trace of the matrix @xmath45 , is expressed simply through the eigenvalues of real orthogonal matrix @xmath46 .
let the set of numbers @xmath47 form the complete set of eigenvalues of the matrix @xmath48 . then ( see appendix ) @xmath49= \prod_{k=1}^{mn/2}\left[2\cos\left(\frac{\phi_k}{2}\right)\right]=
\prod_{k=1}^{mn/2}\left(\sqrt{\rho_k}+\sqrt{\overline{\rho}_k}\right ) , \nonumber \\[8pt ] \rho_k = e^{i\phi_k}. \label{repr4}\end{gathered}\ ] ] in statistical limit , when @xmath50 , the problem of matrix @xmath46 diagonalization simplifies radically since the matrices @xmath51 and @xmath52 become dependent only on @xmath53 at relatively small distances from boundary of the lattice .
this property is known as translational invariance .
therefore the diagonalization of these matrices is performed by means of fourier transformation , i.e. by passing to quasi - momentum representation .
the following complete orthonormal set of functions on the lattice is used for that purpose : @xmath54 p=-\dfrac{\pi(m-2)}{m},\,-\dfrac{\pi(m-4)}{m},\,\ldots,\,0,\,\dfrac{2\pi}{m},\,\ldots,\,\pi , \nonumber \\[8pt ] q=-\dfrac{\pi(n-2)}{n},\,-\dfrac{\pi(n-4)}{n},\,\ldots,\,0,\,\dfrac{2\pi}{n},\,\ldots,\,\pi , \nonumber \\[10pt ] \quad { \bf k}=(p,\,q ) , \nonumber \\[10pt ] \sum_{{\bf x}}\overline{\psi}_{{\bf k}}({\bf x})\psi_{{\bf k}'}({\bf x})=\delta_{{\bf k}{\bf k } ' } \longleftrightarrow \sum_{{\bf
k}}\psi_{{\bf k}}({\bf x})\overline{\psi}_{{\bf k}}({\bf x}')= \delta_{{\bf x}{\bf x}'}. \label{repr5}\end{gathered}\ ] ]
at first let s solve a particular problem : define for @xmath55 the spinor rotation operator transforming only @xmath18-matrices with indexes @xmath56 ) and @xmath57 , where @xmath58 : @xmath59\times \nonumber \\[8pt ] \times\big[(\lambda+\mu\gamma_{{\bf x}+{\bf e}_1}\gamma_{{\bf x}+{\bf e}_1+{\bf e}_2 } ) ( \lambda+\mu\gamma_{{\bf x}+{\bf e}_1}\gamma_{{\bf x}+2{\bf e}_1})\big]\ldots\ , . \label{ev1}\end{gathered}\ ] ] here we have the ordered product of spinor rotation matrixes along only one line @xmath3 . the corresponding orthogonal matrix is defined just as in ( [ repr2 ] ) : @xmath60 according to definitions ( [ z3 ] ) , ( [ repr2 ] ) and ( [ ev1 ] ) @xmath61 in the right hand sides of eqs . ( [ ev3 ] )
the ordered products of matrixes corresponding to lines take place .
let s find the obvious expression for the matrix @xmath62 with the help of ( [ ev2 ] ) .
this calculation is based on the simple relations @xmath63 @xmath64 @xmath65 following directly from eqs .
( [ ac ] ) and ( [ r ] )
. either vector @xmath66 or vector @xmath67 can be taken instead of a vector @xmath68 in eqs .
( [ ev4])-([ev6 ] ) .
consider the first case when @xmath69 .
then in the left hand side of eq .
( [ ev2 ] ) all spinor rotation matrixes forming @xmath70 placed to the right from @xmath71 and all spinor rotation matrixes forming @xmath72 placed to the left from the matrix @xmath73 are cancelled mutually since they do not catch on matrix @xmath74 .
but the `` facings '' @xmath75 transform @xmath74 in accordance with eq .
( [ ev5 ] ) in which one must make the identifications @xmath76 .
so we obtain the matrix element @xmath77 fig .
2 further only eq . ( [ ev4 ] ) is used in which the identifications @xmath78 are made in series , and at each @xmath79 at first @xmath80 and then @xmath81 are put .
this process is pictured symbolically in fig . 2 where the bold point denotes the position of the initial @xmath18-matrix placed in `` facings '' in the left hand side of eq .
( [ ev2 ] ) , and the arrows show where its image comes under the action of the described linear transformations .
every time the result is reduced to multiplication by @xmath82 if the matrix is transfered up or to the right , and to multiplication by @xmath83 if the matrix is left on the former place .
the matrix transfered up further remains unchanged .
thus the following result is obtained : @xmath84^{m'}(\cos\psi)^3 , \quad { \cal o}^{(n)}_{{\bf y},\,{\bf y}+m'{\bf e}_1+{\bf e}_2}=[\sin\psi\cos\psi]^{m ' } , \nonumber \\[10pt ] m'=0,\,1,\ldots , \quad { \bf y}=(m,\,n ) . \label{ev8}\end{gathered}\ ] ] the matrix elements as @xmath85 are calculated similarly . the process of this calculation is pictured in fig .
3 where the bold point again denotes the position of @xmath18-matrix at the beginning of the process , and the arrows show the subsequent movements of the images of initial matrix . at the first step the relation ( [ ev5 ] ) is used in which the identifications @xmath86 are made .
the relation ( [ ev4 ] ) is used at the all subsequent steps . as a result of simple calculation
we obtain : fig . 3 @xmath87^{m ' } , \nonumber \\[10pt ] { \cal o}^{(n)}_{{\bf y}+{\bf e}_2,\,{\bf y}+m'{\bf e}_1}=-[\sin\psi\cos\psi]^{m'+1 } , \quad m'=0,\,1,\ldots , \quad { \bf y}=(m,\,n ) .
\label{ev9}\end{gathered}\ ] ] now it is necessary to make partial diagonalization of the orthogonal matrix using its translational invariance . for that purpose
we make the partial fourier transformation along the lines of @xmath18-matrixes with the help of eqs .
( [ repr5 ] ) : @xmath88_+=2\delta_{n,\,n'}\delta_{p,\,p'}. \label{ev10}\end{gathered}\ ] ] let s pass to fourier transformation along the lines in eq .
( [ ev2 ] ) . using translational invariance we rewrite this equation in the form @xmath89\gamma_{n''}(p ) .
\label{ev11}\end{gathered}\ ] ] the matrix in square bracket in the right hand side of eq .
( [ ev11 ] ) is designated as @xmath90 and it is calculated with the help of eqs .
( [ ev7])-([ev9 ] ) : @xmath91 here @xmath92 } , \quad b_p=\frac{[\sin\psi\cos\psi]}{1-e^{ip}[\sin\psi\cos\psi ] } , \label{ev13}\end{aligned}\ ] ] and diagonal elements @xmath93 and @xmath94 are settled on @xmath3-th and @xmath95-th places , correspondingly .
the direct check shows that the matrix ( [ ev12 ] ) is unitary .
thus , according to ( [ ev3 ] ) @xmath96=\big\{\ldots { \cal o}^{(n-1)}_p{\cal o}^{(n)}_p{\cal o}^{(n+1)}_p\ldots\big\}_{n',n''}. \label{ev14}\end{aligned}\ ] ] the product ( [ ev14 ] ) of matrix of the kind ( [ ev12 ] ) is calculated easily : @xmath97 ( -b_p ) , & \mbox{for } \ \
n'=-1 \\[5pt ] c_pb_p^{n'}a_p , & \mbox{for } \ \
n'\geq 0 \end{array } \right . .
\label{ev15}\end{aligned}\ ] ] to find the eigenvalues of orthogonal matrix ( [ repr2 ] ) one must calculate the fourier components of ( [ ev15 ] ) along the column : @xmath98 \eta_{p , q}=1-\left(e^{-ip}+e^{-iq}\right)(\sin\psi\cos\psi ) .
\label{ev16}\end{aligned}\ ] ] equation ( [ ev16 ] ) is obtained with the help of ( [ ev13 ] ) .
it follows from eqs . ( [ repr4 ] ) , ( [ repr5 ] ) and ( [ ev16 ] )
that free energy is proportional to the following integral @xmath99 = \frac{mn}{4\pi^2}\int_{-\pi}^{+\pi}\d p\int_{-\pi}^{+\pi}\d q\ln\left(\frac{\eta_{p , q}+\overline{\eta}_{p , q } } { \sqrt{\eta_{p , q}\overline{\eta}_{p , q}}}\right)=\frac{mn}{4\pi^2 } \int_{-\pi}^{+\pi}\d p\int_{-\pi}^{+\pi}\d q\ln\left(\eta_{p , q}+\overline{\eta}_{p , q}\right ) . \label{tt10}\end{gathered}\ ] ] the last equality in ( [ tt10 ] ) is true because @xmath100 prove the equalities ( [ tt20 ] ) . according to ( [ ev16 ] ) @xmath101 \alpha=\frac12(\sin2\psi ) , \quad \beta_p=\frac{\sin2\psi}{2-(\sin2\psi)e^{-ip } } , \quad \label{tt30}\end{gathered}\ ] ] and @xmath102 only in the case @xmath103
therefore @xmath104 = \int_{-\pi}^{+\pi}\d q\left\{\int_{-\pi}^{+\pi}\d p\,\ln\left(1-\alpha e^{-ip}\right)\right\}+\int_{-\pi}^{+\pi}\d p\left\{\int_{-\pi}^{+\pi}\d q\,\ln\,\left(1-\beta_pe^{-iq}\right)\right\}= \nonumber \\[8pt ] = -\sum_{n=1}^{\infty}\frac1n\left\{\alpha^n\int_{-\pi}^{+\pi}\d q\int_{-\pi}^{+\pi}\d p\,e^{-inp}+\int_{-\pi}^{+\pi}\d p\,\beta_p^n\int_{-\pi}^{+\pi}\d q\,e^{-inq}\right\}=0 . \label{tt40}\end{gathered}\ ] ] the second equality in ( [ tt20 ] ) is proved similarly .
eventually , we can write out the free energy with the help of formulas ( [ z3 ] ) , ( [ r ] ) , ( [ repr4 ] ) , ( [ ev16 ] ) and ( [ tt10 ] ) : @xmath105 = -mnt\left\{\ln2+\frac{1}{8\pi^2}\int_{-\pi}^{+\pi}\d p\int_{-\pi}^{+\pi}\d q\ln \left[\left(\ch2\theta\right)^2-\left(\sh2\theta\right)(\cos p+\cos q)\right]\right\}= \nonumber \\[8pt ] = -mnt\big\{\ln2-\ln(1-x^2)+ \nonumber \\[8pt ] + \frac{1}{8\pi^2}\int_{-\pi}^{+\pi}\d p\int_{-\pi}^{+\pi}\d q\ln \left[(1+x^2)^2 - 2x(1-x^2)(\cos p+\cos q)\right]\big\ } , \quad x=\th\theta . \label{tt60}\end{gathered}\ ] ] the last expression in ( [ tt60 ] ) coincides with the free energy of 2d ising model given in @xcite . the temperature of phase transition is obtained directly from ( [ tt10 ] ) and ( [ ev16 ] ) .
indeed , free energy has a peculiarity at the phase transition point .
it is seen from ( [ tt10 ] ) that this takes place when @xmath106 for a part of quasi - momenta , which occurs only if ( see ( [ ev16 ] ) ) @xmath107 thus with the help of ( [ tt ] ) and ( [ r ] ) we find that critical temperature is found from the equation @xmath108 let @xmath110 , be the complete orthonormal set of eigenvectors of the matrix @xmath46 , so that the eigenvalue @xmath111 @xmath112 corresponds to the eigenvector @xmath113 @xmath114 . further also the designation @xmath115 is used .
we shall consider the introduced vectors as vector - columns and the upper indices @xmath116 and @xmath117 denote the transposition and hermitian conjugation of vectors and matrices . by definition
@xmath118 the given definitions imply the following formulas : @xmath119 @xmath120 @xmath121 it is shown in @xcite that @xmath122 due to ( [ ap3 ] ) and ( [ ap5 ] ) we have @xmath123 @xmath16-matrixes @xmath124 possess all properties of fermion creation and annihilation operators . indeed , in consequence of ( [ ac ] ) and ( [ ap2 ] ) @xmath125_+=\delta_{kk ' } , \quad [ c_k,\,c_{k'}]_+=[c^{\dag}_k,\,c^{\dag}_{k'}]_+=0 .
\label{ap8}\end{gathered}\ ] ] according to the definitions ( [ ap3 ] ) and ( [ ap7 ] ) we have @xmath126 with the help of eqs .
( [ ap5 ] ) , ( [ ap8 ] ) and ( [ ap9 ] ) the quantity ( [ ap6 ] ) is rewritten as @xmath127= \sum_{k=1}^{mn/2}\left[\left(\ln\rho_k\right)c^{\dag}_kc_k-\frac12\ln\rho_k\right ] .
\label{ap10}\end{gathered}\ ] ] equality ( [ repr4 ] ) follows immediately from ( [ ap10 ] ) since the calculation of the trace in terms of @xmath18-matrixes is equivalent to the calculation of trace in terms of the corresponding fermionic operators ( [ ap7 ] ) . | one more solution of 2d ising model is found . |
the unbounded identifier model were introduced in @xcite . here
we adopt this model with minor changes in formal definitions .
we consider point - to - point synchronous networks and distributed protocols in them .
nodes of a network communicate by sending messages through bi - directed links .
all nodes perform the same communication protocol .
communication speed is bounded by @xmath6 .
the nodes are distinguishable but the amount of information to distinguish them may be arbitrary large . in this section
we provide formal definitions for this model . a _ network _
@xmath12 is a connected graph , where @xmath4 is the set of nodes and @xmath9 is the set of bi - directed links .
the graph is equipped by a set of functions .
an injective function @xmath13 assigns the unique identifier @xmath14 to a node @xmath15 .
links of each node are ordered by index functions @xmath16 .
let @xmath17 be a degree of a node in @xmath12 .
then the range of the index function @xmath18 is exactly the set of edges incident to @xmath15 in @xmath12 .
we assume that initially a node has only local information about a network .
it knows the number of its links ( the degree @xmath17 ) , can distinguish links by its index function , and knows its identifier .
this model uses discrete time .
time moments are numbered by nonnegative integers . at a moment
@xmath19 a node @xmath15 receives messages from all its neighbors .
the messages are bit strings , their lengths are bounded by a constant specified by a protocol .
the empty string @xmath20 is allowed .
it indicates absence of a message .
the list of messages received by the node @xmath15 at a moment @xmath19 is denoted as @xmath21 the messages are ordered by the index function .
usually states of nodes are used in protocol descriptions .
but it is unnecessary and we avoid states in the definitions .
one may take a history of communication in a node as a state of a node . _ a history of communication _
@xmath22 in a node @xmath15 at a moment @xmath19 is the list @xmath23 .
we assume that messages depend on the history of communication only .
( so the protocols are deterministic . ) by definition , a _ protocol _ is the message function @xmath24 that determines the messages sent during the current round of communication .
the first argument in the message function is an index of a link in the link enumeration .
the second argument is the history of communication
. the message function may be arbitrary .
it means that nodes are computationally unlimited .
nevertheless , it is worth to mention that in the protocols presented in this paper the message functions are easy to compute .
initially , at time @xmath25 , the history of communication in a node @xmath15 is @xmath26 .
then the configuration of the network is changed according the following rules : @xmath27 where the link @xmath28 has the index @xmath29 in the node @xmath15 and the index @xmath30 in the node @xmath31 .
( recall that no communication errors are allowed in this model . )
note that the rules are the same for all nodes in the network .
but behavior of a node may depend on a moment of time due to the synchronization assumption .
we say that a node @xmath15 is in the _ final state _ at moment @xmath19 , if @xmath32 for all @xmath29 , @xmath33 .
it means that in the final state a node is not communicating with a network : it sends nothing and ignores all incoming messages .
we say that a node @xmath15 is in a _ sleep _ state at moment @xmath19 , if @xmath34 where @xmath35 is the list composed of the empty strings .
thus a node in a sleep state does not generate messages and does not change its state if it has not received messages from its neighbors at the previous moment of time .
it implies that communication finishes if all nodes are in sleep states . but a node in a sleep state can be ` awaken ' by a suitable message .
a protocol is _ message terminating _
if for any network @xmath12 , @xmath36 , @xmath37 all nodes are in a sleep state at some moment @xmath19 .
a protocol is _ processor terminating _ if for any @xmath12 , @xmath36 , @xmath37 all nodes are in the final state at some moment @xmath19 .
the running time @xmath38 of the protocol in the network @xmath12 , @xmath36 , @xmath37 is the first moment of time when all nodes are in the final ( a sleep ) state .
now we are going to define a useful result of a protocol . loosely speaking ,
it is an arbitrary function of a history of communication .
we say that a protocol @xmath39 _ broadcasts the minimal identifier _ if it terminates in any network and there is a function @xmath40 such that @xmath41 for all @xmath42 . here
@xmath38 is the running time of the protocol and @xmath43 is the minimal identifier in the network with respect to _ the shortlex order _
@xmath44 : compare strings by the length and in the case of equal lengths compare them lexicographically .
we say that a protocol @xmath39 _ elects a leader _ if it terminates in any network and there is a function @xmath45 such that ( i ) @xmath46 and ( ii ) after the protocol finishes , there is exactly one node ( the leader ) with the value @xmath47 : @xmath48 for the leader @xmath49 and @xmath50 for all @xmath51 .
it follows from the definitions that if a protocol broadcasts the minimal identifier then it also elects a leader : the value @xmath52 is determined by @xmath43 and @xmath36 in the straightforward way .
we say that a protocol @xmath39 _ constructs a rooted spanning tree _ if it terminates in any network and there is a function @xmath53 such that ( i ) @xmath54 ; ( ii ) there is exactly one node @xmath55 such that @xmath56 ; ( iii ) for all @xmath42 if @xmath57 then @xmath58 and ( iv ) the links @xmath59 , where @xmath57 , form a spanning tree in the network .
the definition of protocols constructing a spanning tree looks rather artificially .
it is used to simplify formal arguments .
note in advance that an orientation from a child to a parent appears in our protocols naturally as well as the list of childs .
so we skip the list in the definition .
speaking of rooted trees , we always assume edge orientation towards the root .
in this section we discuss informally the protocols presented below .
the basic idea is very simple : to broadcast information as fast as possible .
once a node obtains some useful information , it immediately resends it to all its neighbors .
we illustrate the idea in a simpler situation .
suppose that initially there is the unique _ active _ node in a network ( a leader ) .
it starts communication and sends its identifier to neighbors bit by bit .
an initially nonactive node awakes after receiving a non - empty message and chooses a sender of this message as a parent .
then the node also transmits to its neighbors bits received from the parent . in this way
we obtain a protocol that constructs a rooted tree in time @xmath60 , where @xmath1 is the size of the active node identifier .
the root of the tree is the initially active node .
such protocols are called _
flooding protocols_. in absence of a leader ( the initially active node ) the straightforward realization of this idea takes more time .
we mean the following rules : nodes send identifiers to the neighbors , compare received identifiers and transmit the current minimal value . in this way we obtain a sort of flooding protocol to broadcast the minimal identifier .
the protocol is easily transformed to a protocol constructing a spanning tree .
the running time of this protocol in our settings is @xmath10 . essentially , it is the well - known perlman protocol @xcite adjusted for our model . to speed up this flooding protocol
we will use two ideas . at first
, we encode identifiers in such a way that for encoded strings the result of comparison of string s prefixes does not change after receiving the suffixes of the strings .
it gives a way to compare long strings by their short prefixes .
thus , our protocols transmit _ keys _ ,
i.e. encoded identifiers .
the second idea is to transmit the current information about the minimal key although it might be incorrect .
when incorrectness is revealed , a correction message is sent instead of re - sending the whole information about the minimal key .
of course , corrections make the running time longer . in section [ min - broadcasting ]
we give the exact rules for the protocol and prove the time bound given in theorem [ th : m - t - prot ] . in this way
we get a message terminating protocol . after receiving a set of keys ,
a node can not exclude a possibility that somewhere in a network there is a smaller key .
to make a protocol processor terminating requires additional work .
thus nodes should communicate and establish the fact of termination of the key distribution process . to make this communication faster
, the nodes may use a spanning tree built by the message terminating protocol . for this purpose
we implement a protocol based on the idea of echo protocol ( see , e.g. @xcite ) adjusted to our model .
broadcasting messages through a tree requires a time proportional to the diameter of the tree .
thus we come to the time bound given in theorem [ th : p - t - prot ] : the tree depth does not exceed the running time of the message terminating protocol .
details are presented in section [ finalizing ] .
we encode the identifiers by the map @xmath61 on the set @xmath62 of binary strings .
hereinafter we use the standard notation : @xmath63 is the length of a binary string @xmath64 , @xmath65 is a string @xmath66 repeated @xmath67 times .
all logarithms are binary . for the empty string
@xmath20 we set @xmath68 .
it corresponds to the general rule with @xmath69 .
the image @xmath70 is the set of _
it is easy to see several simple and useful properties of this encoding : * if @xmath71 then @xmath72 ; * the length of @xmath64 is determined by a prefix @xmath73 of the key @xmath74 ; * the encoding is injective ( an identifier can be restored from its key ) ; * any key is not a proper prefix of another key ; * @xmath75 .
we are interested in asymptotic bounds .
so we will ignore the difference between lengths of keys and lengths of identifiers .
keys are compared in _ the partial lexicographical order _ ( pl order ) .
let @xmath76 be the longest common prefix of the strings @xmath77 , @xmath15 .
then by definition @xmath77 is less than @xmath78 in the pl order if @xmath79 , @xmath80 .
we denote this fact by @xmath81 .
note that a prefix of a string is not pl comparable with the string . it justifies the term ` partial ' . in the other direction , if @xmath76 is a proper prefix of both strings @xmath77 and @xmath15 then the strings are comparable in the pl order . we indicate properties of the pl order that will be useful in the analysis of the protocols .
[ greedy - compare ] if @xmath81 then @xmath82 for all @xmath83 , @xmath84 .
the proposition follows easily from the definition of the pl order .
[ length - compare ] if @xmath77 , @xmath15 are keys and @xmath85 then @xmath81 .
moreover , there are prefixes @xmath86 and @xmath87 of the keys @xmath77 , @xmath15 such that @xmath88 and @xmath89 .
note that @xmath90 , where @xmath91 .
so the prefixes @xmath92 , @xmath93 of the keys @xmath77 , @xmath15 are pl compared and @xmath88 .
the second claim follows from the inequality @xmath94 .
there is no linear order on the whole set of binary strings satisfying proposition [ greedy - compare ] and proposition [ length - compare ] .
thus we need to introduce keys .
also note that the pl order is a linear order on the set of the keys .
[ lex - consistent ] the pl order on the keys is consistent with the shortlex order @xmath44 on the identifiers : @xmath95 iff @xmath96 . if @xmath71 then @xmath97 , @xmath98 . in this case
@xmath99 is equivalent to @xmath95 . if @xmath100 then either @xmath101 or @xmath72 .
the former implies @xmath96 by proposition [ length - compare ] . if @xmath72 , then @xmath102 and @xmath103 .
thus @xmath96 . in the other direction
the arguments are similar .
[ min - pl ] let @xmath104 be the set of pl - minimal elements of a set @xmath105 of strings .
then for any pair of strings in @xmath104 one is a prefix of another .
the minimal elements form an antichain in the pl order .
but the only pl uncomparable pairs are ( string @xmath77 , a prefix of @xmath77 ) .
messages in the protocol consist of the _ information bit _ and _ control bits_. there will be finitely many control bits . thus the communication speed is @xmath6 .
the value 1 of a control bit indicates that a specific event occurs in the process of communication .
absence of the information bit in a message is allowed ( the case is indicated by the empty string instead of 1-bit string ) .
we assume that if the information bit is empty and all control bits are @xmath106 , then no message is sent ( formally , the empty string is sent ) .
each node sends the same messages to all its neighbors in this protocol .
the message is determined by the following data . executing the protocol ,
a node maintains _ the candidate _
( to the minimal key ) and the list of _
participants_. candidates and participants are binary strings .
participants are prefixes of the candidates that has been received by a node from its neighbors . in particular
, the number of participants equals the number of neighbors in each node .
initially the candidate of a node is its key and all participants are empty . at any moment of time
the candidate is the longest pl - minimal string among the participants and the node s key .
proposition [ min - pl ] guarantees the uniqueness of the candidate . at _ regular periods of time _
a node sends bits of its candidate one by one until all bits of the candidate are sent . at _ exceptional periods of time _
the participant @xmath107 sent to the neighbors of a node is not a prefix of the current candidate @xmath108 of the node . during an exceptional period the node sends _ a correction message _ : the length of @xmath109 written in binary .
( in other words , a node tells to neighbors how many bits in its data are valid . )
the end of a correction message is indicated in control bits as well as the end of a correction message .
it is possible that the candidate is changed during an exceptional period of time .
it might cause that the current correction message becames incorrect ( more bits in the participant are wrong ) . in this case the current correction message is aborted ( this event is indicated in control bits ) and the node starts transmission of a new correction message . if a node has finished transmission of the candidate it goes to a sleep state .
below we provide the exact rules of the protocol .
we need the following notation .
let @xmath110 $ ] be the substring of a string @xmath49 that starts at the position @xmath29 and ends in the position @xmath30 . a single @xmath30th bit of the string
is denoted by @xmath111 $ ] .
@xmath112 is the key of the node @xmath77 , i.e. , is a shortcut of @xmath113 .
@xmath114 is the candidate of the node @xmath77 after @xmath19 steps of the protocol .
@xmath115 is the participant sent by the node @xmath77 to all its neighbors after @xmath19 steps of the protocol . to define the rules for exceptional periods we need a notation @xmath116 that means the prefix of the correction message that has been sent to neighbors after @xmath19 steps of the protocol . during regular periods
, @xmath116 is the empty string . in this protocol
a node sends the same message to all its neighbors .
so we indicate an information sent by the node .
just the same information is received by all neighbors of the node ( recall that there are no communication errors in our model ) .
thus , an operation of a node at step @xmath117 of protocol is defined by the data @xmath118 i.e. , by the candidate , by the prefixes of the candidate and a correction message sent up to the time , and by participants and prefixes of correction messages received from neighbors .
the node also takes into account the values of control bits . for simplicity
we do not introduce notation for control bits .
instead , we describe a way to use them .
to form a message at step @xmath117 , the node also uses a correction message string @xmath119 .
it is determined by the basic data as follows : if @xmath115 is a prefix of @xmath114 ( a regular period ) , then @xmath119 is the empty string .
otherwise ( an exceptional period ) @xmath120 where @xmath121 is a binary representation of an integer @xmath122 .
if @xmath119 is empty , then , if @xmath123 , the information bit @xmath124 of the next message is the next bit of the candidate , i.e. , @xmath125;\ ] ] otherwise , @xmath126 .
if @xmath119 is non - empty , then the node transmits the following message ( and indicates in control bits that it is a message of an exceptional period ) : 1 .
if @xmath116 is non - empty and @xmath116 is a prefix of @xmath119 , then the information bit @xmath124 of the message is the next bit of the correction message , i.e. , @xmath127;\ ] ] 2 .
if either @xmath116 is empty or @xmath116 is not a prefix of @xmath119 , then the information bit @xmath124 of the message is @xmath128 $ ] , in control bits the fact of starting correction phase is indicated ; 3 .
a transmission of the last bit of a correction message is indicated in control bits also .
figures [ regular - message ] and [ correction - message ] illustrate the rules determining the value of the information bit of a message . depending on the messages received at moment @xmath117 ,
the nodes change the basic data according the following rules . if @xmath129 ( a regular period ) , then @xmath130 if @xmath131 ( an exceptional period ) , then changes in the data are different :
if @xmath132 ( the end of the exceptional period ) , then @xmath133 and @xmath134 ; 2 .
if @xmath135 is a proper prefix of @xmath119 , then @xmath136 and @xmath137 ; 3 . if @xmath135 is not a prefix @xmath119 , then @xmath138 and @xmath137 ( it follows from the rules determining the messages of exceptional period that this case is possible if a correction message is aborted for another correction message ) .
note that , taking into account the received information and control bits , a node is able to update the data of neighbors , i.e. , to determine new values of the lists @xmath139 and @xmath140 . in particular , after receiving the last bit of a correction message from a neighbor @xmath15 , a node @xmath77 sets the participant @xmath141 to be the prefix of @xmath142 of length @xmath143 , where the binary representation of an integer @xmath143 is just the correction message @xmath144 received .
the last piece of the updated data to be determined is the candidate of a node .
the new value @xmath145 is the longest pl - minimal string in the set @xmath146 . initially , at moment @xmath25 , all participants @xmath147 and all prefixes of correction messages @xmath148 are empty and the candidates are the keys : @xmath149 for all @xmath77 .
note that the candidates , the participants and the correction messages are functions of the communication history in the node .
thus the described protocol satisfies the definition .
let s check the important property of the protocol , which has been indicated in the above informal discussion .
[ participant = pref(candidate ) ] a participant @xmath150 is a prefix of some candidate @xmath151 , @xmath152 .
induction by the moments of time .
initially , at @xmath25 , the proposition holds due to the rules of the protocol ( the first step is regular ) .
inductive step is a consideration of possible cases for determining a new value of a participant . a regular period , the information bit is non - empty . in this case
@xmath153 is a prefix of the candidate @xmath114 by the rule of determining the information bit . a regular period , the information bit is empty . in this case @xmath154 the end of an exceptional period : by definition
, @xmath150 is a prefix of @xmath114 . on other steps of an exceptional period , the participant does not change .
thus , by the induction hypothesis , it is a prefix of a previous candidate .
informally , the candidates and the participants provide a sequence of approximations to the minimal key .
therefore it is natural to expect that the sequence is monotone in each node .
the strings are compared in a partial order .
so the only weak monotonicity is possible .
the next proposition formalizes this intuition .
[ monotone0 ] a sequence of the candidates @xmath155 in a node @xmath77 is non - increasing ; moreover , for @xmath156 either @xmath114 is a prefix of @xmath151 or @xmath157 . by the rule of determining the next candidate as the longest possible pl - minimal string over the set that contains the previous candidate
, it follows from @xmath158 that @xmath114 is a prefix @xmath145 due to proposition [ min - pl ] . for arbitrary @xmath156
the statement follows by induction and transitivity of an order relation .
it follows from the rules of the protocol that a node sends the empty message if @xmath159 .
this condition remains true if the node has not received non - empty messages from its neighbor ( the data remain unchanged ) .
therefore , if @xmath159 , then the node @xmath77 is in a sleep state . below we prove the lemma .
[ weak - correctness ] at some moment of time the candidates and the participants in all nodes are equal to the pl - minimal key in a network .
proposition [ lex - consistent ] ( the pl order on the keys is consistent with the short - lex order on the identifiers ) , injectivity of the mapping from the identifiers to the keys , and lemma [ weak - correctness ] imply that the protocol described in the previous subsection is a message terminating protocol broadcasting the minimal identifier . before the proof of lemma [ weak - correctness ]
we prove several facts that will also be useful in evaluating of the running time of the protocol .
[ participant = pref(key ) ] at any moment of time @xmath19 the participant @xmath115 and the candidate @xmath114 of any node @xmath77 are prefixes of node keys .
induction by the moments of time .
it is sufficient to prove the statement for candidates only because participants are prefixes of candidates due to proposition [ participant = pref(candidate ) ] .
the base case holds since initially all candidates are keys .
show that if the statement holds at moment @xmath19 , then it also holds at moment @xmath117 .
participants of all nodes at moment @xmath117 are prefixes of candidates at earlier times due to proposition [ participant = pref(candidate ) ] . by the induction hypothesis they are prefixes of keys .
the candidate @xmath145 coincides either with @xmath114 or with a participant @xmath141 .
let @xmath160 be the minimal key in a network .
we will use the following fact about prefixes of the keys .
[ key - dichotomy ] let @xmath107 be a prefix of a key @xmath161 .
then either @xmath162 or @xmath107 is a prefix of @xmath160 .
look at the position that determines the result of comparison @xmath160 and @xmath161 : @xmath163 let @xmath107 be a prefix of @xmath161 such that @xmath107 is not a prefix of @xmath160 . then @xmath107 is longer than @xmath164
. therefore @xmath165 and @xmath166 .
thus @xmath167 .
[ kmin - stability ] if @xmath168 , then @xmath169 for all @xmath170 , @xmath171 .
show that @xmath169 implies @xmath172 . by definition , @xmath145 is the longest string among all pl - minimal strings in a set containing the candidate @xmath169 and participants .
the participants are prefixes of keys due to proposition [ participant = pref(key ) ] .
now apply proposition [ key - dichotomy ] .
we conclude from this corollary the following fact : if the candidate of a node @xmath77 equals the minimal key at moment @xmath173 , then the node @xmath77 will finish communication at some later moment . indeed ,
if the participant @xmath174 is not a prefix of the minimal key , then a correction message will be transmitted ( since the candidate remains unchanged , this correction message can not be aborted ) .
after that the node resumes transmission of remaining bits of the candidate ( which coincides with the minimal key ) .
the node finishes communication after sending the last bit of the candidate ( the minimal key ) .
due to proposition [ kmin - stability ] it is sufficient to prove that in any node the candidate equals the minimal key at some moment of time .
the proof is by induction on the distance @xmath175 from the node with the minimal key .
the base case @xmath176 is clear . in this case
the candidate is the minimal key from the very beginning . for induction step ,
assume that at time @xmath19 in all nodes at distance @xmath177 from the minimal one the candidates are equal to the minimal key .
take a node @xmath15 at distance @xmath175 from the minimal node .
the node @xmath15 has a neighbor @xmath49 at distance @xmath178 from the minimal node .
as shown above , the node @xmath49 finishes transmission of the minimal key at some moment @xmath179 . at this moment
, the participant @xmath180 equals the minimal key .
it means that @xmath181 is also the minimal key : from propositions [ greedy - compare ] and [ participant = pref(key ) ] we conclude that a prefix of a key is not lesser the minimal key in the pl order ; thus the minimal elements of the set @xmath182 are prefixes of the minimal key and the minimal key is the longest among them . applying the rule for determining the next candidate , we get @xmath183 . by similar arguments one can easily get a simple upper bound on the running time of the protocol , namely , @xmath10 , where @xmath2 is the diameter of the network and @xmath1 is the length of the minimal key . in the next subsection
we improve this bound .
assume that the candidate of a node equals the minimal key at some moment of time @xmath19 .
after that moment the node is transmitting the bits of the minimal key preceding by at most one correction message of the length @xmath184 . for inductive bound similar to the proof of lemma [ weak - correctness ] , we need a lower bound of the length of the prefix of the minimal key that has been sent by the node before time @xmath19 .
this bound is based on monotonicity of the minimal key propagation through a network .
if a node has sent a prefix of the minimal key , then the prefix will not be corrected in future .
we state this property formally and prove it .
but it is convenient to divide it in two statements .
[ monotone1 ] in each node the length of the longest common prefix of the candidate in the node and the minimal key does not decrease .
let @xmath185 , @xmath186 be two subsequent candidates in a node .
we prove that @xmath187 is a prefix of @xmath188 .
if @xmath189 is a prefix of @xmath188 , then the statement is trivial .
so , we assume that the opposite holds
. then @xmath190 due to proposition [ monotone0 ] .
let @xmath191 be @xmath192 . from the definition of the pl order we get @xmath193 by proposition [ participant = pref(key ) ] ,
the string @xmath188 is a prefix of a key @xmath161 .
thus , @xmath194 therefore @xmath195 is not a prefix of @xmath160 .
but it implies that @xmath196 is a prefix of @xmath107 , which is a prefix of @xmath188 .
a similar statement holds for participants .
but the proof is more tricky because participants can be arbitrary truncated .
[ monotone2 ] in each node the length of the longest common prefix of a participant in the node and the minimal key does not decrease .
let @xmath197 , @xmath198 be two subsequent participants in a node . by the rules of the protocol , in a regular period @xmath189
is a prefix of @xmath188 ( possibly , @xmath199 ) . at all moments of an exceptional period , with except for the last ,
@xmath199 too .
this last moment is the only nontrivial case in the proof . at the end of an exceptional period the rules of the protocol
give us @xmath200 now we prove that @xmath201 is a prefix of @xmath188 . by proposition [ participant = pref(candidate ) ] , @xmath115 is a prefix of some earlier candidate @xmath151 , @xmath152 .
a correction message is non - empty iff @xmath115 is not a prefix of @xmath114 .
it selects in the dichotomy of proposition [ monotone0 ] the case @xmath202 .
thus , @xmath203 and @xmath195 is a prefix of @xmath115 ( since @xmath115 is not a prefix of @xmath114 ) . in other words , @xmath204 .
since the candidate @xmath114 is a prefix of a key @xmath161 ( proposition [ participant = pref(key ) ] ) and @xmath205 @xmath195 is not a prefix of @xmath160 .
thus , @xmath201 is a prefix of @xmath206 .
now we introduce the main tool in the time analysis of the protocol : _ the delays _ of the minimal key transmission . by definition , the delay @xmath207 in the node @xmath15 at time @xmath19 is @xmath208 , if the node has not finished the minimal key transmission at moment @xmath19 , i.e. , if @xmath209 .
otherwise , the delay is undefined . by lemma [ weak - correctness ]
, the participant of a node equals the minimal key at some moment of time
. therefore the delays in the node are bounded .
let @xmath210 be the maximal delay in a node @xmath15 .
note that a sequence of the delays @xmath207 in a node is non - decreasing since the only one information bit is transmitted during a step of protocol .
also note that if the node @xmath15 has not finished transmission of the minimal key at time @xmath19 , then its participant @xmath142 has a common prefix of the length at least @xmath211 with the minimal key .
[ delay - lemma ] if nodes @xmath15 and @xmath49 are neighbors in a network , then @xmath212 where @xmath1 is the length of the minimal key .
let @xmath213 be the last moment when a node @xmath15 sends a wrong bit , i.e. @xmath214 after that , during an exceptional period @xmath215 , correction messages are transmitted ( therefore @xmath216 ) .
at the last moment of correction transmission @xmath217 the length of the participant decreases to @xmath218 , since by proposition [ monotone2 ] the length of the longest common prefix of a participant and the minimal key does not decrease but a new participant should be a prefix of the minimal key .
from a moment @xmath219 the delays do not increase : the next bit of the minimal key is transmitted at the next step . as the delay function is monotone ,
we get @xmath220 we upperbound @xmath221 via the delay @xmath222 of a neighbor @xmath49 of the node @xmath15 .
as it shown above , either the participant @xmath223 has a common prefix with the minimal key that is not shorter than @xmath224 , or this participant is the minimal key .
the participant is in the set of strings , which is used in determining the next candidate of the node @xmath15 . therefore the candidate @xmath225 has a common prefix with the minimal key that is not shorter than @xmath226 . on the other hand , this
common prefix coincides with @xmath227 by construction : the moment @xmath213 is regular , thus @xmath228 is a prefix of @xmath225 . from this
we get @xmath229 i.e. , @xmath230
. note that for @xmath231 the length of the candidate increases at step @xmath19 while the candidate differs from the minimal key .
if @xmath232 holds , then the length of the participant also increases while the participant differs from the minimal key .
thus @xmath233 the difference @xmath234 equals the length of the last correction message @xmath235 which is at most @xmath236 . finally , we get @xmath237 it completes the proof .
the running time of the protocol is @xmath238 .
the delay in the node with the minimal key is @xmath106 .
inequality implies that for a node at distance @xmath239 from the minimal node the delay is @xmath240 .
so the maximal delay over all nodes is @xmath241 .
thus each node finishes transmission of the minimal key in time @xmath238 and the running time of the protocol has the same bound . to complete the proof of theorem [ th : m - t - prot ] , we need to show that the above protocol constructs a spanning tree . after finishing the protocol ,
all nodes are in sleep states and their participants and candidates are equal to the minimal key .
one can define a spanning tree at this moment by the following rule : a node looks at the moment , when it has received the last bit of the minimal key , and chooses as a parent a node that has sent this bit ( a choice between several nodes satisfying this condition is arbitrary ) .
the node having the minimal key is the root .
we obtain in this way a directed graph ( a digraph ) .
fan - outs of all nodes in the digraph @xmath242 and there are no directed cycles in it ( to send the last bit of the minimal key , the node should receive it ; thus the moments of transition to a sleep state form a strictly decreasing sequence ) .
but for processor terminating protocol we need a more general construction .
not only the minimal key is propagating through a network .
it is possible that at some moment the protocol is stopped in a part of the network but the candidates in this part differ from the minimal key .
recall that no key is a prefix of another key . by proposition [ participant = pref(key )
] candidates are prefixes of keys .
thus a node can distinguish the moments of time when its candidate is a key as well as neighbors who sent the last bit of the key .
for a node @xmath15 we define a sequence of keys @xmath243 passing through the node at some moment of time
. the sequence is decreasing due to proposition [ monotone0 ] ( recall that the pl order is a linear order on the set of keys ) .
the sequence is non - empty , since initially the candidate is the key of the node . for a key @xmath244
we denote by @xmath245 the first moment @xmath19 such that @xmath246 . if @xmath247 for all @xmath19 , then @xmath248 . by definition , @xmath249 and other values of @xmath250
are either positive or infinite . if @xmath251 and @xmath252 , then @xmath253 . by the rule of determining the candidate
, we get @xmath254 for some neighbor @xmath255 of the node @xmath15 .
actually , it might be several neighbors satisfying this condition .
let choose one of them arbitrary .
we say that the node @xmath15 receives ( the last bit of ) the key @xmath244 from the node @xmath255 .
[ birth - seq ] @xmath256 .
the sequence of candidates is non - increasing in the pl order .
a participant of a node is a prefix of an earlier candidate of the node ( proposition [ participant = pref(candidate ) ] ) .
therefore the key @xmath244 is contained in the sequence of the keys passing through the node @xmath255 .
moreover , it follows from @xmath254 that @xmath256 .
note that @xmath255 is a function of the communication history @xmath257 in the node @xmath15 for @xmath258 .
thus , to define a function @xmath53 from the definition of a protocol constructing a spanning tree , we set @xmath259 , where @xmath244 is the minimal key among the keys that has passed through the node @xmath15 up to the moment @xmath19 .
formally , @xmath260 and @xmath261 is in the key sequence for the node @xmath15 such that @xmath262 and if @xmath263 , then @xmath264 . if @xmath265 , then @xmath266 . at the moment
@xmath38 , when the protocol is finished , the candidate of any node is the minimal key @xmath160 . for the node @xmath55
having the minimal key , the function @xmath267 defined above has the value @xmath268 . in other nodes
@xmath269 points to @xmath270 .
thus @xmath269 defines a digraph and fan - outs of all nodes are @xmath242 . along a directed path in the digraph ,
the values @xmath271 are strictly decreasing ( proposition [ birth - seq ] ) .
so , there are no cycles in the digraph and it is a rooted tree .
this argument completes the proof of theorem [ th : m - t - prot ] .
in this section we prove theorem [ th : p - t - prot ] . to define processor terminating protocols that broadcast the minimal identifier and construct a spanning tree we use the function @xmath53 described in the subsection [ span - tree ] and a structure relying on this function . for the proofs we also need auxiliary trees in a network .
these trees are defined in the next subsection .
after that we provide a protocol description and the proof of correctness .
an analysis of a protocol proposed below is based on a relation @xmath272 `` a node @xmath15 receives a key @xmath244 from a node @xmath77 '' and changes of the relation in time . for this purpose
we define a digraph @xmath273 including in it all edges in the form @xmath274 , @xmath275 .
let @xmath276 be an edge of the digraph @xmath273 .
it means by definition that @xmath277 for some key @xmath244 and the key @xmath244 is the minimal ( in the pl order ) key that has passed through the node @xmath15 up to a moment @xmath19 .
we say that @xmath244 is the _ edge key _ and @xmath245 is the _ birth date _ of the edge .
initially , @xmath278 consists of isolated roots . at the end , as it was shown above , it is a spanning tree . for the proof of correctness we will need a stronger property .
[ forest - lemma ] the digraph @xmath273 is a spanning forest for any @xmath19 .
the fan - outs of the nodes are @xmath242 in this digraph by construction
. therefore the only cycles in the underlying undirected graph are the directed cycles in the digraph @xmath273 . to prove that there are no directed cycles we choose an appropriate monotonicity property .
a transition from an edge @xmath276 to an edge @xmath279 in the digraph @xmath273 decreases either the edge key ( the sequence of keys is monotone in each node ) or the birth date ( proposition [ birth - seq ] ) .
thus the lengths of directed paths in @xmath273 are bounded and the digraph does not contain directed cycles .
connected components of the digraph @xmath273 are rooted trees ( the roots are nodes with fan - out 0 ) .
we call them _ dynamic trees _ in the network . by @xmath280
we denote the connected component that has a root @xmath55 with the key @xmath281 . in the proof below
we will use also _ static _ trees corresponding to keys . let @xmath281 .
then the vertices of the rooted tree @xmath282 are the nodes such that @xmath252 .
the node @xmath55 is the root , a parent of a node @xmath15 is the node @xmath255 .
since @xmath245 is decreasing on a move from a child to a parent , the graph @xmath282 is a tree indeed .
we call these tree static because they are defined by the whole communication history . for some keys the tree @xmath282 consists of the root with the key @xmath244 only . for the minimal key @xmath160
the static tree @xmath283 coincides with the spanning tree constructed by the message terminating protocol from the previous section . on figure [ gt ]
two components of a dynamic tree @xmath273 are shown .
figures [ tk1][tk2 ] show the corresponding parts of static trees @xmath284 and @xmath285 .
the roots of the trees are white circles .
links of the network that are not included in trees are pictured by dashed lines .
it follows from the definition that a node @xmath15 is included in a tree @xmath286 iff the candidate of the node equals the key @xmath244 at some moment .
[ conserve ] moreover , an edge @xmath276 of a static tree @xmath282 is in the digraph @xmath273 at all moments @xmath19 such that the key @xmath244 is the pl - minimal key among the keys passed through the node @xmath15 .
this remark will be used in the proof below .
the required protocol is combined from three protocols @xmath287 , @xmath288 , @xmath289 .
its operation is divided in stages . during each stage
the combined protocol performs one step of each protocol @xmath287 , @xmath288 , @xmath289 in the indicated order .
thus , each protocol is operating 3 times slower
. it does not affect the asymptotic bound in theorem [ th : p - t - prot ] .
in the sequel , we refer to stages of the combined protocol . after performing @xmath19 stages of the combined protocol each protocol @xmath287 , @xmath288 ,
@xmath289 has performed @xmath19 steps .
the protocol @xmath287 is the message terminating protocol described in the previous section .
it broadcasts the minimal identifier ( formally , it broadcasts the minimal key , but an identifier can be recovered by a key ) .
the second protocol @xmath288 maintains the spanning forest @xmath273 defined in subsection [ span - tree ] . in the protocol
@xmath288 nodes send messages of two types : `` i m a child '' and `` i m not a child '' .
the first message is sent by a node @xmath15 at that moment when its candidate @xmath290 becomes a key @xmath251 .
the node @xmath15 sends this message to the neighbor @xmath255 , which sent the last bit of the key @xmath244 .
if the node @xmath15 has a parent @xmath291 at this moment ( i.e. , in the tree @xmath292 ) , then the node @xmath15 also sends the message of the second type to the node @xmath49 .
otherwise , the node sends no message ( formally , the empty message ) . since a step of the protocol @xmath287 is followed by a step of the protocol @xmath288 , before a step of the protocol @xmath289 nodes have correct information about their childs in the forest @xmath273 .
in the finalizing protocol @xmath289 each node @xmath15 checks the _ local termination condition _ : communication with the neighbors in the protocol @xmath287 is finished , the empty message is received in the protocol @xmath288 , and the candidate and all participants of the node @xmath15 are equal to some key @xmath244 . in the protocol
@xmath289 nodes send messages through the edges of the spanning forest @xmath273 .
there are two types of messages : the confirmation ( of local finish ) message and the termination message .
a node also can send no message .
divide nodes in a dynamic tree @xmath280 into three groups : the root has fan - out @xmath106 , the intermediate nodes have childs ( fan - in @xmath293 ) and the leaves have no childs ( fan - in @xmath106 ) . communication in the protocol @xmath289 starts in leaves .
a leaf sends the confirmation message to its parent if the local termination condition becames true for the leaf .
an intermediate node collects confirmation messages from its childs .
it sends the confirmation message to its parent if it has received confirmations from all its childs and the node itself satisfies the local termination condition .
it is possible that the local termination condition changes the value from true to false : a node exchanged a key with all neighbors but the key is not the minimal one . if the candidate of a node is decreased in the pl order after a step of the protocol @xmath287 or the node receives the message `` i m not a child '' after a step of the protocol @xmath288 , then the node removes all confirmations received .
in further operation , it ignores all incoming confirmations until its candidate becomes a key .
the root also collects confirmations from its childs and checks the local termination condition .
the root sends the termination message to all its childs if all childs of the root have sent confirmations and the root satisfies the local termination condition .
the termination messages broadcast from parents to childs through the tree . after sending the termination message a node goes to the final state .
the combined protocol is processor terminating : if a node is in the final state at time @xmath19 , then the protocol @xmath287 has finished at time @xmath19 ; all nodes are in the final state at some moment of time . to prove the first statement of the lemma
, we show that if a root of a dynamic tree @xmath280 sends the termination message , then the tree is spanning .
we say that a confirmation is passed through an edge @xmath276 with the edge key @xmath294 if at some moment @xmath295 the node @xmath15 sends the confirmation message to the node @xmath77 .
nodes remove confirmations if their candidates decrease or the message `` i m not a child '' is received .
thus it is sufficient to take into account only those confirmations that were passed through the edges of the tree @xmath280 with the current edge keys ( but an edge can be in an another tree at the moment of passing confirmation ) .
we conclude from this observation that the root does not send the termination message at moment @xmath19 if there exists an edge @xmath276 of the tree @xmath280 such that no confirmation has passed through the edge . indeed
, the keys in the roots of trees containing the edge @xmath276 form a non - increasing sequence . therefore , while the edge @xmath276 is in the tree @xmath280 , this tree also contains all edges of the path from @xmath77 to the root and no confirmation was passed through edges of this path .
it means that the root does not send the termination message .
if the tree @xmath280 with the root @xmath55 is not a spanning tree , then the following cases are possible . 1 . [ cand - less ] @xmath296 .
2 . [ key - geq ] otherwise , @xmath297 due to monotonicity of the candidates in a node . but
it is possible that the tree @xmath280 includes edges whose keys differ from @xmath244 .
[ good - neighbors ] @xmath297 and the edge key is @xmath244 for each edge in the tree @xmath280 .
we assume that the tree is not spanning .
thus , there exist `` external links '' : between nodes in the tree and outside it ( the network itself assumed to be connected ) .
we call _ tree neighbors _ the endpoints of the second type .
+ suppose that the following condition holds : there is no neighbor of the tree @xmath280 such that its candidate is @xmath244 at some earlier moment .
4 . [ bad - neighbors ]
the last case covers the remaining option .
namely , there exists a neighbor of the tree @xmath280 such that its candidate is @xmath244 at some earlier moment .
* case [ cand - less ] . *
the local termination condition violates in the root @xmath55 in this case : the candidate differs from a key .
thus the root does not send the termination message . for all other cases
we indicate an edge in the tree @xmath280 such that no confirmation has been passed through the edge . * case [ key - geq ] . * by definition of a digraph @xmath273
, the root @xmath55 has not been in any tree @xmath298 , @xmath152 , @xmath299 .
it implies that the only possible value of the edge key for an edge @xmath300 is @xmath244 .
thus , there are two edges in the tree @xmath301 , @xmath302 such that the first edge key is @xmath299 , the second edge key is @xmath244 , and @xmath303 .
indeed , both keys passed through the node @xmath15 up to the moment @xmath19 and inequality @xmath304 contradicts the definition of a digraph @xmath273 .
show that no confirmation has passed through the edge @xmath302 .
suppose the opposite : the node @xmath15 has sent the confirmation message to the node @xmath49 at a moment @xmath305 .
the local termination condition implies @xmath306 .
but @xmath307 , otherwise the node @xmath15 has removed this confirmation up to the moment @xmath19 .
it implies that the key sequence in the node @xmath77 contains the key @xmath308 and the edge key for the edge @xmath301 is less than @xmath294 .
it contradicts the assumption .
this contradiction completes the argument for this case .
* case [ good - neighbors ] .
* suppose that a confirmation passes through an edge @xmath301 .
then the node @xmath77 satisfies the local termination condition .
it implies that candidates of all neighbors are @xmath244 . from this observation
we conclude that nodes in the tree @xmath280 having `` external links '' have not sent the confirmations .
( recall that in the case into consideration candidates of any neighbor of the tree @xmath280 differ from @xmath244 at any moment up to @xmath19 . ) * case [ bad - neighbors ] .
* recall that in this case there are nodes outside the tree @xmath280 such that their candidates were @xmath244 at some earlier moments .
all these nodes are in the static tree @xmath282 .
choose among them the neighbor @xmath49 of the tree @xmath280 that has leaved the dynamic tree with the key @xmath244 at the earliest moment .
let @xmath309 be the path in the static tree @xmath282 from @xmath49 to the root @xmath55 .
show that no node on the path has sent the confirmation through an edge of this path ( with the edge key @xmath244 ) . while the node @xmath49 is in the dynamic tree with the key @xmath244
, the conditions of the case [ good - neighbors ] hold : there is no neighbor of the tree such that a candidate of the neighbor was @xmath244 .
it guarantees that the node @xmath310 has not sent confirmation .
suppose that confirmations were sent by nodes on the path from @xmath49 to @xmath55 at earlier moments than @xmath19 . among the nodes , which sent these confirmations ,
take @xmath311 the nearest to the node @xmath49 .
as it was shown above , the node @xmath311 differs from @xmath49 .
therefore there is the preceding node @xmath312 on the path .
if a node @xmath312 is a child of a node @xmath311 in @xmath313 , then the node @xmath311 has not received the confirmation from @xmath312 . otherwise , the candidate @xmath314 is lesser than @xmath244 in the pl order .
it implies that either the node @xmath311 does not finish communication or it received the message `` i m not a child '' from @xmath312 .
both variants contradict the sending of confirmation by the node @xmath311 .
so , no confirmation has been passed through the edges of the path .
take the minimal key @xmath315 that has been passed through the node @xmath316 .
if @xmath317 , then the edge @xmath318 is in the tree @xmath280 ( see remark [ conserve ] ) and no confirmation has been passed through the edge . otherwise , the local termination condition does not hold in the root @xmath55 because the candidate of @xmath316 is lesser than @xmath244 .
this completes our case analysis .
so , we have proved that if the tree @xmath280 is not spanning , then its root does not send the termination message . to prove the second statement of the lemma , note that the protocol @xmath287 terminates at some time @xmath319 .
the graph @xmath313 is a spanning tree .
all candidates @xmath181 and all participants @xmath320 coincide with the minimal key .
thus all nodes finish communication in the protocol @xmath287 .
it means that in the protocol @xmath289 the confirmations are transmitted upward .
when the root collects the confirmations from all its childs , it sends the termination message .
the termination message will broadcast downward .
the running time of the protocol @xmath287 is @xmath321 due to theorem [ th : m - t - prot ] .
after that all nodes have finished the communication in the protocols @xmath287 , @xmath288 and confirmations will broadcast through the spanning tree without delays as well as termination messages .
so , an additional time to finish the combined protocol is @xmath322 , where @xmath175 is the depth of the spanning tree constructed . to bound the depth of the tree , note that all nodes receive the minimal key in time @xmath238 .
the depth is not greater than this value .
indeed , the moments of receiving the minimal key are different along a path from the root to a node . after receiving the minimal key
the node does not change a parent .
thus , the running time of the combined protocol is @xmath323
the authors are thankful to anonymous referees for their numerous valuable comments and suggestions . | we present fast deterministic distributed protocols in synchronous networks for leader election and spanning tree construction .
the protocols are designed under the assumption that nodes in a network have identifiers but the size of an identifier is unlimited .
so time bounds of protocols depend on the sizes of identifiers .
we present fast protocols running in time @xmath0 , where @xmath1 is the size of the minimal identifier and @xmath2 is the diameter of a network .
we study deterministic distributed protocols for leader election and spanning tree construction in a synchronous network .
these problems play important role in distributed computation theory ( e.g. , see books @xcite ) .
it is well - known that for anonymous networks ( processors are indistinguishable ) there are no deterministic protocols solving these problems .
so we assume that processors in a network ( _ nodes _ ) have unique names ( _ identifiers _ ) . under this assumption ,
a natural way for leader election is _ the minimal identifier broadcast _ : if all nodes in a network know the value of the minimal identifier then the node having this identifier is elected .
the previous study of these problems was based on the assumption that the identifiers are rather short ( say , the nodes are numbered in the range @xmath3 $ ] , where @xmath4 is the size of a network ) and messages are rather long ( the message size is @xmath5 ) . in these settings
near - optimal protocols are known for these problems and for a more general problem of minimum spanning tree construction @xcite .
we address to an intermediate situation between the anonymous case and the well - numbered case .
namely , we assume that the nodes have identifiers but there is no a priori bound on the size of the identifiers .
also we restrict the communication speed by @xmath6 bit per message .
we call this model _ the unbounded identifier model_. here we consider deterministic protocols only and do not analyze possible errors in the process of information transmission . in this model bounds for the bit complexity of protocols ( the total number of bits sent ) are known . for asynchronous ring and chain networks rather close upper and lower bounds of the bit complexity
were obtained in @xcite . in @xcite protocols with near - optimal bit complexity bounds
were presented for arbitrary synchronous networks .
more exactly , for each monotone unbounded function @xmath7 there is a protocol that constructs a spanning tree and sends @xmath8 bits , where @xmath9 is the number of links in a network , @xmath4 is the number of nodes .
these protocols construct rooted trees and the root is the node with the minimal identifier .
so the protocols also elect a leader in a network . in this paper
we are interested in time bounds of the protocols in the unbounded identifier model . in the unbounded identifier model the running time of a protocol
depends on the sizes of identifiers .
we denote by @xmath1 the length of the shortest identifier .
( we always assume that a shorter identifier is lesser . )
straightforward modifications of known protocols for the unbounded identifier model run in time @xmath10 . here
@xmath2 is the diameter of a network .
we present better protocols .
we present the message terminating protocol and the processor terminating protocol for leader election and spanning tree construction .
the definitions for these types of protocols are taken from the paper @xcite .
the message termination means that at some moment of time all nodes are in a _
sleep state_. in such a state a node does not send messages but can change the state after receiving a suitable message .
the processor termination means that at some moment of time all nodes are in the _
final state_. this state can not be changed by any message . in other words , a node in the final state is off - line .
processor terminating is a stronger property .
it is important , e.g. , for compositions of protocols .
the following two results are our main contribution .
[ th : m - t - prot ] there exists a message terminating protocol that broadcasts the minimal identifier and constructs a rooted spanning tree in time @xmath11 , where @xmath1 is the length of the minimal identifier and @xmath2 is the diameter of the network .
[ th : p - t - prot ] there exists a processor terminating protocol that broadcasts the minimal identifier and constructs a rooted spanning tree in time @xmath11 , where @xmath1 is the length of the minimal identifier and @xmath2 is the diameter of the network .
these protocols are modifications of flooding and echo protocols designed for the well - numbered model .
to speed up a flooding protocol in the unbounded identifier model , we introduce a suitable encoding of identifiers ( see section [ id->key ] below ) and add a correction phase to the process of information transmission .
note also that the protocols presented in this paper remain correct in the asynchronous case .
but time bounds are valid for the synchronous case only .
the rest of paper is organized as follows . in section [ defs ]
we give formal definitions of our model and the problems solved by protocols .
section [ informal ] contains an overview of our construction .
section [ min - broadcasting ] contains the proof of theorem [ th : m - t - prot ] : a detailed description of the message terminating protocol that broadcasts the minimal identifier in a network , the proof of correctness , the upper bound of the running time and spanning tree construction based on this protocol . in section
[ finalizing ] we present the proof of theorem [ th : p - t - prot ] based on a suitable modification of an echo protocol combined with the previous protocol that broadcasts the minimal identifier . |
the binary ferromagnetic fe - co alloy has been known for a long time as a system with the maximum spontaneous magnetization among the transition - metal systems .
@xcite its basic magnetic properties , such as the concentration dependence of the alloy magnetization ( the slater - pauling curve ) in the ground - state body - centered cubic ( bcc ) structure , have been reproduced successfully by _ ab initio _ electronic structure calculations in a number of studies .
@xcite this system attracted renewed interest several years ago , after a theoretical prediction of a giant uniaxial magnetic anisotropy energy ( mae ) of the body - centered tetragonal ( bct ) fe@xmath2co@xmath3 alloys .
@xcite the uniaxial mae is defined quantitatively as the difference of total energies for the magnetization direction parallel to the tetragonal @xmath4 and @xmath5 axis , @xmath6 .
the calculated mae reached high values , @xmath7ev / atom , obtained however only in a narrow range of the co concentration , @xmath8 , and of the tetragonal strain , @xmath9 .
the combination of the high mae with the high magnetization makes the bct fe - co system a promising material for fabrication of ferromagnetic thin films with a perpendicular magnetic anisotropy ( the magnetic easy axis perpendicular to the film plane ) , which might be relevant as high - density magnetic recording media .
most of experimental realizations of this tetragonal system employed the possibility to grow epitaxially strained fe - co alloy films of a varying composition on different non - magnetic transition - metal substrates with the face - centered cubic ( fcc ) structure , such as pd(001 ) , rh(001 ) , and ir(001 ) .
@xcite the different lattice parameters of the substrates enable one to scan a relatively wide interval of the @xmath10-ratio of the fe - co films , namely @xmath11 ( note that @xmath12 and @xmath13 refer respectively to the ideal bcc and fcc lattices ) .
the measurements of the _ in situ _ magneto - optical kerr effect @xcite ( moke ) confirmed the theoretical prediction of ref . on a qualitative level ; in particular , a strong perpendicular magnetic anisotropy of the films was observed for compositions and tetragonal distortions in a rough agreement with the calculated trends of the bulk @xmath14 . however , systematic quantitative experimental studies of the mae of the films have not been performed yet ; available values of the @xmath14 for a few selected systems @xcite are appreciably smaller than the calculated ones .
the missing information on the mae has partly been compensated by measurements of x - ray magnetic circular dichroism ( xmcd ) spectra , which provide orbital magnetic moments of both alloy constituents .
@xcite the latter measurements , performed on the rh(001 ) substrate corresponding to @xmath15 , revealed enhanced orbital moments in fe@xmath2co@xmath3 films around @xmath16 , in a close relation to the predicted maximum of the bulk uniaxial mae .
nevertheless , the existing agreement between the theory and the experiment should be taken with caution .
the original approach of ref .
employed the so - called virtual crystal approximation ( vca ) , in which the true species of a random binary alloy are replaced by a single element with an effective atomic number given by the alloy composition .
the vca has been implemented in various _ ab initio _ methods and used for a number of alloys of neighboring elements in the periodic table .
@xcite its application to the bcc fe - co and fcc co - ni systems @xcite yields concentration trends of the spin magnetic moment in a very good agreement with experiment ; @xcite the orbital magnetic moments seem to be described reliably in the vca as well .
however , a recent first - principles study of the random bct fe@xmath2co@xmath3 alloys , @xcite based on a supercell technique applied to a few special co concentrations ( @xmath17 , 0.625 , and 0.75 ) , has shown that a realistic treatment of the chemical disorder can reduce the mae by a factor of 1.5@xmath183 as compared to the simple vca .
the main purpose of this work is to present and discuss the results of a systematic _ ab initio _ calculation of the mae for the disordered bulk tetragonal fe - co systems obtained by using the coherent potential approximation ( cpa ) as a basic tool of the theory of metallic alloys .
@xcite for homogeneous bct alloys , we compare results of the vca and the cpa , show the big difference between them , and identify the underlying physical mechanism in terms of the electronic structure . moreover , we investigate the effect of a complete and an incomplete atomic long - range order on the mae ; this study is motivated by the existing prediction of a large uniaxial mae in stoichiometric perfectly ordered tetragonal feco and feco@xmath19 systems .
@xcite similar theoretical studies have so far been done mainly for tetragonal fept alloys @xcite and very recently also for the feco alloy .
@xcite these works are confined to stoichiometric equiconcentration alloys ; in general , their results reveal a decrease of the mae due to imperfect chemical ordering .
all calculations of this study employed a bct lattice with the wigner - seitz @xmath20 radius equal to that of pure bcc iron , @xmath21 where @xmath22 m denotes the bohr radius .
the neglect of volume relaxations on the mae in a broad range of lattice parameters and of alloy concentrations is justified by a more general study of ref . within the vca .
this structure model
the volume conserving tetragonal distortion ( the bain path)coincides with that used in the original study @xcite and in the recent study of the partially ordered fept alloy .
@xcite the effect of atomic long - range ordering has been studied for the case of @xmath1 ( cuau ) order relevant for compositions not far from the equiconcentration fe@xmath23co@xmath23 system .
the bct structure is partitioned in two simple tetragonal sublattices ( alternating atomic planes perpendicular to the tetragonal @xmath5 axis ) . for the fe@xmath2co@xmath3 system ,
an additional concentration variable @xmath24 is introduced , such that @xmath25 , which defines the chemical composition of both sublattices : fe@xmath26co@xmath27 in the fe - enriched planes and fe@xmath28co@xmath29 in the co - enriched planes .
note that the homogeneous solid solution corresponds to @xmath30 , while the other end of the @xmath24-interval describes the maximum @xmath1 order compatible with the given total co concentration @xmath31 .
this model is a natural generalization of the model used for the stoichiometric fept systems . @xcite
the eletronic structure calculations were done by means of the tight - binding linear muffin - tin orbital ( tb - lmto ) method in the atomic sphere approximation ( asa ) @xcite with a full inclusion of relativistic effects by solving the one - electron dirac equation @xcite with the @xmath32-basis of the valence orbitals .
the effective spin - polarized potential was constructed in the local spin - density approximation ( lsda ) using the parametrization of the exchange - correlation term according to ref . .
the effects of alloying were treated in the cpa for all systems ; @xcite for the homogeneous random alloys ( @xmath30 ) , the simple vca was used as well .
the reliable evaluation of the mae in metallic systems represents a difficult task , especially for cubic @xmath33 transition metals owing to the weakness of the spin - orbit interaction and the high symmetry of cubic structures .
@xcite the situation is more favorable for uniaxial systems ( tetragonal , trigonal , hexagonal ) @xcite and for systems with @xmath34-electrons , see ref . for a review . in a number of existing studies , including the very recent ones , @xcite the magnetic force theorem @xcite is used and the total energy difference @xmath14 is approximated by the change in the sum of occupied valence one - particle eigenvalues . in this study , we go beyond this approximation and evaluate the mae directly from the total energies of the fully self - consistent solutions for both magnetization directions , @xcite which requires high accuracy in total - energy calculations .
we have used uniform sampling meshes of about @xmath35 @xmath36-points for averages over the full brillouin zone ( bz ) of the bct lattices , while about @xmath37 @xmath36-points have been used for the bz of the simple tetragonal bravais lattices of the @xmath1 ordered systems .
the total energies were converged to 0.1 @xmath38ev / atom for all systems ; this accuracy is sufficient in view of the resulting values of @xmath14 ( see below ) .
the dependence of the uniaxial mae of the random fe@xmath2co@xmath3 alloys on the chemical composition and the @xmath10-ratio , calculated in the vca and in the cpa , is shown in fig .
[ f_2d_mae ] ; for better transparency , only the positive values of the @xmath14 are displayed in the plots , while the cases with negative @xmath14 have been omitted ( marked by white color ) .
the vca results ( fig .
[ f_2d_mae]a ) agree nicely with both previous studies ; @xcite in particular , a sharp maximum of the mae , @xmath39ev / atom , obtained for @xmath40 and @xmath41 , is clearly visible .
the maximum mae of @xmath42ev / atom for @xmath40 and for @xmath10 between 1.22 and 1.24 was reported in ref .
( with a slight sensitivity to the particular model of the volume relaxation employed ) and practically the same values were obtained in the original vca study .
@xcite minor quantitative differences might be ascribed to different computational schemes employed .
the mae in the cpa ( fig .
[ f_2d_mae]b ) exhibits a similar trend as in the vca ; however , the cpa maximum is significantly smaller , @xmath43ev / atom .
the latter value is obtained for @xmath40 and @xmath44 , but the maximum is very flat and essentially the same @xmath14 values can be found in a broad region of @xmath45 and @xmath46 , see fig .
[ f_2d_mae]b .
the maximum @xmath14 in the cpa is reduced by a factor of four as compared to that in the vca , which is an even stronger reduction than that reported in ref . .
the small mae values in the cpa might be used to explain the difference between the large @xmath14 values predicted in the vca and the much smaller values inferred from measurements on thin fe - co films .
@xcite undoubtedly , the better description of the chemical disorder in the cpa as compared to the vca contributes to the reduction of the mae of prepared alloy thin films . however , a thorough quantitative analysis of this point is impossible at present and it could even be misleading because of the well - known uncertainty of the lsda to yield quantitatively correct maes in @xmath33 transition - metal systems .
for this reason , an analysis of the trends seems to be more appropriate , see , e.g. , ref . and references therein . co@xmath47 alloy with magnetization along @xmath48 axis as functions of the @xmath10-ratio : the alloy orbital moment in the vca and the species - resolved orbital moments in the cpa .
[ f_orbmm ] ] in general , the lsda underestimates both the maes and the orbital magnetism .
@xcite a counterexample to this rule is represented by the mae of the ordered fept alloy , @xcite which is overestimated in the lsda probably owing to an interplay of intraatomic coulomb correlations , weak exchange splitting and strong spin - orbit interaction of pt atoms .
@xcite the present results for the bct fe@xmath2co@xmath3 alloys seem to confirm the general rule as can be documented by two facts .
first , the effective mae of strained fe - co thin films , given by the difference of the bulk @xmath14 and the magnetostatic shape anisotropy energy @xmath49 , comes out positive for alloys in the vicinity of @xmath16 and @xmath50 ( near the maximum of @xmath14 in the cpa ) .
this situation would thus lead to an out - of - plane orientation of the thin - film magnetization , in a qualitative agreement with existing experiments .
@xcite note that @xmath51 , where @xmath52 is the permeability of the vacuum and where @xmath53 and @xmath54 denote , respectively , the alloy magnetic moment per atom and the atomic volume . in the present case , @xmath55 , where @xmath56 is the bohr magneton , one obtains @xmath57ev / atom , which lies below the bulk mae ( @xmath58ev / atom ) .
however , for the case of @xmath59 and @xmath60 , corresponding roughly to the fe@xmath23co@xmath23 films grown on the pd(001 ) substrate , our calculated values are @xmath61ev / atom , @xmath62 , and @xmath63ev / atom , which indicates an in - plane orientation of the magnetization , in contrast to the out - of - plane orientation observed at low temperatures .
@xcite second , the calculated orbital magnetic moments in the bulk fe@xmath2co@xmath3 alloys , plotted in fig . [ f_orbmm ] for @xmath64 , are appreciably smaller than the values obtained from the measured xmcd spectra by employing the sum rules .
@xcite in particular , the measured values for films with @xmath40 and @xmath41 , namely , @xmath65 and @xmath66 , exceed their bulk theoretical counterparts by a factor of three .
a similar relation of the experiment and the lsda theory ( quantified roughly by a factor between 1.5 and 2 ) was obtained for bcc fe - co alloys .
@xcite moreover , the data in fig .
[ f_orbmm ] prove that the vca and the cpa yield qualitatively different trends of the orbital magnetism versus the @xmath10-ratio : the pronounced maximum at @xmath41 in the vca is replaced by very flat maxima of both local orbital moments in the cpa .
this difference indicates that applicability of the vca to all details of magnetism of the fe - co system is limited , being confined probably only to the cubic structures .
@xcite in view of the above mentioned problems to evaluate reliably the maes and the orbital magnetic moments , we have not attempted to recalculate our results by techniques going beyond the lsda , @xcite but have focused on the strong difference between the vca and the cpa and on the role of disorder on the mae in the tetragonal fe - co systems .
the giant mae obtained in the vca was ascribed originally to the changes in the band structure accompanying the tetragonal distortion of the effective feco crystal .
@xcite the explanation rests on a coincidence of two particular eigenvalues at the @xmath67 point , which occurs in the minority spin ( spin - down ) channel . for a special alloy composition and a special @xmath10-ratio
, this coincidence takes place just at the fermi energy corresponding to the particular filling of the spin - down band .
the large contribution of this eigenvalue pair to the mae can be understood in terms of the second - order perturbation theory , in which the effect of the weak spin - orbit interaction can be safely included .
@xcite in this approach , the enhanced value of the @xmath14 of the bulk fe - co alloys can be explained due to the coupling between @xmath68 and @xmath69 states mediated by the orbital momentum operator @xmath70 , see ref . for details .
the physical origin of the perpendicular magnetic anisotropy of the strained fe - co films is thus similar to that of au covered co monolayers on an au(111 ) substrate .
@xcite in the latter case , however , the resulting positive mae is caused by the @xmath70 coupling between @xmath71 and @xmath72 states .
co@xmath47 alloy with @xmath73 as functions of the energy .
the dotted vertical line denotes the position of the fermi level .
[ f_dos ] ] the usual band structures are not relevant for the electronic structure of random alloys , for which more appropriate quantities , such as the densities of states ( dos ) and the bloch spectral functions ( bsf ) , have to be studied .
@xcite motivated by the above explanation of the giant mae , we present these quantities for the random bct fe@xmath2co@xmath3 alloys in the cpa , evaluated without the spin - orbit interaction , i.e. , in the scalar - relativistic approximation .
@xcite the spin - polarized local doss are shown in fig .
[ f_dos ] for the system with @xmath40 and @xmath73 .
the shapes of the individual doss prove a very weak disorder in the majority spin ( spin - up ) channel , whereas the spin - down band exhibits a regime of a stronger scattering .
this type of spin - dependent disorder was found in the bcc fe - co system a long time ago ; @xcite it is responsible , e.g. , for the observed concentration trend of the residual resistivity .
@xcite note that the strong disorder is present in the spin - down @xmath0-band , which is only partially occupied and which thus contributes a lot to the alloy total energy and , consequently , to the mae .
co@xmath47 alloy at @xmath74 as functions of the energy for three values of the @xmath10-ratio .
the dotted vertical lines denote positions of the fermi levels . for details , see the text .
[ f_bsf ] ] the spin - polarized bsfs at the @xmath67 point for the fe@xmath75co@xmath47 alloy with three tetragonal distortions are plotted in fig .
[ f_bsf ] .
the bsfs were resolved according to the irreducible representations of the point group @xmath76 of the bct lattice .
@xcite the odd ( ungerade ) representations , belonging to @xmath77 orbitals , have been omitted in the plots since their contribution is negligible in the displayed energy interval .
the relevant even ( gerade ) representations are : @xmath78 ( orbitals @xmath20 and @xmath79 ) , @xmath80 ( @xmath68 ) , @xmath81 ( @xmath69 ) , and @xmath82 ( @xmath71 and @xmath72 ) .
one can see that all spin - up bsf s , consisting of narrow lorentzian peaks , can be interpreted as bloch - like eigenstates with finite lifetimes due to a weak disorder .
the spin - down bsf s above the fermi energy ( mainly @xmath78-like ) reflect effects of strong disorder ( non - quasiparticle behavior ) ; the profiles at and below the fermi energy are lorentzian peaks again
. however , their widths are clearly bigger than the spin - up widths , in agreement with the spin - dependent disorder manifested in the dos .
the maximum of the mae in the vca is found for an alloy with @xmath64 and with the tetragonal strain close to @xmath73 ; its spin - down bsf in the cpa ( fig .
[ f_bsf ] , middle panel ) shows an analogy of the coincidence of the @xmath80 ( @xmath68 ) and @xmath81 ( @xmath69 ) levels at the fermi energy .
however , the disorder - induced smearing of both peaks is at least as big as their separation , which suppresses the contribution of this eigenvalue pair to the mae in the framework of the second - order perturbation theory .
this feature proves that the minority - spin fe - co disorder is strong enough to reduce significantly the mae , which explains the very big difference between the vca and the cpa results , see section [ sss_mae ] . the adopted model of the
partially ordered tetragonal fe@xmath2co@xmath3 alloys , namely , the model of two sublattices with chemical compositions fe@xmath26co@xmath27 and fe@xmath28co@xmath29 ( section [ s_mcd ] ) and the use of the cpa ( in contrast to supercell techniques ) enable one to study the mae as a function of three continuous variables : @xmath31 , @xmath24 , and @xmath10 .
such a full three - dimensional scan is , however , computationally very demanding ; we have thus confined our study to co concentrations close to @xmath64 , for which the maximum mae was found in the homogeneous bct alloys both in the vca and in the cpa .
the dependence of the calculated @xmath14 on the tetragonal distortion for @xmath64 is shown in fig .
[ f_maevsr ] for three degrees of the @xmath1 order .
one can see that the very flat low maximum @xmath43ev / atom for the completely disordered alloy ( @xmath30 ) treated in the cpa is replaced by a significantly enhanced pronounced maximum @xmath83ev / atom obtained for the alloy with the maximum order ( @xmath84 ) compatible with the given co concentration ( @xmath64 ) .
the latter maximum is about two times smaller than that in the vca ( @xmath39ev / atom ) , which can be ascribed to the fact that the system with @xmath64 and @xmath84 has one sublattice without disorder ( pure co ) , whereas the other one is disordered , containing the rest co and all fe atoms .
the mae of the system with an intermediate degree of the @xmath1 order ( @xmath85 ) exhibits the maximum and the trend very close to the fully disordered alloy .
co@xmath47 alloy as a function of the tetragonal strain @xmath10 for different degrees of the @xmath1-like atomic order : the maximum order ( @xmath84 , diamonds ) , an intermediate order ( @xmath85 , full circles ) , and the complete randomness ( @xmath30 , crosses ) , all treated in the cpa . for a comparison , the mae of the completely random alloy in the vca is displayed as well ( triangles ) .
[ f_maevsr ] ] co@xmath3 alloys as a function of the degree of the @xmath1-like atomic order : for @xmath17 and @xmath86 ( full circles ) , and for @xmath64 and @xmath87 ( open squares ) .
[ f_maevsy ] ] the detailed dependence of the @xmath14 on the concentration variable @xmath24 is presented in fig .
[ f_maevsy ] for two cases : the stoichiometric alloy ( @xmath17 , @xmath41 ) and the previous off - stoichiometric alloy ( @xmath64 , @xmath88 ) .
the particular values of the @xmath10-ratio chosen in both cases correspond to the maximum @xmath14 value obtained for alloys with the maximum order , i.e. , for @xmath64 , @xmath84 ( fig .
[ f_maevsr ] ) and for @xmath89 .
one can see that the perfectly ordered ( @xmath90 ) stoichiometric alloy leads to a very high mae of @xmath91ev / atom .
this value agrees well with ref . , where the maximum mae for the same @xmath1 ordered system was obtained as @xmath92ev / atom .
the convex shape of both dependences in fig .
[ f_maevsy ] proves that the very high maes can be obtained only for systems with the maximum order ; even a small amount of additional disorder is detrimental to the maes and reduces them to much lower values of the completely random alloys .
@xcite strong sensitivity of the mae to the degree of the @xmath1 atomic order has recently been reported for the stoichometric fept alloy on an fcc lattice with small tetragonal distortions .
@xcite the results of ref . indicate that the chemical ordering is a more important factor for high mae values than the tetragonal distortion .
our results for the fe - co system witness that both factors are of equal importance , see fig .
[ f_maevsr ] .
we believe that validity of this type of conclusions depends also on the range of relevant variables : the tetragonal distortion was varied over a narrower interval ( @xmath93 ) in ref . , whereas a wider interval ( @xmath94 ) has been covered in our study .
another difference between the two systems lies in the dependence of the mae on the @xmath10-ratio : the mae is an ever increasing function of @xmath10 in fept , @xcite in contrast to the maxima found for the fe - co systems around @xmath50 , see fig .
[ f_2d_mae ] and fig .
[ f_maevsr ] .
these trends reflect probably the different origins of the high mae in these systems . in the fept alloys , the iron sites are responsible for strong exchange fields and the platinum sites provide strong spin - orbit interaction , both effects being only little dependent on the tetragonal distortion of the lattice . in the fe - co alloys ,
both species are featured by strong exchange splittings and weak spin - orbit couplings ; the high maes are heavily based on collective properties ( band structure ) of the tetragonal systems .
we have shown by means of first - principles lsda calculations that chemical disorder has a strong influence on the uniaxial mae of the bulk tetragonal fe - co systems .
first , the complete neglect of the disorder , inherent to the simple vca , overestimates the mae by a large factor , whereas the more sophisticated cpa leads to a drastic reduction of the mae .
the latter low maes are quite close to the magnetic shape anisotropy energy and the resulting estimated stability of the perpendicular magnetic anisotropy of thin strained fe - co films comes out probably smaller than the measured one .
second , the disorder - induced reduction of the mae is due to the strong scattering regime in the minority - spin channel , which smears the bloch - like eigenstates at the fermi energy . third , an analysis of the calculated orbital magnetic moments points to a non - negligible underestimation of the orbital magnetism and , most probably , of the maes by the lsda .
fourth , the @xmath1-like atomic long - range order leads to an enhancement of the mae values .
the most pronounced enhancement is obtained for the maximum degree of the order , while imperfect ordering reduces the mae very rapidly as compared to the perfectly ordered systems .
similar correlations between the atomic order and the mae have recently been found in the fept alloy @xcite and observed in artificially synthesized feni films .
@xcite we believe that this interplay should be taken into account in a future search of advanced materials for high - density magnetic recording .
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 , ed .
, @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) in @noop _ _ , , vol . , ( , ) p. @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( ) | we present results of systematic fully relativistic first - principles calculations of the uniaxial magnetic anisotropy energy ( mae ) of a disordered and partially ordered tetragonal fe - co alloy using the coherent potential approximation ( cpa ) .
this alloy has recently become a promising system for thin ferromagnetic films with a perpendicular magnetic anisotropy .
we find that existing theoretical approaches to homogeneous random bulk fe - co alloys , based on a simple virtual crystal approximation ( vca ) , overestimate the maximum mae values obtained in the cpa by a factor of four .
this pronounced difference is ascribed to the strong disorder in the minority spin channel of real alloys , which is neglected in the vca and which leads to a broadening of the @xmath0-like eigenstates at the fermi energy and to the reduction of the mae .
the ordered fe - co alloys with a maximum @xmath1-like atomic long - range order can exhibit high values of the mae , which , however , get dramatically reduced by small perturbations of the perfect order . |
g. a .- c . is supported in part by grant rfp2 - 08 - 02 from the foundational questions institute ( fqxi.org ) .
+ we gratefully acknowledge valuable conversations with f. fiore , g. gubitosi , f. mercati and l. smolin .
g. amelino - camelia , j. ellis , n.e .
mavromatos , d.v .
nanopoulos and s. sarkar , arxiv : astro - ph/9712103 , nature 393 ( 1998 ) 763 - 765 .
j. ellis , n.e .
mavromatos , d.v .
nanopoulos , a.s . sakharov and e.k.g .
sarkisyan , astropart .
phys . 25 , 6 ( 2006 ) 402 .
u. jacob and t. piran , arxiv:0712.2170 , jcap 0801 ( 2008 ) 031 .
j. albert et al .
[ magic collaboration ] and j.r .
ellis , n.e .
mavromatos , d.v .
nanopoulos , a.s . sakharov and e.k.g .
sarkisyan , phys .
b 668 ( 2008 ) 253 .
f. aharonian et al .
[ h.e.s.s .
collaboration ] , phys .
( 2008 ) 170402 . | observations of gamma - ray bursts are being used to test for a momentum dependence of the speed of photons , partly motivated by preliminary results reported in analyses of some quantum - spacetime scenarios .
the relationship between time of arrival , momentum of the photon and redshift of the source which is used for these purposes assumes a breakdown " of relativistic symmetries , meaning that it is a preferred - frame scenario which does not satisfy the relativity principle .
the alternative hypothesis of a deformation " of relativistic symmetries , which preserves the relativity principle by adopting deformed laws of relativistic transformation between observers , could not so far be tested in gamma - ray - burst observations because it was not known how to formulate it in expanding spacetimes .
we here provide such a formulation , and we find that also for the symmetry - deformation scenario the analysis of gamma - ray - burst data take us very close to the desired planck - scale sensitivity " . plus .5pt minus .5pt the famous second postulate of relativity , the one that in high - school textbooks is simply described as a law of constancy of the speed of light , affects the observations performed by a gamma - ray telescope in a somewhat subtle way .
a slight complication is due to the expansion of our universe : within classical general relativity the speed of light takes its ideal value @xmath0 only when the effects of spacetime curvature " , such as the universe expansion , can be neglected . for most earth - bound experiments all such effects are indeed negligible , but the gamma rays emitted by distant astrophysical objects travel for times long enough that the expansion of the universe is tangible . and essentially the expansion causes inbound light to be partly dragged away from us , resulting in an effective speed which is smaller than @xmath1 . for classical spacetimes
there is of course a robust and well - established description of the interplay between speed of light and a possible expansion of spacetime , and this description properly also factors in the first relativity postulate , the relativity principle " which requires the equivalence of the laws of physics in all frames of reference . but amusingly in some areas of quantum - spacetime research for the last decade we only managed to characterize some relativistic properties at the level of a high - school - textbook description , without the ability to account for the universe expansion .
this limitation in particular affects those models of non - expanding quantum spacetimes ( see , _ e.g. _ , refs .
@xcite ) which have provided motivation for the study of the hypothesis of momentum dependence of the speed of photons .
the non - constancy of the speed of light found in some flat / non - expanding quantum spacetimes definitely violates the second einstein postulate , and it also became quickly clear @xcite that the nature and strength of the implications of this unexpected momentum dependence are affected rather strongly by the fate of the first postulate , the postulation of the relativity principle " of equivalence of reference frames .
two alternative scenarios emerged .
on one side there is the lsb scenario ( lorentz symmetry breakdown ) , a quantum - gravity ether scenario " which renounces @xcite also to the validity of the first postulate , and therefore ends up formulating fully nonrelativistic laws with explicit reference to a preferred frame " .
the alternative possibility is a deformation " of relativistic symmetries , in the sense of the doubly - special relativity " @xcite ( dsr ) scenario , which insists on a relativistic formulation of laws , as required by the first postulate , and confines all revisions of relativity to revisions of the second postulate .
the lsb / quantum - gravity - ether scenario is somewhat simpler from a technical perspective , since one is spared the need of enforcing logical compatibility with the relativity of frames of reference .
but , even with this simplification , it was not straightforward to find a description of the momentum dependence of the speed of photons that would be compatible with the expansion of the universe .
after a few steps @xcite of gradual improvement , a consensus @xcite was reached on a formulation , which is centered on a description of the speed of photons in a spacetime with scale factor @xmath2 , which in particular in conformal coordinates takes the form @xmath3 where @xmath4 is the momentum the covariant spatial momentum @xmath5 " , which is a conserved quantity in frw spacetimes . the contravariant spatial momentum
@xmath6 " is of course not conserved , but its scaling along the worldline of a particle is governed by the simple law @xmath7 .
] of the photon and @xmath8 is the conformal time " , related to the cosmological time " @xmath9 by @xmath10 .
most applications of this scenario were centered on the associated prediction of a difference in ( cosmological-)times of arrival @xmath11 of two photons , with momentum difference @xmath12 , emitted simultaneously by a source at redshift @xmath13 : @xmath14 where , as customary , we assumed expansion described by the current cosmological model , with @xmath15 the hubble constant ( @xmath16 ) and the parameters @xmath17 and @xmath18 characteristic of the most successful @xmath19cdm cosmology " ( @xmath20 , @xmath21 ) . over these two years past since a consensus @xcite was reached on this reformulation for expanding spacetimes of the original lsb proposal first given in ref .
@xcite ( for non - expanding spacetimes ) , the prediction ( [ dtzjoccy ] ) was the unchallenged benchmark used in numerous data analyses @xcite testing the hypothesis of momentum dependence of the speed of photons .
one could therefore say that all these studies were exclusively testing the hypothesis of quantum - spacetime -relativity " .
the fact that there was no alternative dsr - relativistic picture represents clearly the most significant limitation of the scope of these studies .
we here set the stage for improving very significantly this state of affairs by reporting success in handling the severe technical challenges @xcite that must be met for a truly relativistic ( no preferred frame ) description of momentum dependence of the speed of photons in an expanding spacetime .
details will be provided in a longer companion paper @xcite . here
we summarize the most intuitive aspects of our findings . and for these purposes
it is useful to take as starting point the eqs .
( [ vjoccy ] ) , ( [ dtzjoccy ] ) of the much - studied lsb proposal .
we wrote ( [ vjoccy ] ) in conformal coordinates because this choice of coordinates of course facilitates the comparison with results obtained for non - expanding spacetimes .
the dsr frameworks which we intend to adapt here to spacetime expansion predict for non - expanding ( @xmath22 ) spacetimes @xmath23 of course , also in the dsr case the description of the momentum dependence of the speed of photons requires the introduction of a length scale , which we denote by @xmath24 . but compatibility with the first relativity postulate imposes that this length scale is an invariant , with the same value in all reference frames . the main non - relativistic feature of ( [ vjoccy ] ) is instead implicitly codified in the properties of the length scale @xmath25 , which is introduced in the lsb scenario as a scale that acquires different values @xcite in different frames of reference .
the observer - independence of @xmath24 was the main objective of our analysis , but we were also somewhat concerned about another aspect of ( [ vjoccy ] ) , which is the dependence of @xmath26 on @xmath27 : it implies that in such
quantum - gravity ether " pictures of quantum spacetime the adoption of conformal coordinates does not actually achieve the usual goal of factoring out the universe expansion for photons .
we should stress that there is no _ a priori _ reason to motivate this feature .
our present ( admittedly limited ) understanding of the quantum - spacetime concept does not suggest in any way that the speed of photons should , unlike what happens in the classical spacetime case ( @xmath28 ) , retain a ( conformal-)time dependence even in conformal coordinates . our search of a dsr framework that would be compatible with spacetime expansion was eventually successful , but only after exploiting the simplification of our task that was provided by the adoption of conformal coordinates .
our intuition was that in conformal coordinates we should be able to use rather directly the lessons learned by applications of the dsr concept in non - expanding spacetimes . and it is noteworthy that some of the most used formulations of such dsr theories are hamiltonian theories which add to the familiar hamiltonian describing a free particle in minkowski spacetime ( @xmath22 ) the correction term @xmath29 : @xmath30 the rather strong connection between the natural coordinates of minkowski spacetime and the conformal coordinates of an expanding ( frw ) spacetime led us to the proposal of adding the correction term @xmath29 also to the hamiltonian of a free particle in a classical frw spacetime : @xmath31 one easily finds @xcite that ( [ hamildesitter ] ) produces worldlines governed by the following description of the speed of photons in conformal coordinates for an expanding spacetime : @xmath32 which is ( conformal-)time independent . and most importantly we find @xcite that this theory governed by the hamiltonian @xmath33 of ( [ hamildesitter ] ) has the length scale @xmath34 with the status of a relativistic invariant .
we find that locally " ( _ i.e. _ in analyses local enough that the time dependence of the scale factor can be neglected ) the equations of motion are covariant , and the scale @xmath24 is invariant , under a rather typical @xcite dsr realization of relativistic symmetries , centered on the following deformed representation of boosts @xmath35 and for the de sitter case , @xmath36 , we rather remarkably also find that globally the equations of motion are covariant , and the scale @xmath24 is invariant , under a novel dsr realization of relativistic symmetries compatible with de sitter expansion , which is centered on the following deformed representation of boosts @xcite @xmath37 we are therefore providing for the first time a fully relativistic theory with momentum dependence of the speed of photons which is compatible with spacetime expansion , and this finally allows one to test the deformed - symmetry scenario with data gathered by gamma - ray telescopes .
the strategy is analogous to the one already in use for the lsb / ether - based predecessor : one relies @xcite on observations of sources that produce short - duration but intense bursts of gamma - ray photons , looking for indirect evidence of momentum - correlated differences in the times of arrival of the photons .
if such momentum - dependent delays are found and can not be ascribed to differences in times of emission , then they would be interpreted as a manifestation of a momentum - dependent speed of propagation .
while the strategy of the data analysis is analogous , there are two profound differences between the lsb framework centered on ( [ vjoccy ] ) and our dsr framework centered on ( [ vnoi ] ) .
the conceptually most significant difference between ( [ vjoccy ] ) and ( [ vnoi ] ) concerns the different status of the scales @xmath25 and @xmath24 , which we already stressed : the fact that our @xmath24 is absolutely invariant allows one to combine straightforwardly the results of all such data analyses , whereas for @xmath25 data obtained by different telescopes , or even by the same telescope at different times ( when , according to the ether picture , the telescope has different velocity with respect to the preferred frame of reference " ) , can be combined only in rather nontrivial way , only by carefully mapping all results to the same ( preferred " ) frame of reference .
this is the core difference between relativistic and nonrelativistic theories , but quantitatively , at least for the present generation of telescopes , it carries very little weight : taking , for example , the natural frame for the description of the cosmic microwave background as the preferred frame , one finds that the speeds of our telescopes in that frame are all very small with respect to @xmath1 , and this implies @xcite that also the manifestations of the non - invariance of @xmath25 are negligibly small at present . since speeds of order @xmath38 are not uncommon for space telescopes one can envision this issue to come into play ( if and ) when the length scale characteristic of the momentum dependence of the speed of light is measured with precision of 1 part in @xmath39 .
this is a possibility which is presently beyond our reach , but intriguingly it is not inconceivable for a not - so - distant future .
the second difference between ( [ vjoccy ] ) and our ( [ vnoi ] ) is less significant conceptually , since it is not an explicit manifestation of the relativistic nature of our proposal , but does affect the analysis in a way that is quantitatively significant .
this is the fact already stressed above that our analysis produces a law for the speed of photons which adopting conformal coordinates is time independent , whereas somewhat surprisingly the lsb / ether - based result ( [ vjoccy ] ) does depend on conformal time , through the expansion factor @xmath27 . as a result
our dsr proposal is applicable also to the phenomenology of the early universe , where ( as stressed above ) the popular lsb framework runs into concerning pathologies . and more importantly , even for the presently most actively pursued opportunity
, the one concerning observations of bursts of gamma rays , the difference between ( [ vjoccy ] ) and our ( [ vnoi ] ) is very tangible for data analysis , as shown in figs . 1 and 2 . ) and our relativistic eq .
( [ vnoi ] ) , by comparing the upper bounds on @xmath40 ( grey ) and @xmath24 ( blue ) that can be inferred from the four most significant relevant analyses of bursts of gamma rays , reported in refs .
@xcite .
different sources are identified by their redshift , with the primary objective of showing that data from sources at redshift up to @xmath41 have quantitatively similar implications for @xmath24 and for @xmath25 , but for @xmath13 significantly greater than @xmath42 ( where however presently the quality of data is not very good ) the differences are rather sizeable.,scaledwidth=48.0% ] in fig .
1 we compare the implications of the same observations on our relativistic proposal ( parameter @xmath24 ) and on the previously - considered ether - based scenario ( parameter @xmath25 ) .
the limits on @xmath24 that can be derived from different observations of bursts of gamma rays are described in units of the planck length @xmath43 ( @xmath44 ) , which according to a few independent ( but all semi - heuristic , see , _
e.g. _ , refs .
@xcite ) arguments should be within one or two orders of magnitude of the characteristic scale of quantum - gravity / quantum - spacetime effects ( so we should expect @xmath24 to be within one or two orders of magnitude of @xmath43 ) .
the main message that our fig . 1 intends to convey is that our analysis provides the basis for testing a relativistic ( no - ether ) description of momentum dependence of the speed of photons with sensitivity which is already presently at the planck - length sensitivity " level .
we find that the most stringent limit on @xmath24 is inferred from the observation of gamma - ray burst grb090510 by the fermi space telescope .
it was already shown in ref .
@xcite that for grb090510 , from a source at redshift @xmath45 , the data analysis could establish that any time delay due to momentum - dependent speeds of photons was not greater than @xmath46 , at momenta of @xmath47 .
this upper bound on momentum - dependence - induced time - of - arrival differences was shown @xcite to imply a limit of @xmath48 for the ether - based case .
we find that for our relativistic scenario the same data imply @xmath49 , and is the best present limit on @xmath24 .
the fact that the data on grb090510 imply a limit on @xmath25 and a limit on @xmath24 with not much difference between them shows that even at redshifts as large as @xmath50 the differences between our new dsr / relativistic result ( [ vnoi ] ) and its lsb / ether - based counterpart ( [ vjoccy ] ) , while tangible , are not very large .
fig .
1 also considers the case of the observation by the hess telescope @xcite of a flare from the active galactic nucleus pks2155 - 304 , whose implications for @xmath25 and for @xmath24 are quantitatively the same ( within @xmath51 ) , because pks2155 - 304 is at the relatively small redshift of @xmath52 . but the cases in fig
.
1 which are at redshift significantly greater than 1 ( grb090902b and grb080916c , observed by the fermi telescope ) illustrate the fact that at large redshifts the effects of our relativistic model are significantly different from the ones of the ether - based model . fig .
2 contributes from a different angle to the same assessment of the dependence on redshift of the differences between the much studied lsb scenario with @xmath25 and our novel dsr scenario . with fig .
2 we point out that , if data of quality roughly comparable with the observation of grb090510 is obtained from sources at somewhat higher redshift ( say , redshifts in the range from @xmath53 to @xmath54 ) the difference between the lsb scenario and our dsr scenario will be observably large . on redshift for our dsr ( blue line ) and the lsb ( grey line ) scenario , in the case when
the parameters @xmath24 and @xmath25 are adjusted in such a way that the two predictions for @xmath55 are the same at @xmath45 ( the redshift of grb090510 ) . at large redshift
the differences are significant enough that one could plausibly discriminate experimentally between the two scenarios .
this however would require much better control of systematic errors than we have presently ( see later comments to fig . 3 ) and
would also require that the statistical errors for data analyses on high - redshift sources ( hypothetically drawn as dashed blue vertical intervals ) are kept comparable to the statistical error found in the analysis of grb090510 ( solid blue vertical interval).,scaledwidth=49.0% ] fig .
3 offers another perspective on presently - available data centered on the fact that , upon re - expressing our conformal - coordinate formulation in terms of cosmological time and redshift measurements , one finds that our result ( [ vnoi ] ) leads to the prediction of a linear relation @xmath56 between the difference in times of arrival @xmath11 of two simultaneously - emitted photons , with momentum difference @xmath12 , and the following function of the redshift @xmath13 : @xmath57 ( where we assumed again @xmath58 , @xmath20 , and @xmath21 ) .
one can exploit the linear dependence on @xmath59 of the @xmath55 ( expected delay of arrival versus momentum difference of simultaneously emitted photons ) .
we show two such linear relations , one ( solid line ) for the value of @xmath60 , which we establish as the best present limit on @xmath24 ( from data @xcite on grb090510 , at redshift @xmath45 , _
i.e. _
@xmath61 ) , and one for @xmath62 , which is the second best limit ( from data @xcite on pks2155 - 304 ) . following the thesis put forward in the recent ref .
@xcite , the figure suggests that in some cases the mere analysis of statistical errors ( blue vertical intervals ) could naively invite a positive / discovery interpretation , but ( as suggested by the brown vertical lines ) essentially we still have no control on a peculiar systematic error that affects these analyses , which is due to our poor knowledge of the mechanisms that produce the structure of the bursts at the source ( the preliminary indications of correlation between momentum of the photon and time of its detection at the telescope are likely to be entirely due to properties of the source).,scaledwidth=48.0% ] another key difference between our new dsr / relativistic framework and the previous lsb / ether - based picture concerns the stability of photons .
our formalization of deformed relativistic symmetries is powerful enough that one can then easily prove that the photon is stable ( for example the requirements of dsr - relativistic kinematics in expanding spacetime are such that the process @xmath63 is never allowed @xcite ) .
the description of dsr - relativistic kinematics in expanding spacetime was one of the key missing pieces of previous attempts of dsr formulation @xcite . and
it marks a very significant difference with respect to lsb / quantum - gravity - ether theories where photon decay can be a significant phenomenological concern @xcite .
the two phenomenological pictures we have been comparing , the previously developed lsb scenario and our novel dsr scenario , clearly present significant differences . and we feel that both of them deserve intrinsic interest , with or without quantum gravity , just because the importance of the development of novel lorentz - symmetry test theories must be assessed as proportional to the pivotal role that exact / classical lorentz symmetry enjoys in our present formulation of the fundamental laws of nature .
still it is of course interesting to ponder on the implications that these two scenarios could have specifically for quantum - gravity research . because of the complexity of the relevant quantum - spacetime theories , at present we only have relatively robust theoretical evidence " ( in some models @xcite ) of modifications of the energy - momentum dispersion relation for non - expanding spacetimes , without any robust indication for or against the emergence of a preferred frame , and without the ability to predict the implications of curvature / expansion of spacetime for the relevant effects .
our analysis provides guidance for future works on the theory side attempting to establish more robustly the links from models of quantum spacetime and phenomenological pictures concerning the fate of lorentz symmetry at the planck scale . on the issue of the possible emergence ( or dsr / non - emergence ) of a preferred frame
there was already a reference conceptual framework , discussed _
e.g. _ in refs .
@xcite .
we here provided several new elements that could play an important role in future analyses of these issues .
in particular we observed that attempts to give sharper quantum - gravity motivation for the previously developed lsb scenario should look for evidence of a mechanism such that the adoption of conformal coordinates would not actually achieve the usual goal of factoring out the universe expansion for photons .
the possibility of such a failure " of conformal coordinates had never been raised before in the quantum - gravity literature and we feel it can be of valuable guidance . more insight can be gained by looking at the specific conformal - time dependence predicted by ( [ vjoccy ] ) , for the previously developed lsb scenario .
if one takes at face value ( [ vjoccy ] ) then the presence of the scale factor in the denominator of the correction term produces , as shown in fig . 4 , some rather puzzling results : for some correspondingly small values of the scale factor the world - lines described by ( [ vjoccy ] ) are rather pathological and the energy of the photons takes imaginary values @xcite .
in order to improve on this state of affairs it appears natural to invoke the presence of further correction terms ( to be added to ( [ vjoccy ] ) ) of type , say , @xmath64 and with sign and magnitude adjusted in such a way to eliminate the unwanted features of the worldlines . in turn this observation can provide a key target for attempts to reproduce the previously developed lsb scenario within some chosen quantum - gravity picture : it appears that these attempts should first of all find evidence of a mechanism such that the strength of quantum - gravity effects is governed by @xmath65 , which for a massless particle is the comoving energy " ( energy in comoving coordinates " , which are like conformal coordinates but with cosmological time rather than conformal time ) . and
it is noteworthy that such a mechanism would be easily distinguished from the type of mechanism that a quantum - gravity theory should host in order to provide support for our novel dsr picture , where the magnitude of quantum - gravity effects is governed by the conserved charge @xmath4 . in our dsr picture
the significance of quantum - gravity effects would be weighed by comparing the @xmath4 that labels the worldline and the ( observer - independent ) scale @xmath66 : some worldlines could be described neglecting the quantum - gravity effects completely , while other worldlines ( with @xmath4 comparable to @xmath66 ) would be severely affected by the quantum - gravity effects . in the previously developed lsb scenario , since the significance of quantum - gravity effects would be determined by comparing @xmath65 to the ( observer - dependent ) scale @xmath67 , one would instead typically find that different portions of the same worldline are affected differently by the quantum - gravity effects : portions where the scale factor is not small would be largely unaffected , while portions where the scale factor is very small would reflect quantum - gravity properties very strongly . ) , ( [ dtzjoccy ] ) . for values of the conformal time
such that @xmath68 one finds that the path of photons of momentum @xmath69 ( much below the planck scale ) from a static source to a static telescope ( both static in comoving coordinates of course ) starts off going in the
wrong direction " , and gets going in the
right direction " only after some time , when the scale factor reaches larger values .
this is shown by the red line , which is also compared to the case ( black line ) of a photon emitted simultaneously by the same source with momentum @xmath70 . during the time of
the world - line anomaly " one also finds @xcite that the energy is imaginary.,scaledwidth=44.0% ] the possibility that the differences we here exposed could be exploited to discriminate between the much - studied lsb framework of eqs .
( [ vjoccy ] ) , ( [ dtzjoccy ] ) and our novel dsr framework is certainly exciting . of course , the most likely scenario is the one with einstein s relativity still prevailing , at least for now .
we assumed here effects suppressed only by one power of the planck length , because this is what is afforded by the sensitivity of presently - operating gamma - ray observatories .
but from a quantum - gravity / quantum - spacetime perspective the case of quadratic ( @xmath71 ) suppression appears to be equally natural @xcite , and could become accessible in a not - too - distant future , exploiting @xcite observations of the ultra - high - energy neutrinos that accompany @xcite gamma - ray bursts .
the techniques we here discussed for discriminating between dsr / relativistic frameworks and lsb / ether - based frameworks should prove valuable for that future challenge , and are already a necessary resource for fully exploiting the present opportunities for challenging einstein relativity with corrections that depend linearly on the planck length . |
tt hydrae ( hd 97528 , hip 54807 , sao 179648 , @xmath1 ) is a typical algol - type eclipsing binary system with an orbital period of @xmath2 days ( kulkarni & abhyankar 1980 ) and consists of a hotter b9.5 v main sequence primary and a cooler evolved k1 iii - iv secondary star filling its roche lobe .
there is evidence of the circumstellar material in the form of an accretion disc ( plavec & polidan 1976 ) surrounding the primary as well as an indication of a gas stream ( peters & polidan 1998 ) .
kulkarni & abhyankar ( 1980 ) obtained ubv photometric observations and photometric elements of the system .
these observations were re - examined by etzel ( 1988 ) who found from photometry , spectrophotometry and spectroscopy : @xmath3k , @xmath4 kms@xmath5 for the primary , and @xmath6k for the secondary .
he also suggested , based on the inglis - teller formula , that the gas producing the uv excess and balmer line emission has an electron number density lower than @xmath7 @xmath8 .
he also estimated a distance of 193 pc to the binary .
eaton & henry ( 1992 ) estimated @xmath9 kms@xmath5 for the secondary .
plavec ( 1988 ) studied the iue spectra of tt hya and concluded from the presence of fe absorption lines that the vertical dimension of the accretion disc is significant and is at least comparable to the diameter of the primary .
he estimated a distance to the system of 194 pc , and pointed out that the presence of super - ionized emission lines of si , c , nv poses a question about the source of ionization .
peters ( 1989 ) estimated from the eclipses of the @xmath0 emission region that a fairly symmetrical disc fills up to 95% of the roche lobe radius .
moreover , she argued that the disc must be rather flat with the vertical dimension comparable or less than the diameter of the primary .
the mean electron number density in the disc was estimated to be @xmath10 @xmath8 .
the depth of the @xmath0 absorption was found to be strongly variable with phase and was deepest shortly before and after primary eclipse ; a feature also observed in other algols ( e.g. , richards 1993 ) .
peters ( 1989 ) also suggested that the additional absorption in the center of the @xmath0 line probably comes from the dense inner region of the accretion disc .
she also observed that the @xmath0 core is blueshifted relative to the primary by about @xmath11 kms@xmath5 and interpreted it as evidence of the mass outflow .
sahade & cesco ( 1946 ) determined the orbital parameters of the primary and identified an eccentric orbit from the cak and h lines .
popper ( 1989 ) measured radial velocities of the primary and secondary and found that the motion of the secondary was consistent with the circular orbit ( @xmath12 kms@xmath5 ) but not in the case of the primary .
van hamme & wilson ( 1993 ) reanalyzed the light and velocity curves separately and simultaneously and used a physical model to obtain the following parameters : separation @xmath13 , effective temperature of the primary @xmath14k , secondary @xmath15k , radius of the primary @xmath16 , secondary @xmath17 , mass of the primary @xmath18 , secondary @xmath19 , and mass ratio @xmath20 .
albright & richards ( 1996 ) obtained the first doppler tomogram to reveal an indirect image of the accretion disc .
hipparcos ( esa 1997 ) obtained a parallax of @xmath21 mas .
peters & polidan ( 1998 ) observed red - shifted absorption in ni , n lines in their fuv spectra at the phase 0.95 and interpreted it as evidence of the gas stream and inferred the rate of inflow to be greater than @xmath22yr@xmath5 .
most recently , richards & albright ( 1999 ) studied the @xmath0 difference profiles and properties of the accretion region and put tt hya into the context of other algol binaries
. it would be interesting to verify the above estimates , conclusions , and interpretations on more physical grounds using some more sophisticated methods involving synthetic spectra which would take into account the circumstellar matter . unfortunately , there are not many tools available for such a task at present .
there are sophisticated computer codes for calculating and inverting light curves or spectra of binary stars with various shapes or geometry including the roche model ( lucy 1968 ; wilson & devinney 1971 ; mochnacki & doughty 1972 ; rucinski 1973 ; hill 1979 ; zhang et al .
1986 ; djurasevic 1992 ; drechsel et al . 1994 ; vink et al . 1996
; hadrava 1997 ; bradstreet & steelman 2002 ; pribulla 2004 ) . in these codes , the stars are assumed to be nontransparent and stripped of any circumstellar matter .
an interesting approach to study extended semitransparent atmospheres of some eclipsing binary components was developed by cherepashchuk et al .
often the 3d model of the circumstellar matter ( behavior of state quantities and velocity field ) is known or expected as a result of hydrodynamic simulations or observational constraints ( see for example karetnikov et al .
1995 ; richards & ratliff 1998 ) .
unfortunately , 3d non - lte ( nlte ) calculations and spectral synthesis including complex hydrodynamics are difficult to carry out
so one alternative has been to perform a simple volume integration of emissivity , which is often too oversimplified for the particular problem . on the other hand , highly sophisticated model atmospheres and spectrum synthesis codes have been developed assuming nlte and plane - parallel atmospheres of hot stars ( hubeny 1988 ; hubeny & lanz 1992 , 1995 ; hubeny et al .
1994 ) , spherically symmetric atmospheres ( kubt 2001 ) , or stellar winds ( krtika & kubt 2002 ) .
there are also sophisticated stationary plane - parallel line - blanketed model atmospheres and spectrum synthesis codes for a large variety of stars assuming lte ( kurucz 1993a , 1993b ; smith & dworetsky 1988 ; piskunov 1992 ; and many others ) . however , these are very specialized codes and their main purpose is to calculate the spectrum emerging from a stellar atmosphere , and it is difficult to apply them to the various cases outlined above .
an exception is the special case of circumstellar matter in the form of accretion discs in cataclysmic variables ( cvs ) . in this case , the disc is either approximated by a set of geometrically thin , but optically thick , static local atmospheres and the output radiation is a sum of properly doppler shifted local emerging intensities ( orosz & wade 2003 ; wade & hubeny 1998 ; la dous 1989 ) or , in case of optically thin discs or accretion disc winds , the sobolev approximation is used ( proga et al . 2002 ; long & knigge 2002 ; rybicki & hummer 1983 ) .
linnell & hubeny ( 1996 ) developed a package of codes which can calculate light curves or spectra of interacting binary stars including optically thick ( nontransparent ) disc .
however , it is often necessary to solve the radiative transfer in a moving disc at least along the line of sight as demonstrated by horne & marsh ( 1986 ) .
nevertheless , their calculations assumed a linear shear and neglected stimulated emission , continuum opacity , scattering , as well as a central star .
in this paper , we attempt to bridge the gap in the previously mentioned approaches using our new code .
it is a tool which solves in lte the simple radiative transfer along the line of sight in an optional optically thin 3d moving medium .
optional transparent ( or non transparent ) objects such as a spot , disc , stream , jet , shell or stars as well as an empty space may be defined ( or embedded ) in 3d and their composite synthetic spectrum calculated .
the stars may have the roche geometry and known intrinsic spectra .
the code is quite a multi - purpose , independent , and flexible tool which can also calculate a light curve or a trailing spectrogram where necessary , or can be used to study various objects or effects . in section [ s1 ] , we describe the basic astrophysics included in the code . in section [ snp ] ,
we briefly mention numerical methods used in the code . in section
[ s2 ] , we study in more detail the effects of various input parameters on the spectrum of a star which has a circumstellar disc .
budaj & richards ( 2004 ) made preliminary calculations for a single star with a disc with parameters similar to tt hya .
however , these were based on the stellar parameters found by etzel ( 1988 ) , a wedge shape geometry of the disc with a relatively large inner radius ( 2.5@xmath23 ) , and no limb darkening of the primary star .
more realistic stellar parameters by van hamme & wilson ( 1993 ) and a slab shaped disc geometry will be used in this work .
then in section [ s3 ] , we will make more detailed calculations of the system for different phases involving both stars and a disc and will compare our new synthetic spectra to real observations of tt hya .
this section describes the basic astrophysics included in the code . for more detailed information and a manual for the code ,
the reader is referred to budaj & richards ( 2004 ) . in the following analysis ,
the calculations are carried out in the observer s cartesian frame with @xmath24 pointing towards the observer .
the radiative transfer equation along the line of sight is : @xmath25 where @xmath26 is the specific monochromatic intensity at the frequency @xmath27 , @xmath28 is the opacity , @xmath29 is the emissivity and @xmath24 is the distance along the beam .
it is convenient to split the opacity into two contributions , the true absorption @xmath30 and the scattering @xmath31 : @xmath32 assuming lte , the line opacity corrected for a stimulated emission is stated simply as : @xmath33 where @xmath34 is the planck constant , @xmath35 is the energy of the transition from the lower level @xmath36 to the upper level @xmath37 , @xmath38 is the boltzmann constant , @xmath39 is the temperature , @xmath40 is the population of the @xmath36-th state of the corresponding ion , and @xmath41 is the einstein coefficient for the whole solid angle .
the velocity field enters the equation via the shifted normalized voigt profile @xmath42 where @xmath43 where @xmath44 is the laboratory frequency of the line and @xmath45 is the radial velocity ( positive towards the observer ) or projection of the local velocity vector @xmath46 to the line of sight unit vector @xmath47 .
the einstein coefficient , @xmath41 , is related to the oscillator strength , @xmath48 , by : @xmath49 where @xmath50 are the electron charge and mass , respectively and @xmath51 is the speed of light .
the shape of the voigt profile is determined by the thermal and the microturbulent broadening , @xmath52 , characterized by the doppler half - width @xmath53 as well as by the damping broadening characterized by the frame damping parameter @xmath54 where the damping constant @xmath55 includes the contribution from the natural , stark and van der waals broadening . in the case of lte ,
all the line opacity is due to the true absorption process i.e. : @xmath56 we also included four other continuum opacity sources : the hi bound - free opacity , the hi free - free opacity , thomson scattering and rayleigh scattering on neutral hydrogen , denoted by @xmath57 , respectively .
+ the total true absorption @xmath30 is the sum of the three opacity sources : @xmath58 the total scattering @xmath31 is : @xmath59 the thermal emissivity associated with the true absorption can then be written as @xmath60 where @xmath61 is the planck function for the local value of the temperature . for scattering emissivity
we have @xmath62 where @xmath63 is the scattering coefficient containing the general redistribution function .
it is this term which causes the main difficulty , since apart from redistributing the frequencies ( @xmath64 ) , it also couples the radiation in one direction @xmath47 with the radiation field in all other directions @xmath65 .
however , in many applications ( e.g. , optically thin shells ) this term can either be neglected or governed by the scattering of light from the central object .
we assume coherent isotropic scattering ( as seen from the scattering particle frame ) from a blackbody or from a central spherical star with precalculated surface intensity @xmath66 or flux @xmath67 .
in this case the emissivity reduces to : @xmath68 where @xmath69 is the mean intensity .
ignoring limb darkening , @xmath69 can be approximated by : @xmath70 where @xmath71 is the solid angle subtended by the central star and @xmath72 where @xmath73 is the radius of the central star and @xmath74 is the distance from the center ( of the star / grid ) and @xmath75 and @xmath76 where @xmath46 is the velocity field vector at the given point specified by the vector @xmath77 and @xmath78 is the velocity of the center of mass of the central object . + for @xmath79 , an approximation including the limb darkening and the non - isotropic dipole phase function @xmath80 can be used : @xmath81 where @xmath82 where @xmath83 where @xmath37 is the limb darkening coefficient .
+ the total emissivity is then @xmath84 and the total source function is : @xmath85 the radiative transfer equation along the line of sight is solved ( see section [ snp ] ) and the flux , @xmath86 , from the object at the earth is then obtained by the integration of the output intensities through the 2d projection surface of the 3d object .
both objects , star and companion , may have shapes according to the roche model for detached or contact systems .
descriptions of the roche model can be found in kopal ( 1959 ) , limber ( 1963 ) , plavec & kratochvil ( 1964 ) , mochnacki & doughty ( 1972 ) , hilditch ( 2001 ) and many other papers and books .
most of them use cylindrical coordinates but we will assume a cartesian coordinate system @xmath87 centered on one of the stars ( labeled as 1 ) such that the companion ( labeled as 2 ) is at ( 1,0,0 ) and revolves around the @xmath24 axis in the direction of the positive @xmath88 axis . let the mass ratio , @xmath89 , always be @xmath90 or ` companion / star ' and @xmath91 will indicate the central star is heavier while @xmath92 means the central star is lighter .
we will also assume synchronous rotation i.e. , the stellar surface rotates with the orbital period .
then , the normalized roche potential , @xmath93 , is expressed as : @xmath94 where @xmath95 and @xmath96 the roche surface of a detached component is defined as an equipotential surface @xmath97 passing through the sub - stellar point @xmath98 ( point on the surface of the star in between the stars , @xmath99 ) which is localized by the ` fill - in ' parameter @xmath100 .
we define this by : @xmath101 for the primary and the secondary , respectively .
@xmath102 is the @xmath103 coordinate of the l1 point at ( @xmath104 ) .
the roche equipotential surface @xmath105 of a contact system will be defined by the fill - out parameter @xmath106 : @xmath107 where potentials @xmath108 correspond to the potentials at the l1 and l2 points , respectively .
first , we calculate l1 , l2 , @xmath105 and @xmath103-boundaries of the object using the newton - raphson iteration method in the @xmath103 direction , and then the 3d shape of the surface is solved using the newton - raphson iteration in @xmath88 and @xmath24 coordinates with the precision of about @xmath109 .
gravity darkening is taken into account by varying the surface temperature according to the following law ( claret 1998 ) : @xmath110 where @xmath111 is the normalized surface gravity , @xmath112 is the gravity darkening exponent , @xmath113 , and @xmath114 are the temperature and gravity at the rotation pole .
the normalized gravity vector is @xmath115 and @xmath116 where : @xmath117 the gravity darkening factor of the surface intensity is then calculated as : @xmath118 notice , that there is an imminent singularity in the calculations in the vicinity of l1 , l2 points since gravity falls to zero which drags temperatures ( a denominator in many equations ) to zero .
we avoid the problem by setting the lowest possible value of @xmath119 . limb darkening is taken into account using eq .
[ eld ] and by calculating the cosine of the angle @xmath120 between the line of sight unit vector @xmath121 and a normal to the surface : @xmath122 the reflection effect is not included in the present version .
a number of optional objects ( including transparent , nontransparent , empty space ) can be defined within the model and the line of sight may cross more of them within a few grid points . a simple and stable method is needed to cope with such velocity , density , temperature fields which are optional and are allowed to be non - continuous .
our problem is that of the integration of a first order ordinary differential equation with known initial values at one boundary .
the equation of the radiative transfer along the line of sight at the frequency @xmath27 is integrated with an analogue of the mid - point method ( budaj & richards 2004 ) .
it can be written in the discretised form : @xmath123 and simply integrated via the following recurrent formula : @xmath124 where @xmath125 and @xmath126 where @xmath127 if the line of sight happens to hit a nontransparent object , the integration starts on the other side of the object with the boundary condition : @xmath128 or @xmath129 where @xmath130 is the doppler shifted frequency in the comoving frame corresponding to the frequency @xmath27 of the observer s frame : @xmath131 and @xmath132 is the radial velocity of the surface of the nontransparent object where it intersects the line of sight .
@xmath133 is the surface intensity of the nontransparent object perpendicular to the surface in the comoving ( frozen to the surface ) frame .
if @xmath133 is not available it can be calculated e.g. , using eq .
[ e3 ] from the known surface flux .
rotation of the nontransparent objects is fully taken into account here by including it into the calculations of @xmath132 .
@xmath134 is a limb darkening factor : @xmath135 where @xmath37 is the limb darkening coefficient and @xmath120 is the angle between the normal to the surface of the nontransparent object and the line of sight .
@xmath136 is a gravity darkening factor which is important in the case of roche geometry ( see the section [ srg ] ) and in which case @xmath137 is the temperature at the rotation pole of a detached star or the temperature at the rotation pole of the more massive star in the case of a contact system , otherwise , it is the common effective temperature of the spherical star . if the line of sight happens to pass through an empty space this region is skipped and the integration continues with @xmath138 .
the code enables the user to look on the object from different points of view and to calculate the corresponding spectra .
the input model of the shell is defined in its ` body frozen ' cartesian coordinates ( @xmath139 ) with the @xmath140 axis corresponding to the intrinsic rotation axis of the model . the spectrum is always calculated in the observer s ` line of sight ' cartesian frame ( @xmath141 ) with @xmath24 pointing to the observer and which has the same center of coordinates . the code assumes the known behavior of state quantities : temperature @xmath39 , density @xmath142 , and electron number density @xmath143 .
solar abundances are assumed ( grevesse & sauval 1998 ) but the user is allowed to change all the element abundances .
the level populations are obtained from the boltzmann and saha equations .
partition functions ( routine ` pfdwor ' ) were taken from the code ( smith & dworetsky 1988 ; smith 1992 ) .
a code containing the partition function routines is also available in budaj , dworetsky & smalley ( 2002 ) .
the gaunt factors ( routines ` gaunt ' , ` gfree ' ) and voigt profile ( routine ` voigt0 ' ) are calculated with the subroutines taken from the code ( hubeny et al .
if the damping constants are not known they are estimated in the code in the way analogous to the code .
we also adopted the atomic data for chemical elements ( routine ` state0 ' ) from and routines ` locate ' , ` hunt ' from numerical recipes ( press et al .
we also used a few sections from our previous original codes for calculations of radiative accelerations in stellar atmospheres of hot stars ( budaj & dworetsky 2002 ) .
apart from the above , the code was written from scratch and provides an independent tool to study a large variety of objects and effects .
cgs units are used within the code and the manuscript , if not specified otherwise .
before we modeled the complex spectrum of the two stars and the disc , we studied the effects of various free parameters on the spectrum . to comprehend the problem more easily we restricted our calculations to the following assumptions .
we neglected the secondary and studied only the primary star surrounded by a disc .
the primary was treated as a spherical blackbody with the following parameters : effective temperature @xmath144 , mass @xmath145 , radius @xmath146 ( van hamme & wilson 1993 ) and limb darkening @xmath147 .
the disc was characterized by a slab of vertical half - width @xmath148 which was further constrained by two spherical surfaces with the inner radius @xmath149 and the outer radius @xmath150 .
the electron number density of the disc was set to @xmath151 @xmath8 , density @xmath152 g@xmath8 , temperature @xmath153k , microturbulence @xmath154 kms@xmath5 , and inclination @xmath155 .
we adopted the following velocity field @xmath46 of the disc : @xmath156 where @xmath157 is the mass of the central object , @xmath158 is the gravitational constant , @xmath159 are the angular velocity and radius vectors , respectively .
the above prescription for the velocity reduces to the well known keplerian velocity field in the limit of zero vertical thickness of the disc .
the calculations were performed on the cartesian 3d grid with the typical resolution in the disc plane of @xmath160 points .
figures [ f1 ] to [ f3 ] illustrate the effects of varying just one of these free parameters of the system .
the top of figure [ f1 ] displays the effect of increasing the vertical half thickness of the disc from @xmath161 to @xmath162 .
the overall emission increases as we increase the emitting space volume , and the central absorption decreases mainly for the thicker discs .
this occurs because the central depression is partly produced by the material projecting onto the star and there is more material in the keplerian disc which does not project onto the stellar disc ( surface ) for thicker discs , thereby filling the central absorption .
this is not true for very thin discs as it essentially causes the disc and its line profile to diminish and both emission and central absorption approach the continuum level .
contrary to the case of the wedge - shaped disc ( budaj & richards 2004 ) , we observe that in a slab disc geometry the blue / red line profile wings of the thick discs ( @xmath163 ) do not change much and emission peaks get slightly closer .
the middle section of figure [ f1 ] displays the effect of changing the outer radius of the disc from 6 to 14 @xmath23
. the emission increases as the emission volume increases .
this parameter has a strong effect on the position of the emission peaks and width of the central depression .
the peaks get closer and the central depression narrower for larger radius as more matter is involved at lower keplerian velocities .
the depth of the central absorption varies very little since the amount of the matter projected onto the stellar surface does not vary as much as in the previous case .
this absorption first deepens as the overall emission becomes more pronounced and then for @xmath164 weakens as the fraction of the gas projecting on the stellar disc decreases .
the blue and red wings of the line are not affected at all .
the bottom part of figure [ f1 ] displays the effect of increasing the inner radius of the disc from 2 to 8 @xmath23 .
contrary to the outer radius , the inner radius has little effect on the emission from the beginning when it is a small fraction of the outer radius since it does not change the emission volume noticeably .
the changes are mainly seen in the far wings of the profile since the higher keplerian velocities are involved . only when it approaches the outer radius , the overall emission starts to decline and the line weakens as the emission volume decreases , and the shape of the profile acquires the new u - type shape with a central hole .
this shape indicates that the central depression is created by two different mechanisms ( see below ) . in figure [ f2 ]
( top left frame ) , we inspect what happens if we vary the stellar radius from 1.5 to 3.0 @xmath23 .
stellar radius certainly changes the amount of matter projected onto the stellar surface but it also raises the continuum level and dilutes the emission from the disc .
this is the main reason why the emission weakens for bigger stars .
at the same time , the central absorption gets wider because the matter projected onto the stellar disc spans a broader radial velocity interval .
when varying the stellar radius , it may happen that the star overlaps the inner radius of the disc .
in such a case , the program sets both radii to be equal .
we also illustrate here the line profiles if the disc were viewed from different inclination angles ( figure [ f2 ] , top right frame ) . starting from edge - on , @xmath165
, the two emission peaks have the lowest intensity and we also observe the largest separation of the emission peaks and the deepest central depression .
this is because we encounter the highest radial velocities and most of the disc which is cooler than the star projects onto the stellar surface . as the inclination decreases , and we begin to see the disc pole - on , the emission peaks increase and their separation and central depression vanish , eventually merging into single peak emission for @xmath166 .
one can also see that the central depression is caused by two different effects .
the velocity field and geometry of the disc itself can produce the u or v type of the depression .
superposed on that is the absorption by the cool matter projected onto the hotter stellar surface which quickly diminishes if viewed out of the disc plane .
it is interesting to explore how the shape of the profile changes if the geometry of the disc changes .
the prescription for the velocity field remains the same . in figure [ f2 ] ( bottom left frame ) , we compare two profiles : one obtained for the slab shaped disc with the standard parameters listed above .
the other disc has the shape of a rotating wedge with the wedge angle @xmath167 , which produces emissions of similar strength to the slab disc model .
all other parameters were the same .
it is clear that the main difference is in the wings , which are more pronounced in the slab model since there is more matter closer to the star in that model than in the wedge model .
apart from the state quantities and velocity field , every space point can be assigned a value of microturbulence .
it is included in the calculation of thermal broadening as an additional thermal motion and can thus be used to model chaotic velocity fields on the scale smaller than the mean free path of a photon .
it turns to be a useful free parameter since the mass transfer may not be a smooth process and the velocity field of the spiraling gas in the disc may well depart from the circular keplerian orbits and be turbulent , and these departures may easily exceed the sound velocity which is of the order of 10 kms@xmath5 .
figure [ f2 ] ( bottom right frame ) illustrates the effect of increasing the microturbulence in the disc from 0 to 60 kms@xmath5 .
note that the inner radius keplerian disc velocity is about 500 kms@xmath5 .
the emission peaks are gradually smoothed and are broader . the behavior of the central absorption is more complicated and interesting . from the beginning it gets narrower and even deeper because of the desaturation effects of the microturbulence but then the smoothing effect prevails .
the microturbulence affects the optical depth along the line of sight .
while the surface of the star is seen at an optical depth of approximately 1.5 ( at the frequency of the line center ) without the microturbulence , it is seen at the optical depth of 0.7 , 0.4 for the microturbulence of 20 and 40 kms@xmath5 , respectively . without the turbulence and assuming the input parameters mentioned above
, the line opacity was found about 3 orders of magnitude higher than the continuum opacity in the line center .
thomson scattering was the most important continuum opacity source and was a factor of 300 higher than the hydrogen bound - free contribution , while h bound - free opacity was a factor of 20 higher than the hydrogen free - free opacity , and h free - free opacity a factor of 30 higher than rayleigh scattering opacity in the disc .
this pattern , however , will be very sensitive to the state quantities , wavelength , and line absorption coefficient .
figure [ f3 ] ( top left frame ) displays the effect of increasing the electron number density of the disc from @xmath168 to @xmath169@xmath8 . here , we assume that hydrogen is almost fully ionized , in which case the density is linearly proportional to the electron number density .
this strongly enhances the emission peaks and slightly deepens the central absorption . this huge impact on the emission
can be understood since the equivalent width in the optically thin case is proportional to the population of the particular level , which in turn is proportional to the total hi population .
however , the ionization fraction of neutral hydrogen is proportional to the electron number density , and if the latter is also proportional to the density it follows that the total hi population increases with the square of the density or the electron number density assuming the hydrogen abundance and temperature are fixed ( @xmath170 ) . as a consequence , the equivalent width , eqw , should behave also like @xmath171 which is seen in figure [ f3 ] .
since the emission is so sensitive to the density we also studied the effect of different density profiles ( figure [ f3 ] , top right frame ) . let the density and electron number density be inhomogeneous and have a power law dependence on the distance from the star , namely , @xmath172 .
this was achieved by varying the exponent @xmath173 from 0.0 to -2.0 and by normalizing the synthetic spectrum to the same emission peak strength .
it is apparent from the figure that the line profiles for lower exponents are broader and have steeper wings while those for higher exponents have narrower and smaller central depression .
it can easily be explained since the lower exponents emphasize the matter close to the star with higher keplerian velocity while the opposite is true for the higher exponents .
contrary , to what was found by budaj & richards ( 2004 ) for a wedge - shaped disc with the inner radius of 2.5@xmath23 the position of the emission peaks in a slab disc model with @xmath174 is much more sensitive to this exponent .
figure [ f3 ] ( bottom left and bottom right frames ) displays also the effect of varying the temperature of the disc over the interval 5000 - 20000k .
the emission is strongest at about 7000k and declines towards higher temperatures as the fraction of neutral hydrogen declines due to ionization .
the emission also declines towards cooler temperatures as it decreases the population of the lower level from which the @xmath0 originates . at the same time
, the temperature of the disc has a strong effect on the depth of the central absorption which grows towards the cooler temperatures . for temperatures @xmath175 , the line source function along the line of sight hitting the stellar surface steps down from the stellar surface to disc and towards the observer . for higher temperatures @xmath176 , this component of the source function steps up reducing the central depression . for both cooler and hotter temperatures than 7000k ,
the line - to - continuum emissivity decreases ( the continuum is provided by the star mainly and is essentially constant ) and the line disappears from the spectrum forcing the central depression to merge with continuum as well .
this is the reason why the central depression ceases also below 6000k although the line source function steps down .
the cooler discs if they were more dense could produce even deeper central absorption . nevertheless , one can see that discs or regions with temperatures below 5000k or above 30000k are not capable of producing strong balmer line emissions and that most of such emissions observed in the algols or cvs must originate from the above mentioned temperature regions .
we would like to emphasize that these calculations do not aim to calculate the physical model of the disc .
the model is assumed to be known .
the previous calculations illustrate the effect of varying just one free parameter at a time while keeping all the others fixed . in many cases
this may not represent the real behavior of the system since some quantities may be more or less interlocked such as temperatures and electron number densities . now that we understand the influence of various parameters on the emerging spectrum , we can proceed to fit the complex spectrum of tt hya .
we used 23 observed spectra from 1994 of richards & albright ( 1999 ) complemented by 24 new spectra from 1996/97 .
a detailed study of these spectra will be published in richards et al .
( 2005 ) and will include the actual measurements of the intensity of the central depression and the emission peaks .
.adopted properties of tt hya . [ cols= " < , < , < " , ] [ t2 ] designation : @xmath37 - limb darkening coefficient , @xmath177 - separation of the components , @xmath113 - temperature at the rotation pole , @xmath112 - gravity darkening exponent , @xmath178 - inclination , @xmath179 - vertical half thickness , @xmath180 - inner radius , @xmath181 - outer radius , @xmath182 - density , @xmath183 - electron number density , @xmath184 - exponent of the density power law , @xmath52 - microturbulence .
secondary was assumed to have synchronous rotation and to fill the roche lobe , vw - van hamme & wilson ( 1993 ) , e - etzel ( 1988 ) , l - lucy ( 1968 ) , brm - this work .
we performed some new and more sophisticated calculations of synthetic spectra in which both stars and a disc were considered .
the primary was treated as a rotating solid sphere while the surface of the secondary was modeled in terms of the roche model .
the stellar properties were based on the absolute parameters of the system obtained by van hamme & wilson ( 1993 ) and are summarized in table [ t2 ] . since our code accepts the polar temperature , @xmath113 , for the roche model rather than the average surface temperature , we adopted the value @xmath185k which is slightly higher than the average temperature @xmath186k obtained by van hamme & wilson .
their v - band limb darkening coefficients were adopted here .
the surface gravity , @xmath187 for the primary and @xmath188 for the secondary were assumed .
first , we obtained the model atmospheres of both stars by interpolating in the kurucz ( 1993b ) @xmath189 grid and assuming solar abundances and a microturbulence of 2 kms@xmath5 .
next , we calculated the intrinsic synthetic spectra emerging from these atmospheric models ( flux from the unit area of the surface ) using the code ( hubeny et al .
1994 ) modified by krtika ( 1998 ) . here
again , solar abundances and a microturbulence of 2 kms@xmath5 were assumed .
these synthetic intrinsic spectra were then assigned to the primary and secondary for the calculation of the complex spectra of both stars and the disc using . based on the lessons learned in the previous section we manipulated the free parameters of the disc so that the observed spectra could be reproduced .
we assumed that the electron number density is equal to the hydrogen number density .
a good fit to the observations was achieved with the realistic parameters for the disc .
these values are also summarized in table [ t2 ] .
figures [ f4 ] and [ f5 ] depict both the observed spectra as well as our synthetic fits for a few representative phases at both quadratures and during both eclipses to illustrate the most interesting effects .
the observed spectra display red and blue emission features as well as a central depression .
the observed far wings of @xmath0 stretching beyond the emission peaks in figure [ f4 ] do not seem as deep as in the synthetic spectra .
this is not a continuum rectification problem .
these shallower observed wings were previously noted by peters ( 1989 ) and interpreted as an emission stretching up to 30 from the line center .
however , these observed far @xmath0 wings in tt hya differ from single star standard spectra or from synthetic ones most probably because the atmospheric regions of the primary are more complicated .
a number of additional effects could be expected here like a suction ( `` spray gun '' ) effect of a disc moving over the atmosphere , centrifugal forces , and disc - atmosphere interaction with higher turbulence or temperatures .
all of them tend to reduce the effective gravity of the atmosphere of the primary and the atmosphere - disc transition region .
note , that the effective gravity is essentially zero in the disc .
this could change @xmath190 behavior of the atmosphere , mimic lower @xmath191 , reduce the pressure broadening , and , consequently , cause shallower line wings or shell spectra .
it might be misleading to interpret these shallower wings as an emission ( peters 1989 ) .
the emission peaks and the central depression depend strongly on the phase , as best illustrated in figure [ f6 ] , for all observed and synthetic spectra . a discussion of the phase dependence of @xmath192 , simultaneously on figures [ f4 ] , [ f5 ] , and [ f6 ] follows .
the first quadrature ( figure [ f4 ] : phase 0.217 ) seems to have the simplest spectrum . here
, we see a remarkably good agreement between the observed and theoretical profiles . both emission peaks have approximately similar strength
. a little bump on the red side of the central depression is the intrinsic @xmath0 line of the secondary .
as we proceed towards secondary eclipse , the blue emission gets apparently stronger than the red emission .
this should not happen if the disc is symmetric and its source function does not depend on the phase .
however , this effect is observed also in the synthetic spectra ( see figure [ f6 ] and figure [ f4 ] : phase 0.475 ) and is caused by the shadowing of the secondary by the disc . just before secondary eclipse the receding parts ( red emission ) of the disc projects onto the surface of the secondary .
this imposes different boundary conditions for intensity which is zero in the blue emission but nonzero in the red emission .
an extra absorption of the secondary s light in the receding part of the disc causes the red emission to be reduced .
this effect should be balanced in the middle of the eclipse as seen in figure [ f6 ] .
if this explanation is correct , the reverse ( stronger red emission ) should be observed after the middle of the eclipse at phases 0.5 - 0.6 .
unfortunately , these phases are not well covered by the observations of richards et al .
( 2005 ) but figure 4 of peters ( 1989 ) confirms that this is really the case . at the same time during secondary eclipse
both emission peaks grow slightly .
this is mainly because the stellar continuum drops during the eclipse and the contribution of the emission to the total light can get more pronounced .
this is also reproduced in the synthetic spectra but the observed emission peaks are still stronger than the theoretical ones which indicates that we may have underestimated the temperature of the secondary at this phase or overestimated the adopted limb darkening of the secondary .
this might also be due to the reflection effect which is not considered in our calculations .
the blue emission extends to higher radial velocities ( figure [ f4 ] ) .
this is not predicted by the synthetic spectra and our interpretation is that this is evidence of the gas stream . at this phase in the blue emission we are looking through more inhomogeneous regions encompassing the gas stream and disc - stream interaction regions which have higher than average radial velocity and
can be rather turbulent exhibiting the strongest departures from circular keplerian velocities .
this is probably also the reason why the observed central depression is displaced slightly to shorter wavelengths .
the observed central depression gets very deep at these phases , and reaches a minimum at about the middle of secondary eclipse .
it is deeper than in the theoretical spectra .
this huge depression could be caused by the combination of several effects : ( 1 ) at this phase the intrinsic @xmath0 lines of both stars overlap which deepens the central depression , ( 2 ) the reduction of the emission peaks by an additional absorption of light from the secondary by the disc might deepen the depression as well , ( 3 ) as pointed out above , if the surface of the secondary were hotter at this phase it would make the intrinsic line of the secondary , as well as the overall central depression , deeper ; ( 4 ) the eclipse of the secondary lowers the stellar continuum and changes both emission and absorption . something unexpected happens after the eclipse during phases 0.6 - 0.7 . the blue emission increases again .
this does not seem to be a temporal effect as it can be seen in the earlier observations of peters ( 1989 ; figure 4 ) .
one could expect to observe the reverse from the opposite line of sight within phases 0.1 - 0.2 but it is not the case .
could there exist some other circumstellar material apart from the disc and the stream in the system that contributes to the emission ? if this source were projected onto the surface of the secondary at phases 0.6 - 0.7 it might have the same effect as the disc at phases 0.4 - 0.5 and might increase the relative strength of the blue emission .
if this were the case , on symmetry grounds , we could expect the reverse ( red emission stronger ) at phases 0.3 - 0.4 .
there is only a very weak indication in figure [ f6 ] that this might be the case .
independent data of peters ( 1989 ) indicate the same effect that the red emission is slightly stronger at phases 0.3 - 0.4 .
however , if this matter projects onto the secondary during 0.6 - 0.7 and 0.3 - 0.4 then it would have to be outside of the roche lobe of the primary .
this may most probably be somewhere between the c1 and c2 roche surfaces .
the emission at the other quadrature ( figure [ f4 ] : phase 0.790 ) looks similar to that at 0.217 and is well reproduced in the calculations but now the observed central depression is weaker .
the central absorption is smallest shortly after the quadrature at about 0.79 - 0.86 ( figure [ f6 ] ) .
our interpretation is that this is evidence of the hot spot or hotter disc regions projecting onto the stellar surface of the primary which increases the source function along the line of sight and reduces the central depression .
a little bump seen now on the blue side of the central depression is the intrinsic secondary @xmath0 line .
the next four phases depicted in figure [ f4 ] ( phases 0.926 - 0.944 ) represent the eclipse of the approaching part of the disc and one can follow in both observed and theoretical spectra how the blue emission progressively ceases .
the behavior of both emission peaks is thus explained fairly well with our model .
red emission seems to be stretching more to the red than predicted but this is just the manifestation of the same effect observed close to the secondary eclipse for the blue emission .
this time , before the primary eclipse in the red emission we observe the gas stream with higher receding velocities than expected from circular keplerian orbits .
however , the observed central absorption is again deeper than expected .
total eclipse which follows is discussed on a separate figure ( figure [ f5 ] ) because of the different intensity scale .
the last four spectra in figure [ f4 ] ( phases 0.066 - 0.085 ) cover egress out of the disc eclipse .
this time , one can observe how the red emission associated with the receding part of the disc emerges .
again , the behavior of emission peaks is reproduced fairly well in the calculations , however , the observed central depression is much deeper than expected from our model and poses the most serious problem .
to some extent it may be caused by the similar effect invoked for the explanation of the depression near secondary eclipse i.e. , ( a ) overlap in radial velocities of stellar components ( which operates only outside of the total eclipse this time ) .
we also encounter ( b ) the eclipse of the low velocity emission regions that spill into the central absorption which deepens the absorption .
this is apparently not enough .
there are several possible explanations of this additional absorption just before and after the primary eclipse .
it can be ascribed ( 1 ) to the main body of the disc and assumed that the cooler disc regions project onto the hot surface of the primary slightly before but , mainly , after the total eclipse .
we reject this idea since such cooler regions in the main body of the disc should manifest in asymmetric emission peaks at the quadratures which is not observed .
another alternative is ( 2 ) to ascribe these deep @xmath0 cores to the inner and more dense regions of the disc as suggested by peters ( 1989 ) .
this might have no significant effect on the emission peaks but there is another problem with this interpretation and it is the strong phase dependence of this feature .
there is no obvious reason why this absorption if originating from inner disc regions should be phase dependent .
one could assume that an inner disc region close to a hot spot or a direct impact of the gas stream is hotter , but the matter would quickly cool towards the equilibrium value on time - scales much shorter than a keplerian period of the inner disc .
moreover , we observe that an enhanced absorption in the cores of @xmath0 are observed only shortly before and after the primary eclipse and quickly change and disappear within 0.1 orbital period between phases 0.85 - 0.95 and 0.05 - 0.15 . such an enhanced absorption is observed at the same phases in other algols e.g. , @xmath112 per ( richards 1993 ) that does not have a permanent accretion disc .
this leads us to suggest a third possibility ( 3 ) that this effect could be associated with some cooler matter in the vicinity of the secondary star or l2 point .
one could speculate that it may have the form of a slightly asymmetric envelope or a `` tail '' accompanying the secondary during its orbital motion or perhaps circumstellar matter populating the region between the c1 and c2 critical roche surfaces .
in either situation , it is plausible that such matter could be present between c1 and c2 , and could get there easily since the secondary already fills the c1 surface and is a magnetically active star .
the disc itself reaches almost up to the roche lobe of the primary and there is no reason to believe why it should have an abrupt end just before the roche surface and not spill beyond the surface
. the leading region ( in the sense of the orbital movement ) of such a secondary s envelope should be slightly hotter or less dense than the rear or tail of the envelope since the excess absorption is deeper after primary eclipse .
there is still more to understand .
note , that just before primary eclipse starts ( 0.90 - 0.95 ) the red emission is enhanced while the blue one is being eclipsed . the opposite ,
even more pronounced effect is observed after the eclipse ( 0.05 - 0.10 ) when the red emission emerges out of eclipse as the blue emission gets fainter .
it is not clear whether these features are permanent or transient or why the non - eclipsed emission should be affected during the eclipse of the other emission .
possible explanations include ( 1 ) that the disc itself contributes significantly to the total continuum and this contribution is reduced during the disc eclipse or ( 2 ) the primary is being shadowed by the extra matter between c1 and c2 ( as suggested above ) .
we made a test calculation for phase 0.0 with and without the disc and found that the contribution of the uneclipsed disc to the whole light in the continuum at @xmath0 at this phase was only about @xmath193 .
thus , the second option would be more plausible .
if the effect is real and the explanation correct one could expect a drop in the r - band photometry light curve at the corresponding phases by about 0.05 mag .
finally , figure [ f5 ] exhibits the spectra during total eclipse .
this is an important verification of the disc model since at this phase the main source of light , the primary star as well as the inner disc regions , are obscured .
the spectral lines of the secondary star can be seen more clearly and they indicate that the temperature of the secondary is probably slightly higher than the value of 4361k found by van hamme & wilson ( 1993 ) .
we adopted a temperature of 4600k at the rotation pole of the secondary which is reduced by about 200k on the back side of the secondary due to gravity darkening .
these phases also demonstrate that a slight decrease in radial density profile of the disc is necessary to get better agreement with the observations .
this radial density profile is described with the power law behavior @xmath194 in our model and the exponent @xmath195 seems to fit the data best .
both emission peaks agree fairly well with the synthetic spectra .
it is interesting that the red emission looks slightly wider than the blue one .
this is the manifestation of the same effect described above at the phases just before primary eclipse or of the reversed effect seen close to secondary eclipse when the blue emission was wider because of the enhanced departures from a keplerian velocity field due to the gas stream and stream - disc interaction .
however , at total eclipse the contribution from the remote part of the disc close to the roche lobe and out of the disc plane are mainly seen .
these areas may exhibit the strongest departures from our simple prescription of the keplerian velocity ( single star model ) which might explain why the synthetic central depression is wider .
the calculated central depression during primary eclipse is slightly stronger than expected .
this can be due to several reasons : firstly , the red emission which is wider overlaps the absorption and also skews it toward the blue .
secondly , the synthetic spectra were calculated for the temperature of 4600k which may be attained at the rotational pole but the actual temperatures we observe at the back side of the secondary are lower by about 200k and the intrinsic balmer lines of the secondary would be noticeably weaker at such temperatures .
we would like to note at this point that the skew of the central depression to the blue at these phases ( close to the primary and secondary eclipse ) can be easily misinterpreted as evidence of mass outflow . at
the other phases , like quadratures , departures from the circular keplerian orbit may occur causing a shift of the central absorption to either side .
that is why we feel that such blueshifted central depression is not proof of the mass outflow from the disc and that this conclusion of peters ( 1989 ) should be revisited .
since there are so many free parameters of the disc we can not claim that our fit is a unique solution .
nevertheless , as shown in the previous section , the parameters affect the spectrum in specific ways that we now understand more clearly ( see figures [ f1]-[f3 ] ) .
consequently , such calculations can be used to put an independent constraint on many of these parameters .
the calculations are not very sensitive to the inner radius of the disc but they seem to indicate that it reaches the surface of the star , moreover , the density of the disc seems to increase towards the star .
this is expressed by the negative exponent @xmath184 .
however , the exponent is best constrained from the emission strength during the total eclipse which is also sensitive to the temperature of the secondary and if the latter is wrong a spurious value of @xmath184 could be obtained .
if we had adopted the cooler temperature of the secondary derived by van hamme & wilson ( 1993 ) , the effect would be to lower the continuum and make the emission peaks relatively stronger and the exponent @xmath184 would have to be even more negative .
the outer radius can be set much more reliably since it affects strongly the separation of the emission peaks .
the roche lobe radius of the primary is about 10.6@xmath23 so our value is realistic and is about 95% of the roche lobe .
this is in perfect agreement with the value suggested by peters ( 1989 ) who estimated that the disc spans 95% of the roche lobe .
the exact agreement is a coincidence since in practice the disc probably does not have an abrupt end but rather dissolves smoothly .
the vertical dimension of the disc is significant and is of the order of the dimension of the primary star .
it agrees with the estimates of peters ( 1989 ) and plavec ( 1988 ) since it is in the region where both their estimates overlap .
the temperature of the disc which we obtained should be viewed as a characteristic temperature of the @xmath0 emitting regions . in reality
, there may be strong local departures from such a homogeneous model .
regions with temperatures between 5000k to 30000k will contribute significantly to the balmer line emission but both cooler and hotter regions may manifest at different wavelengths or spectral lines and an image of the disc can change depending on the spectral feature studied . the temperature was inferred mainly from the central depression which is strongly variable and there may be other effects which are beyond the scope of the paper and may turn out to be important like nlte effects , a better treatment of line absorption coefficient and scattering at optically thick wavelengths . the densities are quite well determined since the overall emission is very sensitive to this parameter .
nevertheless , in reality , again strong local departures can be expected from such a simple density profile .
the densities we obtained throughout the disc decline from @xmath196@xmath8 at the inner radius to @xmath197@xmath8 at the outer radius .
they are in very good agreement with the estimate of peters ( 1989 ; @xmath198@xmath8 ) .
the assumed microturbulence is very uncertain . in the case of an optically thin disc
, it should be viewed as departures from the keplerian velocity field rather than a local turbulence , and we can claim only that these departures will not be higher than about 80 kms@xmath5 . however , these departures may be higher in the gas stream .
consequently , we can claim that the single star disc model with a keplerian velocity field ( as given by eq .
[ e1 ] ) can reproduce the shape of the line fairly well .
the regions of the disc out of the orbital plane and close to the roche lobe are most vulnerable to departures from this model .
the voigt profile is generally not a good approximation for hydrogen lines , however , it can be justified in the case of emission lines originating from low density regions where the line profile is determined mainly by the 3d velocity field .
finally , we would like to point out that the parallax measured by hipparcos corresponds to the distance of about @xmath199pc which is not in very good agreement with the previous estimates of etzel ( 1988 ) and plavec ( 1988 ) who both obtained distances of about 194 pc .
if the new distance is confirmed by future parallax measurements , then either the geometrical scale ( semi - major axis ) and stellar radii are smaller than those obtained by etzel(1988 ) , plavec ( 1988 ) or van hamme & wilson ( 1993 ) or the temperatures of the stars , mainly that of the primary , might have to be reduced slightly .
we have developed a new computer code called shellspec .
this is to our knowledge the only tool which can produce the synthetic light curves and synthetic spectra of interacting binaries taking into account optically thin moving circumstellar matter .
the code was applied to the well - known eclipsing algol - type binary system tt hya which has an accretion disc .
synthetic spectra of the system for all phases were calculated by taking into account spherical primary , a roche lobe filling secondary , as well as a disc .
for the first time , it has been verified by calculations of synthetic spectra that the double peaked emission in tt hya , and thus also in other long period algol - type eclipsing binaries , originates from circumstellar matter in the form of an accretion disc with a keplerian velocity field .
the effect on the spectrum of various free parameters was studied .
the temperature and inclination of the disc have the strongest effect on the depth of the central depression while the outer radius of the disc , the radial density profile , and the inclination affect mainly the position and separation of the emission peaks .
the overall strength of the emission is regulated mainly by the density and temperature .
the central depression was found to be created by two different effects : the geometry and the velocity field of the disc itself and by the cool matter projected onto the hotter stellar surface .
realistic estimates of the densities , temperature , geometry , and dynamics of the disc were obtained as a result of these calculations .
however , there are differences between the observed and synthetic spectra which assume a simple circular keplerian disc .
these can be attributed mainly to the gas stream or stream - disc interaction .
the presence of an additional source of circumstellar matter in the system between the c1 and c2 roche surfaces is suggested to account for the deep absorption in the @xmath0 cores detected near primary eclipse and various other observed features .
we would like to thank drs .
ivan hubeny , alon retter , and the referee , for their comments and suggestions , and dr .
pavel koubsk for kindly allowing us to use our jointly collected 1996/1997 spectra of tt hya prior to publication .
jb gratefully acknowledges grant support from penn state university and thanks dr .
konstatin getman , and department computer staff for their assistance with computer related problems .
this research was supported by the nsf - nato fellowship ( nsf dge-0312144 ) and partly by the vega grant no .
3014 from the slovak academy of sciences and the science and technology assistance agency under the contract no .
51 - 000802 . | the formation and properties of accretion discs and circumstellar material in algol - type systems is not very well understood . in order to study the underlying physics of these structures ,
we have calculated synthetic @xmath0 spectra of tt hya , which is an algol - type eclipsing binary with an accretion disc . both the primary and secondary stars were considered in the calculations as well as a disc surrounding the primary . the roche model for the secondary star
was assumed .
the synthetic spectra cover all the phases including primary eclipse and are compared with the observed spectra . the influence of various effects and free parameters of the disc on the emerging spectrum was studied .
this enabled us to put some constraints on the geometry , temperature , density and velocity fields within the disc .
differences found between the observed and synthetic spectra unravel the existence of a gas stream as well as a hotter disc - gas interaction region .
an additional cooler circumstellar region between the c1 and c2 roche surfaces is suggested to account for various observed effects .
a new computer code called was created for this purpose and is briefly described in this work as well .
it is designed to solve simple radiative transfer along the line of sight in 3d moving media . the scattered light from a central object
is taken into account assuming an optically thin environment .
output intensities are then integrated through the 2d projection surface of a 3d object .
the assumptions of the code include lte and optional known state quantities and velocity fields in 3d . |
recent spectroscopic surveys of the fornax and virgo galaxy clusters have revealed a new class of compact stellar system , `` ultra - compact dwarf galaxies '' ( ucds ) with properties intermediate between the larges globular clusters and the smallest dwarf galaxies .
the defining properties of the first ucds found are that they are significantly more luminous than most known globular clusters ( @xmath0 ) , but they are mostly unresolved in ground - based sky survey images . in this paper
we present detailed observations of a new sample of ucds in the virgo cluster in order to test hypotheses for the formation of these objects . the fornax cluster ucd objects were discovered independently in studies of globular cluster systems around the central galaxy ngc 1399 ( minniti et al . 1998 ,
hilker et al .
1999 ) and in studies of compact dwarf galaxies in the cluster ( drinkwater et al .
2000 , phillipps et al .
2001 ) . confirmed ucds have subsequently been found in the virgo cluster ( haegan et al .
2005 , jones et al . 2006 ) and ucd candidates identified in the more distant cluster abell 1689 ( mieske et al.2004 ) . at low luminosities ( @xmath4 ) the distinction between ucds and globular clusters is not clear ( see discussions in drinkwater et al.2004 , mieske et al .
2006 ) , but in this paper we focus on the most luminous objects ( @xmath0 ) . given their intermediate nature , most formation hypotheses for ucds relate them to either globular clusters or dwarf galaxies
. they could be highly luminous intra - cluster globular clusters ( e.g. mieske et al .
alternatively , ucds may result from the tidal disruption of nucleated dwarf elliptical galaxies .
this process can leave the nucleus intact on intra - cluster orbit ( bekki et al .
2001 , 2003 ) as a ucd .
other formation scenarios include ucds being the evolved products of massive super starclusters formed in galaxy mergers ( fellhauer & kroupa 2002 ) , or primordial objects ( phillipps et al.2001 ) .
it is hard to provide definitive observational tests of these different scenarios for ucd formation .
this is partly because the various scenarios do not always make very different predictions , but the observational picture also remains far from complete .
the observational tests can broadly be divided into statistical population studies ( e.g. the distribution of ucds compared to other objects ) and detailed studies of the internal properties ( e.g. internal velocity dispersion ) of the ucds . in this paper
we focus on the latter .
our previous investigations have focused on the fornax cluster ucds .
we used _ hubble space telescope _ ( hst ) imaging and eso very large telescope spectroscopy to compare the five original ucds to globular clusters and nucleated dwarf galaxies ( drinkwater et al .
we concluded that ucds were distinct from known globular cluster populations as they followed a different relation between internal velocity dispersion and luminosity .
the ucd properties were , however consistent with the threshing model in which the dwarf galaxy nuclei survived tidal disruption with no significant change to their luminosity or velocity dispersion .
the tidal disruption hypothesis was also supported by haegan et al.(2005 ) in their analysis of 10 compact `` dwarf - globular transition objects '' found in hst images of virgo cluster galaxies .
the transition objects of haegan et al . have slightly lower luminosities ( @xmath5 ) than the fornax ucds , but the brighter ones were found to be significantly different to globular clusters , both in size and in following galaxy scaling relations .
haegan et al .
suggest that `` bona fide '' ucds are also distinguished by the presence of dark matter as some of their objects have @xmath6 . a detailed analysis of the stellar populations of compact objects in the fornax cluster ( mieske et al .
2006 ) has reached quite different conclusions .
they measured spectroscopic metallicities for 26 compact objects with luminosities spanning both ucds and globular clusters and found a break in the distribution at about @xmath7 .
the more luminous objects have a narrow metallicity distribution with mean [ fe / h]@xmath8 , whereas the less luminous objects show a much broader range of metallicity and a significantly lower mean ( 0.56 dex lower ) .
there is a break in the size - luminosity relation for these objects at the same luminosity .
mieske et al .
note that the metallicity of the dwarf galaxy nuclei in their sample is significantly lower than that of the ucds which , in turn , are better matched by models of massive young star clusters .
they therefore suggest that the ucds in the fornax cluster are formed as a result of galaxy mergers , but note that the properties of the virgo cluster ucds are more consistent with the stripping model .
in this paper we present new imaging and spectroscopic observations of the virgo cluster ucds listed by jones et al .
these are analysed along with existing data for the luminous fornax cluster ucds ( drinkwater et al .
in section 2 we describe the high - resolution spectroscopic observations , and in section 3 we describe the corresponding hst imaging .
we present an overview of our results in section 4 by comparing the ucds with other objects in various projections of the fundamental plane . in section 5
we investigate the ucd dynamics in more detail , calculating their mass - to - light ratios , and in section 6 we examine the ages and chemical composition of their stellar populations .
our main results and conclusions are given in section 7 .
we use distance moduli of 30.92 mag and 31.39 mag for the virgo and fornax clusters respectively ( freedman et al .
observations of the six virgo ucds , a comparison m87 globular cluster ( strom417 ) , a comparison dwarf galaxy nucleus ( vcc417 ) and the ngc4486b galaxy ( its central part ) were carried out on 2003 april 67 with the echelle spectrograph and imager ( esi ) on the keck ii telescope in the echellette mode ( table 1 ) .
a slit width of 075 was used , providing an instrumental resolution of @xmath9 50 km s@xmath10 ( fwhm ) or @xmath11 6,000 .
the wavelength range is @xmath12 , distributed over 10 echelle orders , with a dispersion of 11.4 km s@xmath10 pixel@xmath10 .
the exposure times are given in table 1 .
in addition , standard stars of spectral types in the range between g8iii and m0iii ( table 2 ) were observed for use as templates for the radial velocity and velocity dispersion measurements and line index calibrators .
the standard stars were observed in two ways : held centered in the slit , and also moving perpendicularly across the slit ( `` smeared '' ) to simulate the appearance of extended sources which overfill the slit and consequently have slightly lower spectral resolution .
the seeing was a stable 08 the first night , and ranged from 08 to 11 the second .
the first night was photometric ; the second had occasional light cirrus .
the data were reduced using scripts in iraf and idl specifically written to handle the esi data format and idiosyncrasies , but otherwise are standard procedures for ccd data .
spectra were extracted over a 15 aperture centered on the peak , taking in nearly all the light from the ucds and globulars .
the relative flux scale was determined using nightly observations of feige 34 .
the s / n for each integration ranges from @xmath13 to 25 per pixel , yielding final s /
n of 30 to 50 after coadding the multiple observations of each target .
the radial velocity and velocity dispersion of our objects were determined using two different techniques : the direct - fitting method ( as described and implemented by van der marel 1994 ) and the cross - correlation method ( tonry & davis 1979 , as implemented in rvsao / iraf ) . in the direct - fitting method the template star spectrum ( table 2 ) is broadened with gaussian functions of variable @xmath14 in velocity space .
the resulting set of spectra are then compared with the object spectrum .
the best - fitting gaussian function is determined by @xmath15 minimization in pixel space .
the second method is to cross - correlate the object spectrum with the stellar template spectrum to determine the width of the cross - correlation peak and the redshift .
the correlation width is used to estimate the velocity dispersion by comparison with results from artificially broadening the template stars by convolution with gaussian functions of known width .
this method is less sensitive to the exact match between template and object spectra , than the direct - fitting method .
velocities and velocity dispersions were obtained from the cat ( @xmath16 ) and mgb ( @xmath17 ) regions .
the measured values are given in tables 3 & 4 and are consistent , within the measurement errors , for the two techniques and the two wavelength regions .
the exception is vucd3 . in the following discussion ,
we use values obtained as the mean of the two wavelength regions and the direct - fitting method as it gives smaller measurement errors . the adopted velocities and velocity dispersions for the observed objects
are shown in table 5 .
our measurements for ngc4486b are in good agreement with bender et al .
( 1992 ) .
there were almost no differences in velocity and velocity dispersion measurements obtained with smeared and unsmeared stellar templates .
the two wavelength regions give significantly different velocity dispersions for vucd3 : @xmath9 49 km s@xmath10 for the mgb region , but only @xmath9 38 km s@xmath10 for the cat region .
this discrepancy appears to be real .
we suspect this may be because the ratio of giant to dwarf population contributions to the continuum is varying rapidly in this object compared to the other ucds .
this might be expected because this object also has a much higher metallicity than the other ucds ( see section 6 below ) , so that the relative contribution of the higher gravity dwarf stars to the blue - green ( mgb ) region of the spectrum is greater , leading to broader lines in this region compared to the other ucds .
the cat region is dominated by giant stars , no matter what the metallicity of the ucds .
modeling of this effect could lead to a better understanding of the stellar populations , but a more extensive library of stars is required .
we obtained images of the virgo ucds in the course of hst snapshot program 10137 .
the data were taken with the advanced camera for surveys ( acs ) , high resolution channel ( hrc ) through the f606w and f814w filters .
exposure times were 870 sec in f606w and 1050 sec in f814w .
the hrc scale is @xmath18 pixel@xmath10 . for the image analysis we used multidrizzle ( mdz ) files retrieved from the hst archive . to measure the total magnitudes , we plotted curves of growth ( integrated magnitude versus circular aperture radius ) to find an aperture radius large enough to enclose all the light from an object .
the instrumental f606w and f814w magnitudes were transformed into landolt @xmath19 and @xmath20 band following sirianni et al .
the resulting @xmath19 magnitudes and @xmath21 colors are listed in table 6 .
we have also used hst imaging data ( program 8685 ) for fornax ucds and a de , n ( fcc303 ) , initially presented in drinkwater et al .
( 2003 ) , as one of the aims of this work is to compare virgo ucds with fornax ucds and dwarf nuclei .
the data consist of 1960 sec exposures taken with the space telescope imaging spectrograph ( stis ) in unfiltered mode ( 50ccd ) .
the stis has a scale of @xmath22 pixel@xmath10 .
we re - processed the stis images with multidrizzle to ensure that the data reduction method for the fornax ucd images was consistent with that of the virgo ucd images .
the instrumental ab magnitudes were transformed into @xmath19 band using the relation @xmath23 ( h.ferguson , private communication ; see also gregg & minniti 1997 ) .
we used @xmath24 colors for ucds and fcc303 from karick et al .
the total @xmath19 magnitudes for fornax ucds and fcc303 are presented in table 7 .
the images of virgo and fornax ucds were modeled using two - dimensional fitting algorithm galfit ( peng et al .
2002 ) and assuming empirical king , sersic and nuker models for the luminosity profile .
the king profile is characterized by the core radius , @xmath25 , and the tidal radius , @xmath26 , and has the following form : @xmath27^{\alpha},\ ] ] where i@xmath28 is the central surface brightness .
we tried both the standard model with @xmath29 and generalized model with variable @xmath1 .
the king model provides a good fit to globular cluster luminosity profiles when the concentration index @xmath30 and to elliptical galaxy luminosity profiles if @xmath31 ( mihalas & binney 1981 ) .
the sersic power law is often used to fit luminosity profiles of galaxies and has the following form : @xmath32,\ ] ] where @xmath33 is the half - light ( effective ) radius , @xmath34 is surface brightness at the effective radius , n is the concentration parameter ( n=4 for de vaucouleurs profile and n=1 for exponential profile ) and k is a constant which depends on n. the nuker law is as follows : @xmath35^{\frac{\gamma - \beta}{\alpha}}.\ ] ] it is a double power law , where @xmath36 is the outer power law slope , @xmath37 is the inner slope , and @xmath1 controls the sharpness of the transition ( at the `` break '' radius @xmath38 ) from the inner to the outer region . @xmath39 .
the nuker law was introduced by lauer et al .
( 1995 ) to fit galaxy centers .
galfit convolves the analytic profile with the psf and determines the best - fitting model parameters by minimizing residuals between the model and original _ two - dimensional _ image .
we derived artificial psfs for acs / hrc images in the f606w and f814w filters using the tinytim software and multidrizzle as follows .
first we generated acs / hrc psfs with tinytim ; these include all the distortions , so they represent the psf in raw images .
we then implanted these psfs in empty distorted images ( flt files ) , _ at the location of each target observed _ , and passed them through multidrizzle using the same parameters as were used for the data .
this produces model psfs that are processed the same way as the real data . for the stis imaging ,
the generation of the psf is more straight - forward and is achieved through a single pass of tinytim .
the quality of the galfit model fits in the inner regions of each object is shown in figures 1 & 2 .
these figures present residual maps after subtracting the ( psf - convolved ) model from each object .
the quality of the models in the outer regions is better shown in figures 3 & 4 .
for these figures we used the ellipse task in iraf to produce _ one - dimensional _ surface brightness profiles for the objects and ( psf - convolved ) galfit models .
the @xmath40 values of the fits ( see peng et al .
2002 ) are shown in tables 8 & 9 .
we use @xmath40 values to choose the _ best model _ for each object ( see the last row of tables 8 & 9 ) . from figures 1 - 4 and the @xmath40 values _ in both filters _ we can see that the _ virgo _ ucds are poorly fitted with the _ standard king model _ ( @xmath29 ) , but are very well fitted with the _ nuker law _ a double power law ( except vucd7 which requires a two - component model ) .
king models predict a truncation radius , beyond which stars are stripped from the cluster by the galactic tidal field . _ none of the ucds show this tidal cutoff down to our limiting surface brightness .
_ also , a main feature of king models is their central cores the regions of constant surface brightness .
we have found from nuker models that only one ucd has a flat core ( vucd5 ) , all other ucds have shallow or steep cusps in the center ( table 8) .
_ generalized king models _
( with a variable @xmath1 parameter ) provide better fits to the data than standard king models .
the parameter @xmath1 controls both the slope of the profile , and the transition from the main body into @xmath26 .
when we relax the @xmath1 parameter , we can better fit the extended outer parts of the ucds .
however , king models have a tidal cutoff and do not fit the data as well as untruncated nuker models do .
to fit a standard king model to the f606w data for vucd3 , we had to fix @xmath25 at 1 pixel ( 1.9 pc ) .
there is no convergency if we leave @xmath25 as a free parameter ( perhaps because @xmath25 becomes too small , @xmath41 1 pixel ) .
the lack of a good king fit to the f606w data for vucd3 is seen in figure 3 .
we also failed to fit any of the king models to the f814w data for this object .
there was no convergency with either @xmath25 fixed or relaxed .
_ sersic models _ do not provide good fits to the virgo ucds either . in the majority of cases , the model profiles drop faster than the data .
the fit seems good in the case of vucd3 , but this object has a very steep profile ( a very high index n ) . models with a sersic index n @xmath42 2 become very sensitive to the sky determination , flatfielding , the accuracy of the psf being used , and how well the assumed profile agrees with the data ( see on - line galfit manualcyp / work / galfit / tfaq.html ] ) .
the more centrally concentrated a galaxy profile is ( the larger the sersic index n is ) , the more extended outer wing it has . because of this behavior ,
if ( for example ) a profile already has a high intrinsic index n , a small underestimation of the background can make n even higher and can cause large errors in the magnitude and size .
when fitting vucd3 data in the two passbands , we obtained quite different parameters : e.g. the difference in @xmath43 was 39% .
this could be due to the above reasons .
we conclude that the nuker law appears to be the best model for all virgo ucds except vucd7 , in which case the king core + sersic halo model is the best one .
the situation is different for the _ fornax ucds_. there is no universal model for them .
two objects ( ucd3 and ucd5 ) are best fitted by two - component models .
the other three have very steep profiles which are hard to model . in the case of ucd1
the best model is sersic .
the best model for ucd2 is the generalized king law and in the case of ucd4 it is the nuker law .
the residual maps of fornax ucd3 reveal faint structure to the north - west of the core .
this is very likely a background spiral galaxy along the line of sight .
this structure affected the profile fits to ucd3 to the extent that the two - component models gave very inconsitent results .
for this object we therefore restricted the model fits to the bottom half of the image .
the model parameters for ucd3 in all the tables are from these fits to half the images , although we show the images and plots in figures 2 & 4 for the whole - image fits to reveal the background object .
the ucd structural parameters obtained from the fits are summarized in tables 8 & 9 .
the nuker fits gave us the outer power law slope @xmath36 , the inner slope @xmath37 , @xmath1 parameter , the `` break '' radius @xmath38 , the surface brightness of the profile at the `` break '' radius @xmath44 and ellipticity @xmath45 ( b / a is the minor - to - major axial ratio ) . from sersic models , we obtained the half - light radius @xmath33 , sersic index n , ellipticity @xmath46 and the integrated magnitude @xmath47 .
the king fits gave us the core radius @xmath25 , tidal radius @xmath26 , concentration parameter @xmath48 , central surface brightness @xmath49 and ellipticity @xmath46 .
all the galfit models assume a constant ellipticity which is fitted at the same time as the other parameters .
the @xmath33 values and the integrated magnitudes for the king and nuker models were obtained via numerical integration of the luminosity profiles ( defined by formulae 1 and 3 ) .
we also obtained half - light radii from the observational data directly ( @xmath50 ) . in this case the data were not psf - deconvolved and the half - light radii may be overestimated . the structural parameters for virgo ucds
were averaged between the two filters . for further analysis we had to decide what @xmath33 to choose for our objects .
we could not take just the best model @xmath33 because , for example , in the case of virgo ucds the best model is nuker .
for all the nuker models in tables 8 & 9 ( except for vucd1 & vucd5 ) the total volume under the profile ( the integrated luminosity ) converges very slowly .
it makes @xmath33 estimations uncertain and results in unphysically large @xmath33 values .
this is why we chose generalized king @xmath33 values .
the generalized king models fit the virgo ucd data better than standard king and sersic models , and they are finite in extent .
in addition to this , generalized king @xmath33 values are consistent with the observational half - light radii ( @xmath50 ) . in the case of the fornax ucds we also chose the generalized king @xmath51 for one - component ucds and the king @xmath33 for the cores of the two - component objects .
the sersic fits seem unreliable as all of the models have very high index
n ( see above ) .
nuker models are not good for @xmath33 estimation because of the reason explained above .
we list the best values of the various parameters of the virgo and fornax objects in table 10 .
these are used in the following section for analysis of the scaling relations . in table 10
we quote three values of effective radius @xmath33 in the case of two - component objects ( vucd7 , ucd3 , ucd5 & fcc303 ) : the total , only the core , and only the halo .
these were obtained via numerical integration of the model luminosity profiles .
table 10 also contains the total v band apparent magnitude @xmath52 , the mean surface brightness within the effective radius @xmath53 and ellipticity @xmath46 .
the @xmath52 values are observational values copied from tables 6 & 7 except for the core and halo magnitudes of the two - component objects , which are model values taken from tables 8 & 9 ( king+sersic models ) .
the @xmath53 values were derived from @xmath33 and @xmath52 as follows : @xmath54 where @xmath33 is measured in arcsec .
the @xmath46 figures are the best model values from tables 8 & 9 .
the ellipticities of the ucds given in table 10 show that some of the objects are significantly non - circular ; the maximum ( core ) ellipticity is 0.24 for fornax ucd5 .
we have compared the distribution of ucd ellipticities with those reported for globular clusters in ngc 5128 ( harris et al .
2002 ) and the milky way ( harris 1996 ; online catalogue version of 2003 february ) .
the two - sample kolmogorov - smirnov test indicates that the ucd ellipticities are consistent with both the mw gc distribution the significance level is 53% or 66% ( depending on the ellipticity used for two - component ucds : core or halo)and the ngc 5128 gc distribution
the significance level is 38% or 63% .
the wilcoxon test gives similar results : the ucd ellipticities are consistent with the mw gc distribution at the 42% or 36% significance level and with the ngc 5128 gc distribution at the 27% or 46% significance level .
in this section we compare ucds with globular clusters and galaxies by their position in both the luminosity velocity dispersion plane and @xmath55-space ( the fundamental plane as defined by bender et al .
1992 ) .
first we revise the luminosity - velocity dispersion correlation for ucds and other types of stellar systems proposed in drinkwater et al .
( 2003 ) . figure 5 represents the absolute v magnitude vs. central velocity dispersion relation for fornax and virgo ucds , globular clusters ( including the most massive and luminous ones : g1 in m31 , the galactic globular cluster @xmath56 cen , and ngc5128 massive gcs ) and galaxies .
the ucd data were obtained in the present work ( tables 10 & 12 ) , except central velocity dispersions for fornax ucds , which were taken from hilker et al .
the velocity dispersion for fornax ucd1 was derived from cat region using the cross - correlation method and only one stellar template ( g6/g8iiiw type ) .
the data for globular clusters are from : m31 gcs djorgovski et al .
( 1997 ) and references therein ; g1 : djorgovski et al . ( 1997 ) and references therein , except half - light radius ( required for figure 6 ) , which was taken from barmby et al .
( 2002 ) ; milky way gcs , @xmath56 cen , lmc and smc gcs ( most of which have old ages ) mclaughlin & van der marel 2005 ( photometry is based on wilson models ) ; ngc5128 gcs martini & ho ( 2004 ) ; strom417 : spectroscopy this work ( table 12 ) , photometry haegan et al .
we also plot `` dwarf - globular transition objects '' from haegan et al .
data for galaxies were obtained from : giant ellipticals
faber et al . ( 1989 ) ; ngc4486b : spectroscopy our data , bender et al . ( 1992 ) , photometry faber et al . ( 1989 ) ; the compact elliptical galaxy m32 faber et al . ( 1989 ) , bender et al . ( 1992 ) ; de , ns and dwarf nuclei
geha et al .
( 2002 , 2003 ) ; fcc303 : photometry this work , spectroscopy hilker et al . ( 2006 ) ; vcc1407 : velocity dispersion this work , magnitude ned .
all the @xmath57 magnitudes were dereddened and all the velocity dispersions for gcs and ucds were aperture - corrected to give standard values ( the central velocity dispersions ) used in the literature for the comparison with galactic gcs . from figure 5 we can see that there is no gap between bright gcs and ucds and that the virgo ucds have velocity dispersions and luminosities similar to the fornax ucds .
vucd3/strom547 and strom417 were previously considered to be m87 gcs ( strom et al .
these are the two brightest m87 gcs according to hanes et al .
( 2001 ) list .
now we see that these gcs lie in the same part of @xmath58 plane as ucds and can be equally considered as ucds .
the ucds , along with transition objects of haegan et al . , appear to follow approximately the same relation between luminosity and velocity dispersion as old globular clusters .
to make a firm conclusion if the ucds lie on the extrapolation of the gc relation or not , more data on the velocity dispersions for bright gcs ( such as ngc5128 and m31 globulars ) or fainter ucds ( e.g. recently discovered in the fornax cluster , drinkwater et al .
2004 ) are required .
there is an overlap in luminosities and velocity dispersions of the de , n nuclei and the properties of bright gcs , transition objects of haegan et al . and
ucds , which is consistent with the stripping hypothesis for gc , transition object and ucd formation .
next we consider the location of ucds relative to other stellar systems in the @xmath55-space ( figure 6 ) .
the @xmath55-space is a space in which axes are combinations of three observable parameters ( central velocity dispersion , effective radius and mean intensity inside effective radius ) into physically meaningful parameters .
the @xmath55-space parameters as defined by bender et al .
( 1992 ) are as follows : @xmath59 @xmath60 @xmath61 where @xmath62 is the central velocity dispersion in @xmath63 , @xmath33 is the half - light radius in kpc , @xmath34 is the mean intensity inside @xmath33 , defined as @xmath64 and measured in v - band solar luminosities pc@xmath65 .
the @xmath55 variables have the following physical meanings : @xmath66 is proportional to the logarithm of the mass , @xmath67 is proportional to the logarithm of the surface brightness cubed times the mass - to - light ratio , and @xmath68 is proportional to the logarithm of the mass - to - light ratio . as expected , objects are distributed more widely in the @xmath66-@xmath67 plane , the face - on projection of the fundamental plane , than in the @xmath66 - @xmath68 , the edge - on projection . in the @xmath66 - @xmath68 plane
we show that the ucds lie on the same tight correlation between mass and mass - to - light ratio as the bright gcs and transition objects of haegan et al . , but the fainter gcs ( @xmath69 ) show little if any correlation in this plane .
this corresponds approximately to a mass of @xmath70 m@xmath71 at which haegan et al .
( 2005 ) also find a turn - over in scaling relations for gcs and other low - mass systems in other projections of the fundamental plane .
consistent with previous discussions ( e.g. burstein et al .
1997 ) we find that this relation does not intersect that of giant elliptical galaxies in the same @xmath66 - @xmath68 plane .
by contrast , in the @xmath66 - @xmath67 plane we find that the ucds are clearly not on the main gc relation ( as defined by the milky way and m31 gcs ) .
the ucds lie in a region away from this sequence in the direction of increasing mass ( @xmath66 ) .
the ngc5128 gcs also lie off the main gc relation between it and the ucds .
we must note here that the ucds are from magnitude - limited samples ( equivalent to @xmath72 ) , so these data do not actually provide evidence for a gap between ucds and gcs .
there is a relatively empty region in the plane at ( @xmath73 ) values of around ( 0,3.54.5 ) at masses intermediate between gcs and ucds but low @xmath67 values . according to bastian et al .
( 2006 ) , young massive star clusters , if evolved to ages of 10 gyr , would occupy this region of the fundamental plane .
bastian et al .
note that not many of these are expected to survive to such ages , but even those not disrupted would be unlikely to appear in the existing observational data sets due to the following selection effects .
first , there are no ucds in this region as our ucd samples are limited to higher masses ( luminosities strictly speaking ) .
secondly , gcs in this region would have high mass and presumably would only be found in external galaxies like the ngc5128 objects , but they would have much lower surface brightness values , so it would not be possible to measure their velocity dispersions even if they could be detected .
it is interesting to note that the more massive objects ( @xmath74 ) show the opposite correlation between @xmath66 and @xmath67 to the lower - mass systems in figure 6 . since @xmath68 ( @xmath75 ) is increasing for the high - mass objects , it must be @xmath76 that is decreasing for these objects .
this agrees with the mass - size relation observed for objects in this mass range ( e.g. kissler - patig et al . 2006 and haegan et al .
2005 ) . in both the fundamental plane projections
we also show the parameters of the nuclei of a sample of nucleated dwarf elliptical galaxies in the virgo cluster ( geha et al .
the nuclei with similar masses to the ucds lie in the same region of both plots as the ucds .
this is consistent with the threshing hypothesis for ucd formation from disrupted dwarf elliptical galaxies .
the ucds are mostly well - separated from complete dwarf elliptical galaxies in the fundamental plane ; the closest galaxy to ucds is m32 , the prototype `` compact elliptical '' galaxy which is also thought to have formed through a disruptive process ( e.g. choi et al . 2002 ) .
in this section we estimate the masses of the ucds using dynamical models .
the masses and mass - to - light ratios of the ucds are important physical parameters for the understanding of their origin .
in particular , the mass - to - light ratio ( @xmath75 ) is an indicator for possibly existing dark matter and/or violation of dynamical equilibrium or isotropy of stellar orbits .
if ucds were the counterparts of globular clusters , one would expect @xmath75 values as predicted by simple stellar population models ( e.g. bruzual & charlot 2003 , maraston 2005 ) .
if ucds were of `` galaxian origin''formed in dark matter halos they might still be dominated by dark matter and show high @xmath75 values .
mass - to - light ratios larger than expected from simple stellar populations can , however , also be caused by objects that are out of dynamical equilibrium , e.g. tidally disturbed stellar systems ( fellhauer & kroupa 2006 ) .
the ucd masses were estimated from the measured velocity dispersions and their structural parameters .
we showed in section 3 that the ucd light profiles can be fitted by various functions .
a simple king profile often is not the best choice to represent ucd surface brightness profiles .
however , most mass estimators available in the literature are based on the assumption of a king profile . in order to be not restricted to king profiles
, we used a more general approach , using software developed by h. baumgardt ( see hilker et al .
2006 ) .
the first stage is to deproject the observed density profile ( either king , generalized king , sersic , nuker , or king+sersic ) , calculate its distribution function @xmath77 under the assumption of spherical symmetry and an underlying isotropic velocity distribution , and finally create an @xmath78-body representation of the ucd .
it is assumed that mass follows light ( e.g. mass segregation is neglected ) . besides the projected profile parameters , the total mass of the stellar system and the number of test particles are needed .
the resulting model is a list of @xmath79 , @xmath80 and @xmath81 positions and @xmath82 , @xmath83 and @xmath84 velocities for all particles that correspond to the specified structural parameters and the given mass . from this model , the central as well as global projected velocity dispersion can be calculated .
the projected half - mass radii for all the models were also derived and were in very good agreement with the half - light radius values in tables 8 & 9 . in the second stage ,
the velocity dispersion seen by an observer is simulated . in order to do this
, all test particles are convolved with a gaussian whose full - width half - maximum ( fwhm ) corresponds to the observed seeing .
the fraction of the light ( gaussian ) falling into the slit at the projected distance of the observed object ( the size of the slit in arcsec and the distance to the object in mpc are input parameters ) is then calculated .
these fractions are used as weighting factors for the velocities .
the weighted velocities of all particles whose `` light '' falls into the slit region are then used to calculate the mimicked observed velocity dispersion @xmath85 .
the total `` true '' mass of the modelled object , @xmath86 , that corresponds to the observed velocity dispersion , @xmath87 is not known a priori .
one has to start with a first guess of the total mass , @xmath88 , from which the `` true '' mass can be calculated as @xmath89 . in the case of the nuker and sersic functions
, the models were truncated at large radii to avoid the unphysical infinite extensions of ucd light profiles .
the truncation radius of the nuker model was fixed to 2 kpc .
the truncation radius of the sersic model was set to 20 times the effective radius , thus ranging between a few hundred parsecs and a few kiloparsecs for the ucds in our sample .
the true tidal radii of the ucds depend on their distances to the cluster center @xmath90 and the enclosed mass @xmath91 of the potential they are living in .
they can be estimated by the formula : @xmath92 , where @xmath93 is the circular velocity of the cluster potential and @xmath94 is the gravitational constant .
the estimated tidal radii of the ucds range between 1 and 4 kpc , thus justifying the chosen truncation radii .
a more detailed description of the mass determination process and mass values for fornax ucds are presented in hilker et al .
( 2006 ) . in this paper
the discussion will be focused on the virgo ucds .
the uncertainties of the modelled masses were estimated from the uncertainty in the observed velocity dispersion , @xmath95 and the differences of the model parameters for surface brightness profiles in the @xmath19 and @xmath20 band .
on one hand , maximum and minimum model masses were simulated that correspond to the observed velocity dispersion of @xmath96 and @xmath97 , respectively .
the average of the differences @xmath98 and @xmath99 defines the first mass uncertainty .
on the other hand , models were created that simulate @xmath87 from the profile parameters in @xmath19 and @xmath20 separately as well as from combinations of their parameters ( i.e. @xmath25 of @xmath19 and @xmath26 of @xmath20 ) to mimic the uncertainties in the profile parameters .
the maximum and minimum mass deviations from @xmath86 define the second uncertainty .
both uncertainties then were summed to derive the total mass uncertainty .
the uncertainties for the mass - to - light ratios were propagated from the mass uncertainties and the uncertainty in the luminosity ( assumed to be 0.05 mag in the absolute magnitude ) .
the luminosities were derived from the apparent @xmath19 magnitudes given in table 10 , the distance to the virgo cluster as mentioned in the introduction and a solar absolute @xmath19 magnitude of @xmath100 mag .
the uncertainty of the central velocity dispersion was estimated from the observational uncertainty plus the scatter of modelled velocity dispersions in annuli of 0.5 parsec within the central 5 pc for each object .
the uncertainty of the global velocity dispersion is the sum of the observational uncertainty and the uncertainties as propagated from the mass uncertainty of the profile fitting parameters .
the results of the modelled object masses and velocity dispersions for the king , generalized king , sersic , nuker and king+sersic functions are presented in table 11 .
these results are based on the surface brightness profile parameters in table 8 and the observed velocity dispersions in table 5 . for strom417 we used surface brightness profile parameters obtained by haegan et al .
( 2005 ) from king model fits .
the masses and @xmath75 values of the different models in general agree with each other within the uncertainties . on average ,
the masses of the nuker profile models are slightly higher than those derived from the other profiles , whereas the masses of the sersic profile models are on the low side .
as discussed in section 3 , the generalized king models give the most stable estimates in the case of one - component profile fits .
therefore , the masses from the generalized king models and from the king core plus sersic halo models in the case of vucd7 were adopted for further analyses ( see table 12 ) .
the virgo ucds have masses and mass - to - light ratios in the range @xmath101 and @xmath102 .
the fornax ucd masses and mass - to - light ratios are in the same range ( hilker et al .
2006 ) . in order to compare our results with other dynamical mass estimators we calculated masses using the king mass estimator ( e.g. queloz et al . 1995 ) and the virial mass estimator ( spitzer 1987 ) .
both methods assume a constant @xmath75 ratio as function of radius , an isotropic velocity distribution and that the object is in virial equilibrium .
we consider the ucds are in virial equilibrium as their ages ( estimated in the following section ) are much greater than their crossing times ( @xmath103 myr ) .
the king mass estimator takes the form : @xmath104 where @xmath62 is the central projected velocity dispersion ( from table 11 , standard king model ) , @xmath25 is the core radius ( from tables 8 & 9 , standard king model ) , @xmath94 is the gravitational constant and @xmath105 , @xmath1 and @xmath106 are constants which depend on the concentration , @xmath107 , and are tabulated in king ( 1966 ) and peterson & king ( 1975 ) .
@xmath105 , @xmath1 and @xmath106 are the mean of the two passbands for virgo ucds , except for vucd6 . in the case of vucd6
we used @xmath19 passband only , @xmath20 passband gives unreasonably high mass estimate .
@xmath25 , @xmath105 , @xmath1 and @xmath106 for strom417 were taken from haegan et al .
( 2005 ) .
the virial mass estimator is as follows : @xmath108 where @xmath14 is the global projected velocity dispersion ( from table 12 ) and @xmath33 is the half - light radius ( from table 10 ) .
@xmath33 for strom417 was taken from haegan et al .
( 2005 ) .
the results of these two mass estimators are given in table 13 .
the uncertainties of the calculated masses were propagated from the uncertainties in @xmath25 and @xmath62 for the king mass estimator and the uncertainties in @xmath33 and @xmath14 for the virial mass estimator where the uncertainty in @xmath25 was estimated from the difference of @xmath25 in the @xmath19 and @xmath20 profile .
both the king and virial masses are consistent with the masses and mass - to - light ratios indicated in figure 6 and with the masses derived from the dynamical models ( table 12 ) .
the exception is vucd7 for which the virial mass estimate is 1.8 times larger than the dynamical model mass .
this may be due to the inability of the simple virial estimator to correctly model the prominent core structure of this object , but the difference is not significant given the large uncertainty in the virial mass estimate for this object . in table 14
we compare the dynamical @xmath75 estimates for virgo ucds with the predicted stellar mass - to - light ratios of simple stellar population ( ssp ) models by maraston ( 2005 ) . to obtain the ssp model values
, we used ucd ages and metallicities derived in the following section ( figure 9 ) .
the uncertainties of the @xmath75 values are based on the age and metallicity ranges .
the dynamical @xmath75 values are consistent with the ssp model predictions within the uncertainties for both salpeter and kroupa imf .
it implies that virgo ucds do not require dark matter to explain their mass - to - light ratios .
this conclusion applies to the central region where we have velocity dispersion data covered by our spectroscopic observations .
an increasing dark matter contribution towards larger radii can not be ruled out with the present data .
the mass - to - light ratios of fornax ucds are discussed in hilker et al .
. the mass - to - light ratio of strom417 ( @xmath109 ) is larger than the @xmath75 value predicted by the ssp models with kroupa imf , but is in agreement ( within the uncertainties ) with the predictions from the models with salpeter imf .
this mass - to - light ratio is consistent with the high value reported for this object by haegan et al .
( 2005 ) , but our result is based on their king model fit to this object . for this reason our measurement is not an independent confirmation of their result
in this section we estimate the ages , metallicities , and abundances of our objects using the lick / ids index analysis .
lick / ids absorption - line indices were measured as defined by worthey et al .
( 1994 ) in the wavelength region @xmath110 .
we could not use the spectral region @xmath111 due to low s / n .
the line indices h@xmath36 , mgb , fe5270 , and fe5335 together with @xmath112 uncertainties are listed in table 15 . before measuring the indices we shifted the spectra to the rest frame wavelengths . since our spectra have much higher resolution ( @xmath113 in the blue wavelength range ) than the lick / ids system ( @xmath114 at the h@xmath36 - fe5335 wavelength range ) , we smoothed our spectra to the resolution of the lick data with a gaussian of dispersion @xmath115 .
the estimated size of the smoothing kernel is @xmath116 17 esi pixels ( gaussian sigma ) . after broadening ,
the lick indices for our objects and for the lick / ids standards ( 9 stars ) were measured as described in worthey et al .
the uncertainties on the indices were calculated according to cardiel et al .
( 1998 ) based on the noise spectrum of each galaxy / gc . to check the agreement between our instrumental system and lick / ids system , we calculated the difference between measured and published ( worthey et al .
1994 ) indices for all observations of the 9 calibration stars .
figure 7 shows our standard star measurements versus the published values ( worthey et al .
the mean offsets between our instrumental system and the lick / ids system are listed in table 16 .
the index measurements were corrected for the offsets .
the lick index measurements for the ngc4486b galaxy were also corrected for the effects of internal velocity dispersion as described in davies et al .
ucds , gcs and de , ns have small internal velocity dispersions compared to the lick / ids broadening function .
there is no need to apply velocity dispersion correction for these objects .
we have not corrected the de , n ( vcc1407 ) spectrum for any halo contribution because the nuclear light dominates in the central 15 of our extraction ; this object is already on the old envelope and its indices agree well with the other virgo dwarf elliptical galaxies measured by geha et al .
( 2003 ) .
the n4486b galaxy does not have a distinct halo , like the des do and the problem is the opposite of the de problem . when subtracting `` sky '' from along the short slit like esi ( @xmath117 ) , one really subtracts a component of galaxy light from farther out , so then one ends up subtracting too much stellar light , which then alters the actual physical extent of what one samples .
however , since ellipticals ( especially the likes of n4486b ) are very peaked in the center and have very modest ( if any ) line strength gradients , the error in the measured indices is very small . to translate measured line indices into age and metallicity estimates , we used the ssp models of thomas et al .
these models predict lick indices for a wide range of ages ( 1 - 15 gyr ) and metallicities ( [ z / h ] = -2.25,-1.35,-0.33,0.0,+0.35,+0.67 dex ) , and are tabulated for several different abundance ratios ( [ @xmath1/fe ] = -0.3,0.0,+0.3,+0.5 ) . to estimate [ @xmath1/fe ] for the ucds we plot mgb ( an indicator of @xmath1elements ) versus @xmath118fe@xmath42 ( an average of the indices fe5270 and fe5335 ) in figure 8 , overlaid with isochrones and isometallicity lines from thomas et al . ( 2003 ) . five ucds and strom417 ( a gc ) have super - solar abundance ratio , [ @xmath1/fe ] @xmath9 + 0.3 + 0.5 , and one ucd appears to have solar abundances , [ @xmath1/fe ] @xmath9 0.0 .
the super - solar abundances ( [ @xmath1/fe ] @xmath9 + 0.3 ) are typical of old stellar populations like globular clusters and elliptical galaxies .
[ @xmath1/fe ] traces the timescale of star formation activity in galaxies .
the majority of @xmath1-elements is produced rapidly by type ii supernovae , while fe is produced by type ia sne on longer timescales .
supersolar [ @xmath1/fe ] indicates rapid enrichment from type ii supernovae and implies that the galaxy / gc has undergone a short burst of star formation activity .
the solar and subsolar abundance ratios indicate slower chemical enrichment or a more quiescent star formation history ( van zee , barton & skillman 2004 ) . the nuclei of nucleated dwarf ellipticals , taken from geha et al .
( 2003 ) , are also shown in the same plot . the majority of the de nuclei data are consistent with solar [ @xmath1/fe ] abundance ratios , while the majority of ucds have super - solar [ @xmath1/fe ] abundances .
this provides evidence that the virgo ucds and typical de , n nuclei are different in that they have different formation histories . our de , n ( vcc1407 ) lies , however , along the ucd relation together with two de , ns from geha et al .
( 2003 ) . in figure 9
we show the age - sensitive h@xmath36 index versus the metallicity sensitive [ mgfe] index ( [ mgfe]=@xmath119 ) and compare them with the ssp models of thomas et al .
[ mgfe] is largely independent of [ @xmath1/fe ] and serves best as a tracer of total metalicity ( thomas et al .
2003 ) ; and h@xmath36 is less [ @xmath1/fe]-sensitive than other lick balmer line indices ( thomas et al .
2004 ) . as we can see from the plot ,
_ the virgo ucds are old ( older than 8 gyr ) and have metallicities between [ z / h ] = -1.35 and + 0.35 dex . _
the ssp models in figure 9 are shown for the abundance ratio [ @xmath1/fe ] = + 0.3 .
the conclusion about ucd ages and metallicities remains the same if we use [ @xmath1/fe ] = 0.0 and + 0.5 models . as a consistency test
, we used the virgo ucd ages and metallicities derived from figure 9 and maraston ( 2005 ) ssp models to predict photometric colors and to compare them with the observed ones .
the results are summarized in table 14 .
the observed colors of virgo ucds are in very good agreement with the colors predicted from the derived ages and metallicities .
the ages , metallicities and abundances of virgo ucds are similar to those found for gcs in the galaxies m49 and m87 in the virgo cluster by cohen et al .
( 2003 , 1998 ) .
according to cohen et al .
( 2003 ) , the m49 gcs have metallicities in the range from [ z / h ] = -1.3 to + 0.5 dex and in mean are older than 10 gyr .
the metallicity and age parameters for m49 and m87 gcs are basically identical .
the gc systems of both of these galaxies are @xmath1-enhanced by a factor of about 2 above the solar value .
we also find that the virgo ucds have older integrated stellar populations on average than the present - day de , n nuclei .
however we note that ucds are not distinct from the oldest de , n nuclei of geha et al .
( 2003 ) and our de , n , vcc1407 , lies with the ucds .
the general trend of ucds to lower metallicities and older ages than the de nuclei in figure 9 is not consistent with the naive threshing model in which ucds are identical to the present - day nuclei of de galaxies .
these results may , however , be consistent with variations of the threshing hypothesis in which the parent objects are disrupted at an early time when star formation is still going on and gas is present ( e.g. miske et al .
2006 ) . in this scenario ,
the stripping selectively halts the star formation in the stripped objects ( ucds ) giving them the lower metallicities and older ages compared to the nuclei which continue to form stars
. however this may not be consistent with the [ @xmath1/fe ] abundances found in virgo ucds : their super - solar abundances imply rapid enrichment in a short burst of star formation .
this seems to be inconsistent with gas stripping over an extended period , unless the stripping process caused a very sudden halt to the star formation .
we also measured the near - infrared ca ii triplet ( cat ) index for virgo ucds ( as defined in cenarro et al .
2001 ) to compare it with the metallicity derived from the lick indices above .
ssp models predict a strong dependence of the cat index on metallicity for sub - solar metallicities ( vazdekis et al .
2003 , maraston 2005 ) .
globular clusters are found to follow the model predictions very well for the metallicities typical of galactic gcs up to about a solar metallicity ( saglia et al .
2002 , maraston 2005 ) , whereas normal and dwarf elliptical galaxies deviate from ssp model predictions ( saglia et al .
2002 , michielsen et al .
michielsen et al .
( 2003 ) obtained cat values for a sample of des and found that four of five des with independent metallicity estimates have cat @xmath120 , which is much higher than expected from their low metallicities ( -1.5@xmath118[z / h]@xmath118 - 0.5 ) .
saglia et al .
( 2002 ) found that the cat values for bright ellipticals ( @xmath121 ) are lower than predicted by ssp models for their ages and metallicities ( 0.0@xmath118[z / h]@xmath118 + 0.7 ) .
the cat index values for our virgo objects are listed in table 17 .
the uncertainties were calculated based on the noise spectrum of each object .
the resolution of the cenarro et al .
stellar library is very close to our spectral resolution , so no correction for resolution was needed .
the measured indices for stars in common with the cenarro et al .
library showed good agreement between our instrumental system and cenarro et al .
the cat index for the ngc4486b galaxy was corrected for the effects of internal velocity dispersion using a k - type stellar template ( see cenarro et al .
2001 ) .
figure 10 presents [ z / h ] metallicity vs. cat index for the virgo ucds , strom417 ( a gc ) , vcc1407 ( a de , n ) and the ngc4486b galaxy . the metallicities were derived from the lick indices ( figure 9 ) . in figure 10
we also plot maraston ( 2005 ) ssp model predictions with a salpeter imf for ages 4 , 9 and 15 gyr .
it is hard to make any strong conclusions from this figure due to the large uncertainties in the data , but it appears that for sub - solar metallicities the virgo ucds and strom417 follow the ssp model predictions ( within the uncertainties ) .
vucd3 deviates strongly from the model predictions .
it has super - solar metallicity and lies in the plot with ngc4486b and other es from saglia et al .
there are no cat data for gcs at these metallicities available in the literature yet , so we can not conclude if vucd3 is globular - like or not .
our de , n ( vcc1407 ) lies with the ucds in the [ z / h ] vs. cat plot .
the cat value for vcc1407 is consistent with the ssp model predictions ( within the uncertainties ) .
however , as we already mentioned above , one of the five des studied by michielsen et al .
( 2003 ) also have cat in agreement with the model predictions for its metallicity , but all the rest do not .
given that globular clusters are known to follow ssp model predictions ( for sub - solar metallicities ) , whereas de and e galaxies do not ( saglia et al .
2002 , michielsen et al .
2003 ) , then figure 10 suggests that the virgo ucds have cat indices more like globular clusters than galaxies .
in this paper we have presented new imaging and spectroscopic observations of six virgo cluster ucds ( discovery reported by jones et al .
2006 ) , along with re - analysed data for five fornax cluster ucds ( initially presented by drinkwater et al .
these are the most luminous ucds : @xmath0 .
the main results of our analysis of these data are as follows . 1 . from the hst imaging we find that most of the ucds have shallow or steep cusps in their cores ; only one ucd has a flat `` king '' core .
we also find that none of the ucds show tidal cutoffs down to our limiting surface brightness .
these properties are not consistent with the standard king models with flat cores and tidal cutoffs used for most globular clusters .
however , recent work has shown that gcs can have such parameters .
noyola & gebhart ( 2003 ) obtained inner logarithmic slopes of profiles for 28 galactic gcs and found that the slopes span a continious range from zero to 0.6 featuring central cusps as well as flat king cores .
it is known that young gcs can have extended halos ( e.g. elson et al .
1987 ) , but mclaughlin & van der marel ( 2005 ) have now shown that extended halos are a generic characteristic of massive gcs both young and old in the magellanic clouds .
fundamental plane projections reveal a ) that virgo ucds have properties similar to fornax ucds and b ) that ucds and transition objects of haegan et al .
appear to follow the same relation between luminosity and velocity dispersion as old globular clusters .
+ in the @xmath66 - @xmath68 plane the ucds lie on the same tight correlation between mass and mass - to - light ratio as the bright gcs and transition objects of haegan et al . , but the fainter gcs ( @xmath69 ) show little if any correlation in this plane .
this corresponds to a mass of @xmath122 at which haegan et al .
( 2005 ) find a turn - over in scaling relations for low - mass systems in other projections of the fundamental plane .
+ in the @xmath66 - @xmath67 plane the ucds are not on the main gc relation as defined by the mw and m31 gcs , but the available data do not provide any evidence for a gap between ucds and gcs in this plane . + the de , n
nuclei in the virgo cluster with similar masses / luminosities to the ucds lie in the same region of all fundamental plane projections as the ucds .
this is consistent with the threshing hypothesis for ucd formation from early - type dwarf galaxies by the removal of low surface brightness envelope .
the age and metallicity analysis shows that virgo ucds are old ( older than 8 gyr ) and have metallicities ranging from [ z / h ] = -1.35 to + 0.35 dex .
+ the observed colors of virgo ucds are in agreement with the colors predicted from the derived ages and metallicities .
+ five ucds and strom417 , a gc , have super - solar abundance ratio , [ @xmath1/fe ] @xmath9 + 0.3 + 0.5 , and one ucd has solar abundance ratio , [ @xmath1/fe ] @xmath9 0.0 . the super - solar [ @xmath1/fe ]
abundances are typical of old stellar populations found in globular clusters and elliptical galaxies . + virgo ucds and typical present - day de , n nuclei are different in that they have different [ @xmath1/fe ] abundance ratios and , therefore , have different formation histories .
+ the ages , metallicities and abundances of virgo ucds are similar to those found for gcs in the galaxies m49 and m87 in the virgo cluster .
+ ucds generally have lower metallicities and older ages than de nuclei : this is not consistent with the naive threshing model in which ucds are identical to the present - day nuclei of de galaxies .
+ measurements of the near - ir cat index suggest that virgo ucds have stellar populations more like those found in globular clusters than in de and e galaxies .
the virgo and fornax ucds all have masses @xmath123 and mass - to - light ratios @xmath124 .
+ ucds are more massive than transition objects of haegan et al . and all known gcs for which dynamical mass estimates are available .
( meylan et al . 2001 ) , @xmath56 cen : @xmath125 ( meylan et al . 1995 ) , ngc5128 gcs : @xmath126 ( martini & ho 2004 ) , transition objects of haegan et al .
( 2005 ) : @xmath127 . ]
+ although the ucds are more massive than known globular clusters , they are within the theoretical limits for the most massive globular clusters formed in galaxies as large as m87 and ngc 1399 ( see formula 8 of kravtsov & gnedin 2005 ) according to wu & tremaine ( 2006 ) and the mass of ngc1399 within 50 kpc is @xmath128 as found by richtler et al .
( 2004 ) . ] .
recent simulations ( yahagi & bekki 2005 , bekki & yahagi 2006 ) confirm that globular clusters can escape the potential of these galaxies .
+ the ucd masses are close to the estimated masses of some young massive gcs ( ymgcs ) such as ngc7252:w3 , ngc7252:w30 and ngc1316:g114 ( maraston et al . 2004 , bastian et al . 2006 ) , whose origin is suggested to be by early mergers of lower mass stellar clusters ( kissler - patig et al .
as these objects evolve they will lose mass and their structural parameters will also change ( e.g. fellhauer & kroupa 2005 ) .
it is not clear that they will still have the same masses ( and other parameters ) as ucds after the very long evolution times demanded by our old age estimates for the virgo ucds . +
the dynamical mass - to - light ratios for virgo ucds are consistent with the simple stellar population model predictions ( by maraston 2005 ) within the uncertainties .
it implies that virgo ucds do not require dark matter to explain their mass - to - light ratios .
this conclusion applies to the central region where we have velocity dispersion data covered by our spectroscopic observations .
an increasing dark matter contribution towards larger radii can not be ruled out with the present data .
note that whilst the structural properties and internal velocity dispersions have been measured for all the ucds , the ages and metallicities ( and the interpretation of the mass - to - light ratios ) are limited to the virgo ucds for which we have this data .
the high resolution spectra for fornax ucds used by hilker et al .
( 2006 ) have too low s / n for the lick index measurements .
also , no lick standards were observed for the calibration onto the lick / ids system .
mieske et al .
( 2006 ) present [ fe / h ] metallicities and h@xmath36 indices for 26 compact objects in fornax , including ucd2 , ucd3 and ucd4 and four more objects with luminosities similar to our five fornax ucds .
the metallicities derived for virgo ucds are total metallicities [ z / h ] .
[ z / h ] and [ fe / h ] are related as follows : [ fe / h]=[z / h]-0.94@xmath129[@xmath1/fe ] ( thomas et al . 2003 ) .
assuming fornax and virgo ucds have similar mean @xmath1-abundances , [ @xmath1/fe]=+0.3 , the bright fornax compact objects in mieske et al .
( 2006 ) have mean metallicity [ z / h ] @xmath9 -0.34 , which is similar to the mean metallicity for virgo ucds .
the h@xmath36 indices in mieske et al .
( 2006 ) are not calibrated , so no reliable conclusions can be drawn , but the appearance at least is for the fornax objects to have higher h@xmath36 ( @xmath130 ) , hence younger mean ages compared to the virgo ucds .
this suggests a difference in formation time , if not mechanism , for ucds in the two galaxy clusters .
however , we prefer to refrain from any firm conclusions on fornax ucd origins until accurate age , metallicity and @xmath1-abundance estimates are obtained for them ; these should in turn be compared to the properties of _ fornax cluster _ globular clusters and dwarf galaxy nuclei . in the following discussion ,
our conclusions focus mainly on the virgo ucds .
the common feature in all the above results is that our detailed measurements of the internal ucd properties give values consistent with observed properties of globular clusters . in all the parameters we have investigated
there is no evidence for any gap between globular clusters and ucds .
the ages , metallicities and abundances of the virgo ucds are similar to those found for gcs in the two brightest virgo galaxies ( m49 and m87 ) .
this suggests that ucds and gcs could have the same formation epoch and the same star formation history .
theoretical work shows that such massive objects as ucds could form in m87 and ncg 1399 and subsequently escape the host galaxy potential ( kravtsov & gnedin 2005 ; yahagi & bekki 2005 , bekki & yahagi 2006 ) .
the surface brightness structure of the ucds is not different to that of globular clusters in the milky way and magellanic clouds .
the mass - to - light ratios of the virgo ucds are consistent with simple stellar populations as in globular clusters .
some of our results , notably the fundamental plane projections are consistent with the formation of ucds by the simple removal of the halo from the nuclei of nucleated dwarf galaxies
. however the ages , metallicities and abundances for virgo ucds are not consistent with this simple stripping model
. it might be consistent with more sophisticated models of the stripping process that include the effects of gas removal on the chemical evolution of the nuclei .
as we have shown that the virgo ucds are old , we note that definitive tests of theories of their formation by stripping processes or the evolution of merger - formed massive star clusters will need to consider the effects of gas processes ( especially gas removal ) over these long timescales .
eae and mjd acknowledge support from the australian research council .
mdg gratefully acknowledges support by the national science foundation under grant no .
some of the work reported here was done at the institute of geophysics and planetary physics , under the auspices of the u.s .
department of energy by lawrence livermore national laboratory under contract no . w-7405-eng-48 .
the authors particularly wish to thank chien peng ( stsci ) for very valuable assistance and discussions concerning the use of his galfit software .
we also wish to thank harry ferguson ( stsci ) and anton koekemoer ( stsci ) for assistance with the hst data analysis , holger baumgardt ( university of bonn ) for help with the dynamical calculations and richard white ( stsci ) for allowing us to use his esi reduction scripts , which we then modified for our specific purposes .
we are grateful to the referee for very helpful comments and suggestions which have improved this paper .
this work is based on data obtained at the w. m. keck observatory , which is operated as a scientific partnership among the california institute of technology , the university of california , and the national aeronautics and space administration .
the observatory was made possible by the generous financial support of the w. m. keck foundation .
barmby , p. , holland , s. , & huchra , j.p .
2002 , aj , 123 , 1937 bastian , n. , saglia , r.p . ,
goudfrooij , p. , kissler - patig , m. , maraston , c. , schweizer , f. , & zoccali , m. 2006 , a&a , 448 , 881 bekki , k. , couch , w.j . , & drinkwater , m.j .
2001 , apj , 522 , l105 bekki , k. , couch , w.j . ,
drinkwater , m.j . , & shioya , y. 2003 , mnras , 344 , 399 bekki , k. , & yahagi , h. 2006 , mnras , 372 , 1019 bender , r. , burstein , d. , & faber , s.m .
1992 , apj , 399 , 462 bruzual , g. , & charlot , s. 2003 , mnras , 344 , 1000 burstein , d. , bender , r. , faber , s.m . , & nolthenius , r. 1997 , aj , 114 , 1365 cardiel , n. , gorgas , j. , cenarro , a.j . ,
gonzlez , j.j .
1998 , a&as , 127 , 597 cenarro , a.j .
, cardiel , n. , gorgas , j. , peletier , r.f . , vazdekis , a. , & prada , f. 2001 , mnras , 326 , 959 choi , p.i . ,
guhathakurta , p. , johnston , k.v .
2002 , aj , 124 , 310 cohen , j.g . ,
blakeslee , j.p . , & ryzhov , a. 1998 , apj , 496 , 808 cohen , j.g . ,
blakeslee , j.p . , & ct , p. 2003
, apj , 592 , 866 cottrell , p.l .
, & sneden , c. 1986 , a&a , 161 , 314 davies , r.l . , sadler , e.m . , & peletier , r.f .
1993 , mnras , 262 , 650 djorgovski , s.g . , gal , r.r . ,
mccarthy , j.k .
, cohen , j.g . ,
de carvalho , r.r . ,
meylan , g. , bendinelly , o. , & parmeggiani , g. 1997 , apj , 474 , l19 drinkwater , m.j . ,
jones , j.b . , gregg , m.d . , & phillipps , s. 2000 , pasa , 17 , 227 drinkwater , m.j . ,
gregg , m.d .
, hilker , m. , bekki , k. , couch , w.j . ,
ferguson , h.c . ,
jones , j.b . , &
phillipps , s. 2003 , nature , 423 , 519 drinkwater , m.j . ,
gregg , m.d . ,
couch , w.j . ,
ferguson , h.c .
, hilker , m. , jones , j.b . ,
karick , a. , & phillipps , s. 2004 , pasa , 21 , 375 elson , r.a.w . ,
fall , s.m . , & freeman , k.c .
1987 , apj , 323 , 54 faber , s.m . , wegner , g. , burstein , d. , davies , r.l . , dressler , a. , lynden - bell , d. , & terlevich , r.j .
1989 , apjs , 69 , 763 fellhauer , m. , & kroupa , p. 2002 , mnras , 330 , 642 fellhauer , m. , & kroupa , p. 2005 , mnras , 359 , 223 fellhauer , m. & kroupa , p. 2006 ,
mnras , 367 , 1577 freedman , w.l . ,
madore , b.f . ,
gibson , b.k .
, et al . 2001 , apj , 553 , 47 geha , m. , guhathakurta , p. , & van der marel , r.p .
2002 , aj , 124 , 3073 geha , m. , guhathakurta , p. , & van der marel , r.p .
2003 , aj , 126 , 1794 gregg , m.d . , & minniti , d. 1997 , pasp , 109 , 1026 hanes , d.a . , ct , p. , bridges , t.j . , mclaughlin , d.e .
, geisler , d. , harris , g.l.h . ,
hesser , j.e . , & lee , m.g .
2001 , apj , 559 , 812 harris , w.e .
1996 , aj , 112 , 1487 harris , w.e . ,
harris , g.l.h . ,
holland , s.t . , & mclaughlin , d.e .
2002 , aj , 124 , 1435 haegan , m. , jordn , a. , ct , p. , et al .
2005 , apj , 627 , 203 hilker , m. , infante , l. , vieira , g. , kissler - patig , m. , & richtler , t. 1999 , a&as , 134 , 75 hilker , m. , baumgardt , h. , infante , l. , drinkwater , m. , evstigneeva , e. , gregg , m. 2006 , a&a , in press jones , j. b. , drinkwater , m. j. , jurek , r. ; phillipps , s. , gregg , m. d. , bekki , k. , couch , w. j. , karick , a. , parker , q. a. , & smith , r. m. 2006 , aj , 131 , 312 karick , a.m. , drinkwater , m.j . , &
gregg , m.d .
2003 , mnras , 344 , 188 king , i.r . 1966 , aj , 71 , 64 kissler - patig , m. , jordn , a. , & bastian , n. 2006 , a&a , 448 , 1031 kravtsov , a.v . , & gnedin , o.y .
2005 , apj , 623 , 650 lauer , t.r . ,
ajhar , e.a . ,
byun , y .-
i . , dressler , a. , faber , s.m .
, grillmair , c. , kormendy , j. , richstone , d. , & tremaine , s 1995 , aj , 110 , 2622 luck , r.e . , & challener , s.l .
1995 , aj , 110 , 2968 mackey , a.d . &
van den bergh , s. 2005 , mnras , 360 , 631 maraston , c. , bastian , n. , saglia , r.p .
, kissler - patig , m. , schweizer , f. , & goudfrooij , p. 2004 , a&a , 416 , 467 maraston , c. 2005 , mnras , 362 , 799 van der marel , r.p .
1994 , mnras , 270 , 271 martini , p. , & ho , l.c .
2004 , apj , 610 , 233 mclaughlin , d. e. , & van der marel , r. p. 2005 , apjs , 161 , 304 mcwilliam , a. 1990 , apjs , 74 , 1075 meylan , g. , mayor , m. , duquennoy , a. , & dubath , p. 1995
, a&a , 303 , 761 meylan , g. , sarajedini , a. , jablonka , p. , djorgovski , s.g . , bridges , t. , & rich , r.m . 2001 , aj , 122 , 830 michielsen , d. , de rijcke , s. , dejonghe , h. , zeilinger , w.w . , & hau , g.k.t .
2003 , apj , 597 , l21 mieske , s. , hilker , m. , & infante , l. 2002 , a&a , 383 , 823 mieske , s. , infante , l. , benitez , n. , et al .
2004 , aj , 128 , 1529 mieske , s. , hilker , m. , infante , l. , & jordn , a. 2006 , aj , 131 , 2442 mihalas , d. , & binney , j. 1981 , galactic astronomy , san francisco minniti , d. , kissler - patig , m. , goudfrooij , p. , & meylan , g. 1998 , aj , 115 , 121 noyola , e. , & gebhardt .
k. 2003 , revmexaa ( serie de conferencias ) , 18 , 78 peng , c.y .
, ho , l.c . , impey , c.d . , & rix , h .- w .
2002 , aj , 124 , 266 peterson , r.c .
1978 , apj , 224 , 595 peterson , r.c . , & king , i.r . 1975 , aj , 80 , 427 richtler , t. , dirsch , b. , gebhardt , k. et al .
2004 , aj , 127 , 2094 phillipps , s. , drinkwater , m. j. , gregg , m. d. , & jones , j. b. 2001 , apj , 560 , 201 saglia , r.p . ,
maraston , c. , thomas , d. , bender , r. , & colless , m. 2002 , apj , 597 , l13 schlegel , d.j . ,
finkbeiner , d.p .
, & davis , m. 1998 , apj , 500 , 525 sirianni , m. , jee , m.j . , benitez , n. , et al .
2005 , pasp , 117 , 1049 spitzer , l. 1987 , dynamical evolution of globular clusters ( princeton : princeton univ . press ) strom , s.e . , forte , j.c . ,
harris , w.e . ,
strom , k.e . , wells , d.c . , &
smith , m.g .
1981 , apj , 245 , 416 thomas , d. , maraston , c. , & bender , r. 2003 , mnras , 339 , 897 thomas , d. , maraston , c. , & korn , a. 2004 , mnras , 351 , l19 tonry , j. , & davis , m. 1979 , aj , 84 , 1511 trager , s.c . ,
faber , s.m . ,
worthey , g. , & gonzlez , j.j .
2000 , aj , 119 , 1645 queloz , d. , dubath , p. , & pasquini , l. 1995 , a&a , 300 , 31 valdes f. , gupta r. , rose j.a .
, singh h.p . , & bell d.j .
2004 , apjs , 152 , 251 vazdekis , a. , cenarro , a.j . , gorgas , j. , cardiel , n. , & peletier , r.f .
2003 , mnras , 340 , 1317 worthey , g. , faber , s.m . ,
gonzlez , j.j . , & burstein , d. 1994 , apjs , 94 , 687 wu , x. , & tremaine , s. 2006 , apj , 643 , 210 yahagi , h. , & bekki , k. 2005 , mnras , 364l , 86 van zee , l. , barton , e.j . , & skillman , e.d .
2004 , aj , 128 , 2797 ccccc strom417 & m87 gc & 12:31:01.29 & + 12:19:25.6 & @xmath131 + vucd1 & ucd & 12:30:07.61 & + 12:36:31.1 & @xmath131 + vucd3/strom547 & ucd / m87 gc & 12:30:57.40 & + 12:25:44.8 & @xmath131 + vucd4 & ucd & 12:31:04.51 & + 11:56:36.8 & @xmath131 + vucd5 & ucd & 12:31:11.90 & + 12:41:01.2 & @xmath131 + vucd6 & ucd & 12:31:28.41 & + 12:25:03.3 & @xmath131 + vucd7 & ucd & 12:31:52.93 & + 12:15:59.5 & @xmath132 + vcc1407 & de , n & 12:32:02.70 & + 11:53:25.0 & @xmath133 + ngc4486b & e & 12:30:31.92 & + 12:29:27.4 & @xmath134 rclcc 1 & hd040460 & k1iii & -0.50 & 2 + 2 & hd048433 & k1iii & -0.26 & 1 + 3 & hd137704 & k4iii & -0.43 & 1 + 4 & hd139195 & k0iii & -0.17 & 1 + 5 & hd139669 & k5iii & -0.13 & 5 + 6 & hd141680 & g8iii & -0.28 & 1 + 7 & hd142574 & m0iii & ... & ... + 8 & hd143107 & k2iii & -0.32 & 1 + 9 & hd145148 & k1.5iv & ... & ... + 10 & hd145675 & k0v & 0.31 & 3 + 11 & hd147677 & k0iii & -0.08 & 1 + 12 & hd148513 & k4iii & 0.04/-0.31 & 1/4 + 13 & hd149161 & k4iii & -0.23 & 1 cccccc vucd1 & 1227.8 @xmath135 1.7 & @xmath136 33.8 @xmath135 1.7 & 6 & 1225.4 @xmath135 3.7 & 34.0 @xmath135 1.6 + vucd3/strom547 & @xmath136 710.6 @xmath135 3.5 & @xmath136 37.7 @xmath135 1.4 & 2,8 & @xmath136 711.4 @xmath135 3.4 & 37.8 @xmath135 1.6 + vucd4 & @xmath136 919.7 @xmath135 1.7 & @xmath136 23.9 @xmath135 2.2 & 6 & @xmath136 916.5 @xmath135 4.2 & 21.3 @xmath135 2.2 + vucd5 & 1293.1 @xmath135 1.7 & @xmath136 27.4 @xmath135 1.7 & 6 & 1290.3 @xmath135 3.0 & 24.9 @xmath135 2.0 + vucd6 & 2105.3 @xmath135 1.7 & @xmath136 24.6 @xmath135 1.8 & 6 & 2101.7 @xmath135 3.9 & 22.2 @xmath135 2.2 + vucd7 & @xmath136 988.3 @xmath135 2.7 & @xmath136 36.7 @xmath135 3.7 & 6 & @xmath136 985.7 @xmath135 5.0 & 35.6 @xmath135 1.6 + strom417 & 1863.5 @xmath135 1.6 & @xmath136 26.4 @xmath135 2.1 & 6 & 1860.6 @xmath135 3.0 & 25.8
@xmath135 2.0 + vcc1407 & 1018.9 @xmath135 3.2 & @xmath136 29.3 @xmath135 2.5 & 1,2,3,4,6 & 1019.4 @xmath135 5.7 & 26.6 @xmath135 2.1 + ngc4486b & 1558.4 @xmath135 4.2 & 211.3 @xmath135 4.8 & 2,3,7,8,11,12 & @xmath136 1546.9 @xmath135 32.8 & 198.1 @xmath135 15.5 cccccc vucd1 & 1219.5 @xmath135 1.3 & @xmath136 34.4 @xmath135 1.6 & 2 & 1225.1 @xmath135 9.3 & 32.8 @xmath135 6.1 + vucd3/strom547 & @xmath136 716.4 @xmath135 1.2 & @xmath136 46.5 @xmath135 1.6 & 9 & @xmath137 717.4 @xmath135 12.7 & 51.4 @xmath135 4.7 + vucd4 & @xmath136 912.5 @xmath135 1.5 & @xmath136 23.9 @xmath135 1.8 & 2 & @xmath136 916.8 @xmath135 5.8 & 15.7 @xmath135 8.3 + vucd5 & 1285.8 @xmath135 1.0 & @xmath136 28.9 @xmath135 1.5 & 2 & 1291.8 @xmath135 6.3 & 26.0 @xmath135 6.8 + vucd6 & 2096.7 @xmath135 1.4 & @xmath136 25.7 @xmath135 1.8 & 2 & 2103.1 @xmath135 6.8 & 20.4 @xmath135 7.6 + vucd7 & @xmath136 980.7 @xmath135 3.5 & @xmath136 39.2 @xmath135 4.4 & 2 & @xmath136 986.6 @xmath135 7.7 & 33.3 @xmath135 6.0 + strom417 & 1857.1 @xmath135 1.4 & @xmath136 28.8 @xmath135 1.6 & 2 & 1863.4 @xmath135 6.7 & 25.1 @xmath135 6.9 + vcc1407 & 1018.8 @xmath135 3.3 & @xmath136 31.4 @xmath135 2.7 & 1,2,6,11 & 1020.2 @xmath135 7.5 & 28.5 @xmath135 6.5 + ngc4486b & 1556.3 @xmath135 8.7 & 199.1 @xmath135 5.0 & 2,6,8,9 & @xmath136 1558.7 @xmath135 52.1 & 230.4 @xmath135 26.3 ccc strom417 & 1860.3 @xmath135 1.5 & @xmath136 27.6 @xmath135 1.9 + vucd1 & 1223.7 @xmath135 1.5 & @xmath136 34.1 @xmath135 1.7 + vucd3/strom547 & @xmath136 713.5 @xmath135 2.4 & @xmath136 42.1 @xmath135 1.5 + vucd4 & @xmath136 916.1 @xmath135 1.6 & @xmath136 23.9 @xmath135 2.0 + vucd5 & 1289.5 @xmath135 1.4 & @xmath136 28.2 @xmath135 1.6 + vucd6 & 2101.0 @xmath135 1.6 & @xmath136 25.2 @xmath135 1.8 + vucd7 & @xmath136 984.5 @xmath135 3.1 & @xmath136 38.0 @xmath135 4.1 + vcc1407 & 1018.9 @xmath135 3.3 & @xmath136 30.4 @xmath135 2.6 + ngc4486b & 1557.4 @xmath135 6.5 & 205.2 @xmath135 4.9 cccc vucd1 & 18.66 & -12.26 & 0.96 + vucd3/strom547 & 18.34 & -12.58 & 1.27 + vucd4 & 18.62 & -12.30 & 0.99 + vucd5 & 18.60 & -12.32 & 1.11 + vucd6 & 18.82 & -12.10 & 1.02 + vucd7 & 17.48 & -13.44 & 1.13 lcccc ucd1 & 3:37:03.30 & -35:38:04.6 & 19.20 & -12.19 + ucd2 & 3:38:06.33 & -35:28:58.8 & 19.12 & -12.27 + ucd3 & 3:38:54.10 & -35:33:33.6 & 17.82 & -13.57 + ucd4 & 3:39:35.95 & -35:28:24.5 & 18.94 & -12.45 + ucd5 & 3:39:52.58 & -35:04:24.1 & 19.40 & -11.99 + fcc303 & 3:45:14.08 & -36:56:12.4 & 15.90 & -15.49 lccccccc @xmath50 & @xmath136 15.4 & @xmath136 28.1 & @xmath136 29.4 & @xmath136 22.5 & @xmath136 22.1 & ... & ... + + * nuker * & & & & & & & + @xmath40 ( f606w ) & @xmath136 0.39 & @xmath136 0.41 & @xmath136 0.35 & @xmath136 0.38 & @xmath136 0.24 & ... & ... + @xmath40 ( f814w ) & @xmath136 0.38 & @xmath136 0.43 & @xmath136 0.34 & @xmath136 0.36 & @xmath136 0.27 & ... & ... + @xmath43 & @xmath136 13.3 & ... & ... & @xmath136 19.2 & ... & ... & ...
+ @xmath138 & 18.49 & ... & ... & 18.55 & ... & ... & ... + @xmath38 & @xmath137 6.3 & @xmath136 73.3 & @xmath137 9.1 & @xmath136 19.3 & @xmath137 5.2 & ... & ...
+ @xmath44 & 16.42 & 22.00 & 17.24 & 18.73 & 16.45 & ... & ... + @xmath1 & @xmath136 2.52 & @xmath136 1.95 & @xmath136 3.92 & @xmath136 0.99 & 19.48 & ... & ... + @xmath36 & @xmath136 2.69 & @xmath136 3.17 & @xmath136 2.34 & @xmath136 4.02 & @xmath136 2.24 & ... & ... + @xmath37 & @xmath136 0.23 & @xmath136 1.53 & @xmath136 0.33 & @xmath136 0.00 & @xmath136 0.50 & ... & ... + @xmath46 & @xmath136 0.06 & @xmath136 0.15 & @xmath136 0.15 & @xmath136 0.01 & @xmath136 0.04 & ... & ... + * sersic * & & & & & & & + @xmath40 ( f606w ) & @xmath136 0.48 & @xmath136 0.42 & @xmath136 0.40 & @xmath136 0.39 & @xmath136 0.29 & @xmath136 0.37 & @xmath136 0.37/0.37 + @xmath40 ( f814w ) & @xmath136 0.44 & @xmath136 0.45 & @xmath136 0.38 & @xmath136 0.38 & @xmath136 0.31 & @xmath136 0.37 & @xmath136 0.36/0.37 + @xmath43 & @xmath136 11.1 & @xmath136 64.3 & @xmath136 19.3 & @xmath136 18.1 & @xmath136 12.9 & @xmath137 9.2 & @xmath136 214.1/223.4 + @xmath138 & 18.75 & 17.69 & 18.85 & 18.64 & 19.05 & 18.68 & @xmath136
18.01/17.85 + n & @xmath137 2.2 & @xmath136 10.9 & @xmath137 2.1 & @xmath137 1.9 & @xmath137 3.1 & @xmath137 2.2 & @xmath136 1.4/2.1 + @xmath46 & @xmath136 0.07 & @xmath136 0.16 & @xmath136 0.16 & @xmath136 0.01 & @xmath136 0.04 & @xmath136 0.12 & @xmath136 0.05/0.05 + * king , @xmath29 * & & & & & & & + @xmath40 ( f606w ) & @xmath136 0.45 & @xmath136 0.55 & @xmath136 0.37 & @xmath136 0.37 & @xmath136 0.27 & @xmath136 0.37 & ...
+ @xmath40 ( f814w ) & @xmath136 0.42 & ... & @xmath136 0.36 & @xmath136 0.37 & @xmath136 0.29 & @xmath136 0.36 & ... + @xmath43 & @xmath136 11.2 & ... & @xmath136 21.8 & @xmath136 17.8 & @xmath136 15.2 & @xmath136 10.4 & ...
+ @xmath138 & 18.66 & ... & 18.54 & 18.59 & 18.93 & 18.42 & ...
+ @xmath25 & @xmath137 3.6 & @xmath137 1.9 & @xmath137 5.8 & @xmath137 6.6 & @xmath137 2.7 & @xmath137 3.1 & ...
+ @xmath26 & 124.0 & @xmath137 @xmath139 & 302.7 & 172.5 & 355.6 & 130.2 & ... + c & @xmath136 1.54 & ... & @xmath136 1.71 & @xmath136 1.42 & @xmath136 2.12 & @xmath136 1.62 & ...
+ @xmath49 & 14.91 & 13.79 & 15.96 & 15.98 & 14.98 & 14.38 & ... + @xmath46 & @xmath136 0.06 & @xmath136 0.17 & @xmath136 0.15 & @xmath136 0.01 & @xmath136 0.04 & @xmath136 0.11 & ... + & & & & & & + @xmath40 ( f606w ) & @xmath136 0.41 & @xmath136 0.53 & @xmath136 0.36 & @xmath136 0.37 & @xmath136 0.25 & ... & ... + @xmath40 ( f814w ) & @xmath136 0.40 & ... & @xmath136 0.35 & @xmath136 0.37 & @xmath136 0.28 & ... & ... + @xmath1 & @xmath136 3.74 & @xmath136 0.63 & @xmath136 3.95 & @xmath136 2.24 & @xmath136 4.45 & ... & ... + @xmath43 & @xmath136 11.3 & @xmath136 18.7 & @xmath136 22.0 & @xmath136 17.9 & @xmath136 14.8 & ... & ...
+ @xmath138 & 18.63 & 18.18 & 18.52 & 18.60 & 18.84 & ... & ... + @xmath25 & @xmath137 4.3 & @xmath137 1.8 & @xmath137 6.7 & @xmath137 6.7 & @xmath137 3.2 & ... & ...
+ @xmath26 & 360.0 & 247.5 & 1217.4 & 200.5 & 2352.5 & ... & ... + c & @xmath136 1.92 & @xmath136 2.14 & @xmath136 2.26 & @xmath136 1.48 & @xmath136 2.87 & ... & ...
+ @xmath49 & 14.68 & 13.78 & 15.76 & 15.97 & 14.81 & ... & ... + @xmath46 & @xmath136 0.06 & @xmath136 0.17 & @xmath136 0.15 & @xmath136 0.01 & @xmath136 0.04 & ... & ... + + best model & @xmath136 n & @xmath136 n & @xmath136 n & @xmath136 n & @xmath136 n & @xmath136 k & @xmath136 s lcccccccccc @xmath50 & @xmath136 33.2 & @xmath136 29.5 & @xmath136 80.7 & ... & ... & @xmath136 31.3 & ... & ... & ... & ... + + * nuker * & & & & & & & & & & + @xmath40 & @xmath136 0.61 & @xmath136 0.64 & @xmath136 0.51 & ... & ... & @xmath136 0.50 & ... & ... & ... & ...
+ @xmath43 & ... & ... & ... & ... & ... & ... & ... & ... & ... & ...
+ @xmath138 & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... + @xmath38 & @xmath136 36.7 & @xmath137 4.3 & 318.7 & ... & ... & @xmath137 7.3 & ... & ... & ... & ...
+ @xmath44 & 20.67 & 16.53 & 24.38 & ... & ... & 16.92 & ... & ... & ... & ... + @xmath1 & @xmath136 0.34 & @xmath136 0.73 & @xmath136 1.90 & ... & ... & 20.00 & ... & ... & ... & ... + @xmath36 & @xmath136 3.03 & @xmath136 2.38 & @xmath136 7.04 & ... & ... & @xmath136 2.13 & ... & ... & ... & ... + @xmath37 & @xmath136 0.91 & @xmath136 0.58 & @xmath136 1.10 & ... & ... & @xmath136 0.88 & ... & ... & ... & ... + @xmath46 & @xmath136 0.18 & @xmath136 0.01 & @xmath136 0.03 & ... & ... & @xmath136 0.05 & ... & ... & ... & ... + * sersic * & & & & & & & & & & + @xmath40 & @xmath136 0.60 & @xmath136 0.66 & ... & @xmath136 0.47 & @xmath136 0.47/0.47 & @xmath136 0.69 & @xmath136 0.47 & @xmath136 0.47/0.47 & @xmath136 1.03 & @xmath136 1.03/1.03 + @xmath43 & @xmath136 36.9 & @xmath136 26.6 & ... & @xmath137 8.6 & @xmath136 106.6/103.2 & @xmath136 24.1 & @xmath137 6.0 & @xmath136 134.5/135.5 & @xmath136 25.2 & @xmath136 696.3/692.8 + @xmath138 & 19.01 & 19.03 & ... & 20.29 & @xmath136 17.98/17.94 & 18.97 & 20.45 & @xmath136 19.54/19.58 & 18.76 & @xmath136 15.95/15.95 + n & @xmath137 9.9 & @xmath137 6.8 & ... & @xmath137 1.7 & @xmath136 1.3/1.5 & @xmath137 5.5 & @xmath137 1.1 & @xmath136 6.9/6.3 & @xmath136 10.7 & @xmath136 0.6/0.6 + @xmath46 & @xmath136 0.19 & @xmath136 0.01 & ... & @xmath136 0.02 & @xmath136 0.03/0.03 & @xmath136 0.05 & @xmath136 0.24 & @xmath136 0.16/0.15 & @xmath136 0.03 & @xmath136 0.10/0.10 + * king , @xmath29 * & & & & & & & & & & + @xmath40 & @xmath136 0.67 & @xmath136 0.63 & ... & @xmath136 0.47 & ... & @xmath136 0.52 & @xmath136 0.47 & ... & @xmath136 1.03 & ... + @xmath43 & @xmath136 48.9 & @xmath136 28.3 & ... & @xmath136 10.9 & ... & @xmath136 26.0 & @xmath137 6.0 & ... & @xmath136 25.9 & ...
+ @xmath138 & 18.71 & 18.98 & ... & 20.00 & ... & 18.88 & 20.19 & ... & 18.78 & ...
+ @xmath25 & @xmath137 1.8 & @xmath137 2.3 & ... & @xmath137 3.6 & ... & @xmath137 2.8 & @xmath137 4.0 & ... & @xmath137 1.2 & ...
+ @xmath26 & 5761.2 & 1457.5 & ... & 119.2 & ... & 987.6 & @xmath136 30.7 & ... & 2403.9 & ... + c & @xmath136 3.51 & @xmath136 2.80 & ... & @xmath136 1.52 & ... & @xmath136 2.55 & @xmath136 0.89 & ... & @xmath136 3.30 & ...
+ @xmath49 & 14.21 & 14.70 & ... & 15.74 & ... & 14.89 & 15.13 & ... & 13.32 & ... + @xmath46 & @xmath136 0.18 & @xmath136 0.01 & ... & @xmath136 0.03 & ... & @xmath136 0.05 & @xmath136 0.24 & ... & @xmath136 0.03 & ... + & & & & & & & & & + @xmath40 & @xmath136 0.64 & @xmath136 0.62 & ... & ... & ... & @xmath136 0.51 & ... & ... & ... & ... + @xmath1 & @xmath136 0.74 & @xmath136 1.23 & ... & ... & ... & @xmath136 3.32 & ... & ... & ... & ... + @xmath43 & @xmath136 22.4 & @xmath136 23.1 & ... & ... & ... & @xmath136 29.5 & ... & ... & ... & ...
+ @xmath138 & 19.00 & 19.10 & ... & ... & ... & 18.82 & ... & ... & ... & ... + @xmath25 & @xmath137 1.8 & @xmath137 2.2 & ... & ... & ... & @xmath137 3.0 & ... & ... & ... & ...
+ @xmath26 & 378.0 & 487.1 & ... & ... & ... & 4501.2 & ... & ... & ... & ... +
c & @xmath136 2.32 & @xmath136 2.35 & ... & ... & ... & @xmath136 3.18 & ... & ... & ... & ...
+ @xmath49 & 14.21 & 14.67 & ... & ... & ... & 14.94 & ... & ... & ... & ... + @xmath46 & @xmath136 0.18 & @xmath136 0.01 & ... & ... & ... & @xmath136 0.05 & ... & ... & ... & ... + + best model & @xmath136 s & gen .
k & k+s / s+s & @xmath136 k & @xmath136 s & @xmath136 n & @xmath136 k / s & @xmath136 s & @xmath136 k / s & @xmath136 s lcccc & & & & + vucd1 & @xmath136 11.3 & 18.66 & 16.58 & 0.06 + vucd3 & @xmath136 18.7 & 18.34 & 17.34 & 0.15 + vucd4 & @xmath136 22.0 & 18.62 & 17.98 & 0.15 + vucd5 & @xmath136 17.9 & 18.60 & 17.51 & 0.01 + vucd6 & @xmath136 14.8 & 18.82 & 17.32 & 0.04 + vucd7 & @xmath136 96.8 & 17.48 & 20.06 & ... + vucd7 core & @xmath136 10.4 & 18.42 & 16.15 & 0.11 + vucd7 halo & 214.1 & 18.01 & 22.31 & 0.05 + + * fornax : * & & & & + ucd1 & @xmath136 22.4 & 19.20 & 18.13 & 0.19 + ucd2 & @xmath136 23.1 & 19.12 & 18.12 & 0.01 + ucd3 & @xmath136 89.7 & 17.82 & 19.76 & ...
+ ucd3 core & @xmath136 10.9 & 20.00 & 17.36 & 0.03 + ucd3 halo & 106.6 & 17.98 & 20.30 & 0.03 + ucd4 & @xmath136 29.5 & 18.94 & 18.47 & 0.05 + ucd5 & @xmath136 31.2 & 19.40 & 19.05 & ...
+ ucd5 core & @xmath137 6.0 & 20.19 & 16.25 & 0.24 + ucd5 halo & 134.5 & 19.54 & 22.37 & 0.16 + fcc303 & 660.0 & 15.90 & 22.18 & ... + fcc303 core & @xmath136 25.9 & 18.78 & 18.03 & 0.03 + fcc303 halo & 696.3 & 15.95 & 22.35 & 0.10 lcccc & & & + vucd1 & @xmath140 & @xmath141 & @xmath142 & @xmath143 + vucd3 & @xmath144 & @xmath145 & @xmath146 & @xmath147 + vucd4 & @xmath148 & @xmath149 & @xmath150 & @xmath151 + vucd5 & @xmath152 & @xmath153 & @xmath154 & @xmath155 + vucd6 & @xmath156 & @xmath157 & @xmath158 & @xmath159 + & & & + vucd1 & @xmath160 & @xmath161 & @xmath162 & @xmath163 + vucd3 & @xmath164 & @xmath165 & @xmath166 & @xmath167 + vucd4 & @xmath168 & @xmath169 & @xmath170 & @xmath171 + vucd5 & @xmath172 & @xmath173 & @xmath174 & @xmath175 + vucd6 & @xmath176 & @xmath177 & @xmath178 & @xmath179 + & & & + vucd1 & @xmath180 & @xmath181 & @xmath182 & @xmath183 + vucd4 & @xmath184 & @xmath185 & @xmath186 & @xmath187 + vucd5 & @xmath188 & @xmath189 & @xmath174 & @xmath175 + vucd6 & @xmath190 & @xmath191 & @xmath192 & @xmath193 + strom417 & @xmath194 & @xmath195 & @xmath196 & @xmath109 + & & & + vucd1 & @xmath197 & @xmath198 & @xmath199 & @xmath200 + vucd3 & @xmath201 & @xmath202 & @xmath203 & @xmath204 + vucd4 & @xmath205 & @xmath206 & @xmath186 & @xmath207 + vucd5 & @xmath208 & @xmath209 & @xmath174 & @xmath175 + vucd6 & @xmath210 & @xmath211 & @xmath212 & @xmath213 + & & & + vucd7 & @xmath214 & @xmath215 & @xmath216 & @xmath217 lcccc vucd1 & @xmath197 & @xmath198 & @xmath199 & @xmath200 + vucd3 & @xmath201 & @xmath202 & @xmath203 & @xmath204 + vucd4 & @xmath205 & @xmath206 & @xmath186 & @xmath207 + vucd5 & @xmath208 & @xmath209 & @xmath174 & @xmath175 + vucd6 & @xmath210 & @xmath211 & @xmath212 & @xmath213 + vucd7 & @xmath214 & @xmath215 & @xmath216 & @xmath217 + strom417 & @xmath194 & @xmath195 & @xmath196 & @xmath109 lcccc vucd1 & @xmath218 & @xmath219 & @xmath136 @xmath182 & @xmath220 + vucd3 & ... & ... & @xmath136 @xmath221 & @xmath222 + vucd4 & @xmath223 & @xmath224 & @xmath136 @xmath225 & @xmath226 + vucd5 & @xmath227 & @xmath228 & @xmath136 @xmath229 & @xmath163 + vucd6 & @xmath230 & @xmath224 & @xmath136 @xmath231 & @xmath232 + vucd7 & ... & ... & @xmath233 & @xmath234 + strom417 & @xmath235 & @xmath236 & @xmath136 @xmath237 & @xmath238 lcccccccc vucd1 & @xmath136 8 ... 15 & @xmath136 @xmath239 ...
@xmath240 & 0.96 & 0.91 ... 1.16 & 0.89 ... 1.13 & @xmath200 & @xmath241 & @xmath242 + vucd3 & 12 ... 15 & @xmath136 @xmath243 ...
@xmath244 & 1.27 & 1.19 ... 1.30 & 1.16 ... 1.27 & @xmath204 & @xmath245 & @xmath246 + vucd4 & @xmath136 8 ... 15 & @xmath136 @xmath239 ...
@xmath240 & 0.99 & 0.91 ... 1.16 & 0.89 ... 1.13 & @xmath207 & @xmath241 & @xmath242 + vucd5 & @xmath136 8 ... 15 & @xmath240 ...
@xmath243 & 1.11 & 1.07 ... 1.22 & 1.05 ... 1.20 & @xmath175 & @xmath247 & @xmath248 + vucd6 & @xmath136 8 ... 15 & @xmath136 @xmath239 ...
@xmath240 & 1.02 & 0.91 ... 1.16 & 0.89 ... 1.13 & @xmath213 & @xmath241 & @xmath242 + vucd7 & @xmath136 8 ... 15 & @xmath136 @xmath239 ...
@xmath240 & 1.13 & 0.91 ... 1.16 & 0.89 ... 1.13 & @xmath217 & @xmath241 & @xmath242 + strom417 & @xmath136 4 ... 12 & @xmath239 ...
@xmath243 & ... & 0.82 ... 1.19 & 0.81 ... 1.16 & @xmath109 & @xmath249 & @xmath250 lcccccc vucd1 & 2.06 @xmath135 0.14 & 2.20 @xmath135 0.13 & 1.56 @xmath135 0.15 & 1.19 @xmath135 0.17 & 1.38 @xmath135 0.16 & 1.79 @xmath135 0.15 + vucd3/strom547 & 1.40 @xmath135 0.14 & 4.99 @xmath135 0.12 & 2.69 @xmath135 0.13 & 2.37 @xmath135 0.15 & 2.53 @xmath135 0.14 & 3.60 @xmath135 0.14 + vucd4 & 2.16 @xmath135 0.18 & 1.42 @xmath135 0.16 & 1.41 @xmath135 0.18 & 1.18 @xmath135 0.21 & 1.30 @xmath135 0.20 & 1.38 @xmath135 0.17 + vucd5 & 1.82 @xmath135 0.17 & 3.30 @xmath135 0.15 & 2.11 @xmath135 0.17 & 1.90 @xmath135 0.19 & 2.01 @xmath135 0.18 & 2.60 @xmath135 0.17 + vucd6 & 2.27 @xmath135 0.16 & 1.58 @xmath135 0.15 & 1.40 @xmath135 0.17 & 0.87 @xmath135 0.19 & 1.14 @xmath135 0.18 & 1.41 @xmath135 0.17 + vucd7 & 1.90 @xmath135 0.32 & 2.45 @xmath135 0.29 & 1.75 @xmath135 0.33 & 1.09 @xmath135 0.38 & 1.42 @xmath135 0.36 & 1.96 @xmath135 0.33 + strom417 & 2.18 @xmath135 0.23 & 2.86 @xmath135 0.20 & 1.82 @xmath135 0.23 & 1.47 @xmath135 0.25 & 1.65 @xmath135 0.24 & 2.22 @xmath135 0.23 + vcc1407 & 2.08 @xmath135 0.24 & 2.20 @xmath135 0.21 & 1.63 @xmath135 0.24 & 1.53 @xmath135 0.27 & 1.58 @xmath135 0.26 & 1.88 @xmath135 0.24 + ngc4486b & 1.39 @xmath135 0.04 & 5.24 @xmath135 0.04 & 2.93 @xmath135 0.04 & 2.60 @xmath135 0.05 & 2.77 @xmath135 0.05 & 3.86 @xmath135 0.04 lcccccc vucd1 & 7.51 @xmath135 0.32 + vucd3/strom547 & 6.92 @xmath135 0.27 + vucd4 & 5.71 @xmath135 0.45 + vucd5 & 7.97 @xmath135 0.35 + vucd6 & 6.13 @xmath135 0.41 + vucd7 & 6.37 @xmath135 0.71 + strom417 & 8.62 @xmath135 0.45 + vcc1407 & 6.90 @xmath135 0.51 + ngc4486b & 7.32 @xmath135 0.09 | we present new imaging and spectroscopic observations of six ultra - compact dwarf ( ucd ) galaxies in the virgo cluster , along with re - analysed data for five fornax cluster ucds .
these are the most luminous ucds : @xmath0 mag . our _ hubble space telescope _
imaging shows that most of the ucds have shallow or steep cusps in their cores ; only one ucd has a flat `` king '' core .
none of the ucds show tidal cutoffs down to our limiting surface brightness .
spectroscopic analysis shows that virgo ucds are old ( older than 8 gyr ) and have metallicities in the range from [ z / h ] = -1.35 to + 0.35 dex .
five virgo ucds have super - solar [ @xmath1/fe ] abundance ratios and one virgo ucd has a solar abundance ratio .
the super - solar [ @xmath1/fe ] abundances are typical of old stellar populations found in globular clusters and elliptical galaxies .
we find that virgo ucds have structural and dynamical properties similar to fornax ucds .
the virgo and fornax ucds all have masses @xmath2 and mass - to - light ratios @xmath3 .
the dynamical mass - to - light ratios for virgo ucds are consistent with simple stellar population model predictions : the virgo ucds do not require dark matter to explain their mass - to - light ratios .
we conclude that the internal properties of virgo ucds are consistent with them being the high - mass / high - luminosity extreme of known globular cluster populations .
we refrain from any firm conclusions on fornax ucd origins until accurate age , metallicity and @xmath1-abundance estimates are obtained for them .
some of our results , notably the fundamental plane projections are consistent with the formation of ucds by the simple removal of the halo from the nuclei of nucleated dwarf galaxies .
however the ages , metallicities and abundances for virgo ucds are not consistent with this simple stripping model
. it might be consistent with more sophisticated models of the stripping process that include the effects of gas removal on the chemical evolution of the nuclei . |
we need eigenvalues of the shape matrix @xmath69 ( [ eq : a ] ) of dimension @xmath70 .
@xmath69 is block tridiagonal with blocks of size @xmath114 .
we first observe @xcite that in @xmath136 as well as in @xmath137 the matrix on the 1st tensorial subspace is either @xmath138 or the identity .
therefore , by simple @xmath139 dimensional rotation @xmath69 can be brought to an upper triangular form , showing that all eigenvalues of @xmath69 are at least twice degenerate and that all eigenvalues can be obtained from smaller @xmath140 matrix @xmath141 ( each eigenvalue of @xmath141 appears twice in the spectrum of @xmath69 ) , @xmath142 where @xmath143 matrices are @xmath144_{jk}=\delta_{j , r}\delta_{k , r}$ ] , @xmath145_{jk}=\ii(\delta_{j-1,k}+\delta_{j+1,k})$ ] , while @xmath146 and @xmath147 .
@xmath141 is again block - tridiagonal .
characteristic polynomial of a block - tridiagonal matrix can be written in terms of transfer matrices @xcite . defining a @xmath148 matrix @xmath149 as @xmath150 where @xmath151 , the determinant of @xmath152 is equal to @xmath153 times the determinant of the upper - left @xmath154 block of @xmath149 , giving @xmath155}=(-1)^{n-1}{\rm det}\ , [ -\ii p_{n-2}(b_n^\lambda b_1^\lambda)+\nonumber \\ & + & p_{n-3}(b_n^\lambda+b_1^\lambda)+\ii p_{n-4}\mathbbm{1}_2 ] , \end{aligned}\ ] ] where @xmath156 are polynomials in @xmath100 given by the recursion @xmath157 , with starting points @xmath158 .
recursion can be solved , getting @xmath159 , where @xmath160 is the divisor and @xmath161 the remainder .
up to a prefactor these are in fact fibonacci polynomials @xmath162 , and we have @xmath163 .
characteristic polynomial can therefore be finally written as @xmath78,\nonumber \\ a&=&4\gamma^2\left(\cosh{\frac{s}{2}}+\mu\sinh{\frac{s}{2}}\right)^2 - 4\gamma^2,\nonumber \\
r_n(\lambda)&=&\gamma^2 f_{n-1}(\lambda)+2\gamma f_n(\lambda)+f_{n+1}(\lambda).\end{aligned}\ ] ] fibonacci polynomials @xmath79 satisfy recursion @xmath164 , with starting @xmath165 , @xmath166 .
@xmath79 is of order @xmath167 in @xmath168 . because all coefficients of @xmath81 are
positive complex roots come in conjugate pairs @xmath80 , @xmath169 , and because @xmath76 is even in @xmath100 all roots come in pairs @xmath170 , see also illustration in fig . [
fig : roots ] .
critical current @xmath135 is equal to the asymptotic slope of @xmath31 ( in the limit @xmath171 and @xmath172 ) .
instead of directly calculating the derivative of @xmath31 we can evaluate it in a simpler way as @xmath173 , @xmath174 with @xmath175 as always being the roots of @xmath76 ( [ eq : qn ] ) .
beware that the order of the two limits is crucial exchanging them gives infinity .
as we shall see , at fixed @xmath4 the roots @xmath175 , as one increases @xmath45 , become increasingly close to those in the case @xmath176 .
the idea then is to use perturbation theory to obtain @xmath175 in the tdl . before tackling that problem
let us have a look at a simpler case of perturbation theory at @xmath65 for which @xmath177 . setting also @xmath176
the situation is trivial , with the characteristic polynomial being @xmath178 , where we used that @xmath179 are even / odd functions of @xmath168 for odd / even @xmath180 .
roots of @xmath181 are just @xmath182 , and we get that @xmath183 , with @xmath184 ( every @xmath80 being a double root of @xmath76 ) . these are of course nothing but the eigenvalues of an uncoupled tight binding model ( i.e. , of a harmonic chain ) . for small @xmath52 one can use perturbation theory . doing that we have @xmath185 . using that @xmath186
, we get in the lowest order in @xmath52 the expression @xmath187 , with @xmath184 .
important here is a prefactor @xmath188 , comming from @xmath189 , and causing the correction to become small in the tdl .
incidentally , from perturbative expression for @xmath80 we can also read out the value of the gap @xmath190 of the liouvillian in the even subsector , @xmath191 . note that this expression is not necessarily exactly equal to the gap of the whole spin liouvillian @xmath62 because of the symmetry - subspace complications connected with a string of @xmath66 in @xmath63 . for nonzero @xmath4 and large @xmath45 we can use the same reasoning .
roots become closer and closer to the imaginary axis ( because of an @xmath188 prefactor in the correction ) , so that eventually one can use perturbation theory , see fig . [
fig : roots ] . to evaluate @xmath135 ( [ eq : jc ] )
we also only need real parts of @xmath80 ( due to complex - conjugate pairs of roots imaginary parts cancel ) . for those we get ( fixing for simplicity
@xmath119 , @xmath91 ) @xmath192 , @xmath193 $ ] . replacing summation in ( [ eq : jc ] ) by an integral we finally get @xmath194 . from fig .
[ fig : phi ] one can see that this value nicely agrees with the location of the divergence in @xmath18 .
the procedure to calculate higher order cumulants is analogous to calculation of @xmath199 .
intermediate steps are a bit messy and we shall write only final results , @xmath200,\nonumber\\ j_5&=&\frac{\mu}{8\epsilon^9}\left[\epsilon^4 d_4 - 30\mu^2\epsilon^2 d_2 + 90\mu^4d_0 \right],\end{aligned}\ ] ] where @xmath201 , @xmath202 , and @xmath203 ( remember that @xmath204 ) . for @xmath205 one gets @xmath206,\ ] ] where @xmath207 , @xmath208 , @xmath209 , and @xmath210 .
one could also calculate few even higher cumulants , however , already @xmath205 is rather complicated and such exact expressions , per se , do not offer much physical insight .
an interesting observation that we made is that @xmath44 is independent of @xmath45 for @xmath104 .
this is a consequence of various identities involving roots of linear combinations of fibonacci polynomials .
the simplest one , needed in @xmath118 , states that , if @xmath211 ( @xmath212 are arbitrary complex numbers ) and @xmath213 are solutions of @xmath214 , then @xmath215 for any @xmath216 .
b. derrida and j. l. lebowitz , phys .
* 80 * , 209 ( 1998 ) .
b. derrida , j. l. lebowitz , and e. r. speer , phys .
lett . * 87 * , 150601 ( 2001 ) .
j. de gier and f. h. l. essler , phys .
lett . * 107 * , 010602 ( 2011 ) .
m. gorrissen , a. lazarescu , k. mallick , and c. vanderzande , phys .
lett . * 109 * , 170601 ( 2012 ) .
a. lazarescu , j. phys .
a * 46 * , 145003 ( 2013 ) .
k. saito and a. dhar , phys .
lett . * 99 * , 180601 ( 2007 ) .
j. li , y. liu , j. ping , s .- s .
li , x .- q .
li , and y. yan , phys .
b * 84 * , 115319 ( 2011 ) .
j. m. hickey , s. genway , i. lesanovksy , and j. p. garrahan , phys .
a * 86 * , 063824 ( 2012 ) . c. ates , b. olmos , j. p. garrahan , and i. lesanovsky , phys .
a * 85 * , 043620 ( 2012 ) .
d. karevski and t. platini , phys .
lett . * 102 * , 207207 ( 2009 ) .
m. nidari , j. phys .
a * 43 * , 415004 ( 2010 ) .
m. nidari , b. unkovi , and t. prosen , phys .
e * 84 * , 051115 ( 2011 ) .
t. antal , p. l. krapivsky , and a. r ' akos , phys .
e * 78 * , 061115 ( 2008 ) .
note that , because we use @xmath220 instead of @xmath221 in our definition of @xmath23 , we are in fact studying the particle current and not true magnetization current , being equal to @xmath222 . the central moment @xmath223 of the magnetization current
can be simply obtained as @xmath224 .
d. j. evans , e. g. d. cohen , and g. p. morriss , phys . rev
. lett . *
71 * , 2401 ( 1993 ) ; g. gallavotti and e. g. d. cohen , j. stat . phys . * 80 * , 931 ( 1995 ) ; j. l. lebowitz and h. spohn , j. stat . phys . * 95 * , 333 ( 1999 ) . | we consider a one - dimensional xx spin chain in a nonequilibrium setting with a lindblad - type boundary driving . by calculating large deviation rate function in the thermodynamic limit ,
being a generalization of free energy to a nonequilibrium setting , we obtain a complete distribution of current , including closed expressions for lower - order cumulants .
we also identify two phase - transition - like behaviors in either the thermodynamic limit , at which the current probability distribution becomes discontinuous , or , at maximal driving , when the range of possible current values changes discontinuously . in the thermodynamic limit
the current has a finite upper and lower bound .
we also explicitly confirm nonequilibrium fluctuation relation and show that the current distribution is the same under mapping of the coupling strength @xmath0 .
systems out of equilibrium are of wide importance . on a practical side ,
all dynamical processes , for instance in devices performing some useful work , involve nonequilibrium states supporting nonzero currents .
also , phenomena not possible in equilibrium , for instance finite - temperature phase transitions in local one - dimensional systems , become possible out of equilibrium .
unfortunately , understanding nonequilibrium physics is more difficult than understanding equilibrium . for a start
, there is no general nonequilibrium formalism analogous to equilibrium measures and thermodynamic potentials .
the best method available for study of stationary nonequilibrium systems is the so - called large deviation ( ld ) formalism @xcite , which enables us to assess , in addition to most probable values , also the whole distribution function .
because the formal mathematical frame - set used in the ld approach is very similar to the one used in equilibrium statistical physics , one can argue that the functions provided by the ld method are generalizations of equilibrium thermodynamic potentials like the entropy and free energy . in the last 15 years ld
formalism has been successfully applied to classical nonequilibrium systems , most notably to exclusion processes @xcite .
a ld function of heat current in a chain of harmonic oscillators described by a quantum langevin equation has been calculated in ref .
@xcite . for quantum systems
much less is known because exact quantum solutions are typically more difficult to obtain .
in fact , full ld formalism has been applied to quantum systems only very recently @xcite , mainly to single or two qubit systems @xcite , or , within a mean field approximation also in an infinite ising spin chain @xcite . because interesting phenomena , like phase transitions , are possible only in the thermodynamic limit ( tdl ) it is important to have solutions , possibly exact ones , for thermodynamically large many - body systems .
no exact ld statistics are known for quantum systems in the tdl , and the goal of this work is to provide one .
hoping for an an exact solution we shall study the simplest possible nonequilibrium quantum system and that is a boundary driven one - dimensional xx spin chain ( being equivalent to a tight binding model ) .
it is one of the first driven thermodynamically large quantum systems for which an exact nonequilibrium steady state ( ness ) solution has been found @xcite , including a compact matrix product form of the ness @xcite . despite being very simple , out of equilibrium it can exhibit interesting phenomena , for instance , when periodically driven its transport properties can vary drastically @xcite .
fluctuations in a closed xx chain , starting from a domain wall initial condition have been studied in ref .
@xcite .
we are going to study ld statistics in a ness that results from the evolution with the lindblad equation @xcite , @xmath1+\sum_{l_j } \cl_0(\rho;l_j)+\cl_{\rm j}(\rho;l_j,0 ) \equiv \cl(\rho;0 ) , \label{eq : lin}\ ] ] where we expressed the dissipator as a sum of the `` jump '' term @xmath2 , and the rest @xmath3 ( an extra parameter `` @xmath4 '' in @xmath5 will be needed in the ld formalism ) .
the whole liouvillian propagator is denoted by @xmath6 .
an especially important state in a ness @xmath7 which is a solution of a stationary lindblad equation , @xmath8 , and to which , in the absence of degeneracies , any initial state converges , @xmath9 .
expectation value of any observable @xmath10 in the ness is @xmath11 .
if one wants to calculate fluctuations of @xmath10 though , or even its complete distribution function , the information contained in @xmath7 is not enough .
namely , fluctuations in a ness are connected with nonequilibrium steady - state correlations between @xmath12 at different times .
one way to calculate these is via the ld formalism .
_ large deviation formalism. _ the ld formalism has been developed by mathematicians to study properties of distributions beyond the most probable values @xcite . to be concrete ,
let us consider a current @xmath13 and its integrated quantity called the number of particles , @xmath14 .
provided @xmath15 has a so - called ld property its probability distribution decays exponentially for large @xmath16 and can be written as @xmath17 , where @xmath18 is called a ld function and encodes information about the distribution of @xmath15 , or , equivalently , of the average current @xmath19 .
provided lindblad operators are such that the jump term @xmath20 either changes @xmath15 by @xmath21 , or leaves it the same , @xmath22 can be obtained from the largest eigenvalue of the so - called tilted liouvillian @xmath23 @xcite . to obtain @xmath23
we divide lindblad operators @xmath24 into three sets : the set @xmath25 of those @xmath24 that do not change @xmath15 , @xmath26 of @xmath24 that change @xmath15 by @xmath27 , and @xmath28 of operators that change @xmath15 by @xmath29 , and then form @xmath30 . denoting by @xmath31 the eigenvalue of @xmath23 having the largest real part ( we always have @xmath32 , with the corresponding eigenvector being @xmath7 )
, one can show that , provided @xmath31 is differentiable for all real @xmath4 , it is equal to the scaled cumulant generating function of @xmath15 . in other words , @xmath33 .
sometimes it is more practical to operate with the distribution function @xmath34 instead of with @xmath31 , which can be obtained from @xmath31 by the legendre - fenchel transform @xcite , @xmath35 denoting by @xmath36 the leading contribution in the limit @xmath37 , we can write ( for @xmath38 up - to normalization ) @xmath39 the scaled cumulants of @xmath15 are given by the derivatives of @xmath31 , and , because the measured current @xmath40 after time @xmath16 is simply @xmath41 , they are also equal to @xmath42^r}_{\rm c}$ ] , @xmath43^r}_{\rm c } = \left .
\frac{{\rm d}^r \lambda(s)}{{\rm d}s^r}\right|_{s=0}. \label{eq : jr}\ ] ] we shall call @xmath44 simply the ( scaled ) current cumulants .
one can see that the ld formalism essentially boils down to the calculation of @xmath31 , more details can be found in , e.g. , ref .
@xcite .
formally the ld formalism looks at the outset similar to a more recent full counting statistics @xcite , which also leads to cumulants , and is often employed in mesoscopic physics . with the ld approach though there is an important bonus
we get a complete distribution function , including statistical physics formalism relating @xmath18 and @xmath31 . among many works on full counting statistics
we shall only mention ref .
@xcite , studying a tight binding model with hamiltonian reservoirs , and the recent @xcite , which uses perturbation theory ( in the coupling strength of a boundary lindblad driving ) to derive expansion of current cumulants in spin chains . _ xx spin chain. _ the hamiltonian of the xx chain with @xmath45 sites is @xmath46 , where @xmath47 are pauli matrices .
boundary driving is described by 4 lindblad operators , @xmath48 , @xmath49 , @xmath50 , and @xmath51 .
parameters are the coupling strength @xmath52 and the driving strength @xmath53 .
such lindblad operators try to induce magnetization @xmath54 at the left end and @xmath55 at the right chain end . for
expectation values in @xmath7 see ref .
@xcite ; here we are interested in statistics of magnetization current @xcite ( i.e. , particle current in fermionic language ) , defined at site @xmath56 by @xmath57 , that can be determined by counting the amount of transferred magnetization ( particles ) at right chain end .
therefore , the tilted liouvillian is obtained by taking @xmath58 , @xmath59 and @xmath60 . using the jordan - wigner transformation both @xmath61 and the lindblad operators can be written in terms of fermionic operators .
the superoperator @xmath23 is quadratic and thereby its linear action on the space of operators can be as well expressed as a quadratic function of fermionic - like operators acting on the operator space .
such a procedure has been introduced in ref .
@xcite and has shown that the eigenvalues of @xmath62 can be expressed as simple sums over occupation numbers of super - eigenmodes , with the factors being the eigenvalues of the shape matrix . for our driving there
is one subtle technical issue : @xmath63 are not strictly linear in fermions ( they contain a string of @xmath64 ) .
one can show @xcite though that for @xmath65 the largest eigenvalue ( i.e. , the ness ) comes from the even parity sector with an even ( zero ) number of super - excitations , meaning that the action of a string of @xmath66 gives a trivial phase factor of @xmath67 . because @xmath31 is a smooth function of @xmath4 we know that we can get it also for @xmath68 by just considering that symmetry sector ( there are no degeneracies as a function of @xmath4 ) .
the shape matrix @xcite @xmath69 of size @xmath70 , whose eigenvalues we need , can be calculated and is in the even sector @xmath71 where @xmath72 and @xmath73 , with @xmath74 and @xmath75 .
one can first show @xcite that all eigenvalues are at least doubly degenerate and then calculate the characteristic polynomial @xmath76 of a smaller @xmath77 block tridiagonal matrix , resulting in , @xmath78,\nonumber \\ a&=&4\gamma^2\left(\cosh{\frac{s}{2}}+\mu\sinh{\frac{s}{2}}\right)^2 - 4\gamma^2,\nonumber \\
r_n(\lambda)&=&\gamma^2 f_{n-1}(\lambda)+2\gamma f_n(\lambda)+f_{n+1}(\lambda ) , \label{eq : qn}\end{aligned}\ ] ] where @xmath79 are fibonacci polynomials . denoting by @xmath80 the roots of polynomial @xmath81 ( [ eq : qn ] ) , we can express the largest eigenvalue @xmath31 of @xmath23 by a sum over all @xmath80 with positive real parts , @xmath82 _ fluctuation relation. _ the first important observation , which though rather trivially follows from an explicit form of @xmath81 that we derived ( [ eq : qn ] ) , is that the dependence on @xmath53 and @xmath4 is only via a constant coefficient @xmath83 .
one consequence is that the characteristic polynomial is invariant , i.e. , eigenvalues @xmath80 do not change , under the change @xmath84 , where @xmath85 .
as a consequence , symmetries @xmath86 as well as @xmath87 hold , or , equivalently , @xmath88 .
one can show that the same symmetry holds also in other symmetry sectors and therefore that the whole spectrum of @xmath23 is the same as the spectrum of @xmath89 @xcite .
this is nothing but a manifestation of a gallavotti - cohen - like symmetry or a nonequilibrium fluctuation theorem @xcite .
as we shall see , the divergence of @xmath90 at @xmath91 will be reflected in a discontinuous change of the definition range of current @xmath92 .
_ current cumulants. _ .
bottom : dependence of @xmath44 on @xmath52 for @xmath93 ( cross - section along the dashed line in the upper plot).,scaledwidth=45.0% ] .
bottom : dependence of @xmath44 on @xmath52 for @xmath93 ( cross - section along the dashed line in the upper plot).,scaledwidth=45.0% ] current cumulants ( [ eq : jr ] ) are obtained from eq .
( [ eq : lambda ] ) .
we therefore need derivatives of roots @xmath80 of @xmath76 at @xmath65 .
these can in turn be obtained by taking derivatives over @xmath4 of the characteristic polynomial @xmath81 . doing that we obtain @xmath94 where @xmath95 ( note that @xmath96 ) , @xmath97 and @xmath98 is a derivative of @xmath99 ( [ eq : qn ] ) with respect to @xmath100 . for higher cumulants expressions
get more complicated ; we will write out just the one for @xmath101 , @xmath102 for small @xmath45 one can explicitly express zeros @xmath103 and therefore also cumulants in terms of parameters @xmath53 and @xmath52 .
importantly , one notes that @xmath44 is independent of @xmath45 provided @xmath104 @xcite !
therefore , one can obtain @xmath44 in the tdl already by calculating derivatives of @xmath31 for @xmath105 , enabling one to get exact expressions for lower cumulants , @xmath106 , \label{eq : j123}\end{aligned}\ ] ] where they are expressed as functions of @xmath107 .
exact expressions for @xmath108 in the tdl can be found in @xcite .
the general form of @xmath44 is @xmath109 , where @xmath110 for odd @xmath111 while @xmath112 for even @xmath111 , and @xmath113 is an even polynomial of degree @xmath114 in @xmath53 and of degree @xmath115 in @xmath116 .
an important observation is also that all cumulants are functions of only @xmath116 , meaning that @xmath44 , ld function @xmath18 as well as @xmath31 are , in the tdl , invariant under mapping @xmath117 ( note that this invariance does not hold for some other observables in @xmath7 , for instance for magnetization @xcite ) . observe that @xmath44 have , apart from @xmath118 , a rather nontrivial dependence on @xmath53 and @xmath52 , see fig .
[ fig : nicle ] .
_ large deviation function. _ we were not able to obtain an analytic expression for @xmath31 in the tdl and for large @xmath4 .
however , zeros of @xmath81 , and therefore also @xmath31 , can be computed numerically for sizes @xmath45 of several hundred . for not too large @xmath4 the convergence with @xmath45 is actually quite fast ( see fig .
[ fig : lambda ] ) . in the thermodynamic limit ( full red lines ) for different values of driving @xmath53 ( @xmath119 ) .
the asymptotic linear behavior of @xmath31 reflects a strict upper and lower bound on the possible current values ( [ eq : lf ] ) .
two dotted lines for @xmath120 are finite-@xmath45 results ( blue for @xmath121 and black for @xmath122).,scaledwidth=45.0% ] an important property of @xmath31 is that for large @xmath123 the derivative @xmath124 converges to a constant .
a consequence of that is that using the legendre - fenchel transform ( [ eq : lf ] ) results in @xmath18 being infinite for @xmath92 that are larger / smaller than the minimal / maximal slope of @xmath31 .
this can be nicely seen in fig .
[ fig : phi ] where we show @xmath18 in the tdl .
remember that the zeros of @xmath18 correspond to the most probable values ( in our case the ness expectation value @xmath118 ) . in the thermodynamic limit ( @xmath119 ) .
@xmath18 is a non - smooth function of @xmath92 .
the definition range discontinuously changes from @xmath125 for @xmath126 to @xmath127 at @xmath91.,scaledwidth=45.0% ] several interesting things can be observed .
first , @xmath18 is finite , and therefore also the distribution function of the current , only within the interval @xmath125 for all @xmath126 and within @xmath127 for @xmath91 .
definition range of the current discontinuously changes at @xmath91 .
one can calculate @xcite that @xmath128 and is independent of @xmath53 and @xmath52 .
overall , @xmath18 and @xmath31 have an inessential dependence on @xmath52 . at the boundaries of this definition range
@xmath18 has an infinite jump .
this is in turn reflected in a discontinuity of the distribution function @xmath38 , for instance , @xmath129 is finite while @xmath130 .
a nonanalytic behavior of a thermodynamic potential is a characteristic feature of phase transitions and one can say that the open xx chain is a model residing at a phase transition point .
this explains and confirms a little mysterious finding in ref .
@xcite that the xx model exhibits a discontinuous change from being a ballistic conductor to being a diffusive conductor as one adds nonzero dephasing . just from @xmath7 , studied there , it was not clear where that discontinuity came from .
( @xmath120 , @xmath119 ) . dashed red lines are unnormalized distributions @xmath131 ( shown for @xmath132 and @xmath133 , and @xmath134 ) , trying to indicate how the current distribution looks.,scaledwidth=45.0% ] it is actually instructive also to look at finite-@xmath45 values of @xmath31 and @xmath18 , see fig .
[ fig : converg ] and [ fig : lambda ] .
one observes that for any finite @xmath45 , at a sufficiently large @xmath123 , @xmath31 eventually begins to increase / decrease faster than with the slope @xmath135 .
this means that for finite @xmath45 thermodynamic functions are ( expectedly ) smooth and that the definition range of the current is the whole real axis for @xmath126 , or the whole positive real axis for @xmath91 . _
conclusion. _ we present results for the large deviation statistics of a nonequilibrium quantum xx spin chain driven at the boundaries .
despite being a simple textbook model ( a driven tight binding model ) , we find a rather rich behavior out of equilibrium . in the thermodynamic limit the large deviation function exhibits nonanalyticity , signaling a phase - transition - like behavior , reflected for instance in a discontinuity of the current distribution function .
for a few lower - order cumulants we provide exact closed expressions .
we also explicitly show the validity of a nonequilibrium fluctuation theorem .
the study elucidates the importance of large deviation formalism for quantum systems , since , for instance , the nonanalytic properties discovered here are not visible in simple steady - state expectation values . |
the fermi spacecraft supports two gamma - ray instruments ; the large area telescope ( lat ) @xcite and the gamma - ray burst monitor ( gbm ) @xcite .
the lat is a wide - field gamma - ray telescope ( 20 mev - 300 gev ) that continuously scans the sky , providing all - sky coverage every two orbits .
the gbm is an all - sky monitor ( 10 kev - 25 mev ) that detects transient events such as occultations and gamma - ray bursts ( grb ) .
gbm detections of strong grbs can result in an autonomous re - point of the observatory to allow the lat to obtain afterglow observations .
the satellite sends data to the ground every 3 hours .
data is transferred via relay satellites at 40 mb / s to the white sands ground station .
it then follows a leased line to the mission operations center ( moc ) at goddard space flight center where data is split into two parts and sent for science processing to both the gbm and the lat teams , the latter being located at stanford national accelerator laboaratory ( slac ) .
the processing of downlinked satellite data is a time - critical operation .
it is , therefore , necessary to automatically trigger the pipeline for each newly arrived block of data , and to exploit parallel processing in a batch farm to achieve the required latency for the production of the various data products .
this processing is complex and is abstracted in a process graph .
an xml representation of this process graph is interpreted by the pipeline to become a _ task_. once defined , a task is exercised by the creation of _ streams _ , each of which is one instance of the process graph and which consists of an arbitrary number of interconnected batch jobs ( or scripts ) known as _ process instances _ @xcite . to that end
, the software is designed to be able to handle and monitor thousands of streams being processed at separate sites with a daily average throughput of about 1/2 cpu - year of processing . to date peak usage has been 45,000 streams in a single day and 167 cpu - years of processing in a single month @xcite .
the pipeline was designed with the l1 data processing task ( `` level 1 processing '' , the core data processing of raw data that comes from the satellite ) , automatic science processing ( asp ) and monte carlo simulations as principle task types in mind and to ensure the tight connection to the fermi data catalog .
it is literally impossible to depict the whole l1 task scheme in one single figure as it contains many dozens of sub - stream creations , dependencies and automatic re - run mechanisms .
thus we refrain from including them in this paper . for details
the reader is referred to @xcite .
instead we show the simple layout of a monte carlo task ( as our efforts initially are geared towards porting them to grid sites ) in fig .
[ [ fig : mctask ] ] that does not rely on external dependencies such as local databases .
this task simply consists of 3 steps , the generation of monte carlo data on a computing node , the transfer of the mc products to slac and their registration in the data catalog .
the first 2 steps are batch operations where the registration step takes only split seconds and is achieved through running a dedicated jython @xcite scriptlet .
we use a three tier architecture as shown in fig . [ [ fig : keytech ] ] comprising of back - end components , middle - ware and front - end user interfaces .
we describe them more in detail below .
core of the back - end components is the oracle database that stores all processing states .
in addition we make extensive use of oracle technologies such as the scheduler that is used to run periodic jobs for system monitoring including resource monitoring of the oracle server itself .
most quantities are made available to the user through trending plots that allow quick judgment about the state of the system along with its resource usage .
another back - end component is the pipeline job control service . using remote method invocation ( rmi @xcite )
they publish the uniform interface and communicate with the pipeline server .
we discuss some more details on this component in the next section .
to provide asynchronous persistent messaging from batch jobs to the pipeline server we use email messaging . at the beginning of a job an email
is sent detailing the host name and other worker node specific information .
another email is sent to indicate that a job has finished .
this email may also contain additional commands that invoke new pipeline commands , such as creation of sub - streams as the next step . in order to avoid overloading the slac email server , by relaying tens of thousands emails per day
, we use a dedicated email server running the free apache james @xcite software .
the pipeline server is the core of the system and contains two pools for threads , a worker and an admin pool . when a batch process is ready to run on one of the farms , a thread is allocated on the worker pool to perform the submission using the appropriate job control service .
extensions to the pipeline ( called _ plugins _ ) can be used to add additional functionality , for example to provide access to experiment specific databases or to communicate with other middleware services without compromising the experiment independent design of the core pipeline software .
plugins are written in java and loaded dynamically when the pipeline starts .
the user can also provide jython scriptlets to run within the pipeline threads to perform simple calculations and to communicate with plugins .
the fermi data catalog is implemented as a plugin .
the pipeline server api allows queries for processes , stream management as well as means to get or set environment variables .
the admin thread pool is used to identify work to be delegated to the worker pool .
this includes gathering processes which are ready to run as well as various database queries .
we provide a subset of the pipeline api as java management extension ( jmx ) , that provides a call interface to various user - interface applications .
these come both as web interface and line command applications .
the web interface provides password protected world - wide access to the pipeline and its control interfaces and allow simultaneous monitoring of tens of thousands of jobs in various states . for detailed technical information on
the pipeline components the reader is referred to @xcite . each job control service ( refer to fig . [
[ fig : jcd ] ] ) implements job control and status methods that are specific to the batch or grid system . to that
end each job control service needs to provide the following commands : _
submitjob , getjobstatus , killjob_. the code for the job control service is written in java and runs as a daemon on a dedicated service machine at each computing site .
to date the pipeline supports lsf @xcite , bqs and more recently also condor as well as sun grid engine ( sge @xcite ) . at
present uses lsf at slac and sge at the lyon computing center .
the code can easily be adapted to any other desired batch system that follows the same job logic as the currently supported ones . to that end a java class _ batchjobcontrolservice _ and _ batchstatus _ need to be implemented for the desired new batch system , where _ batch _ denotes the system to be implemented . for the use with dirac ( distributed infrastructure with remote agent control ) the job control code wraps python scripts that provide the bridge in both format and language by using the dedicated dirac api .
we describe further aspects of the dirac system and motivate our choice to use it in section [ dirac ] . in the past
a task was defined on top level to be handled by a specific job control service .
recently the pipeline has been enabled to support multi - site tasks that allow the pipeline server to delegate the running of commands to our dirac job control service while we leave the transfer step to the lyon - based sge job control service .
as of this writing the lat has been granted resources both at slac ( dynamic allocation scheme ) and lyon ( guaranteed 1200 cores allocation ) . at lyon
all resources are used for monte carlo productions while at slac the total allocation is shared between l1 , monte carlo production and individual user jobs from the collaboration . as a recent challenge we have started reprocessing all of our data that was taken from the beginning of the mission up until now and are reprocessing it with our current state of knowledge about the experiment .
the reprocessing requires a significant amount of our computing resources .
as allocations are dynamic , users running science analysis may be directly impacted through less available slots on the batch farm .
another recent challenge was a massive monte carlo production of proton runs that occupied our resources both at slac and lyon for several months .
while not being particularly storage intensive , we do require significant amounts of cpu time .
since this production run was setup with the previous interation of the instrument response , dubbed `` pass 6 '' , it is likely that simulation requests of this kind may be repeated as our knowledge of the experiment grows .
it is thus important to investigate possibilities to extend our resources that can be utilized within the current pipeline framework .
although we perform standard computing tasks at our two sites on local batch farms , there exists the virtual organization ( vo ) glast.org .
this organization was founded in 2009 to provide access to glite @xcite resources granted by participating institutions in italy and france . at present
the vo includes 13 sites that are partially enabled for use by .
its use has however been limited to non - pipeline operations .
most notably , our existing grid resources have been used for stand - alone pulsar blind searches and some large monte carlo simulations .
these stand - alone tasks were unable to take advantage of the pipline s integration with the batch system and data catalog which made them more man - power intensive than their pipeline counterparts . in order to optimize the resource usage provided by the egi @xcite sites supporting the glast.org vo
, we are exploring the use of the dirac system as potential connection between the pipeline and the grid resources @xcite .
the dirac system , originally developed to support production activities of the lhcb experiment , is today a general solution to manage the distributed computing activities of several communities .
one of its main components is the workload management system ( wms ) .
the key feature of the dirac wms is the implementation of the ` pilot job ' mechanism , which is widely exploited by all lhc communities as a way to guarantee high success rate for user jobs ( workloads ) .
more in detail , workloads are submitted to the dirac wms and inserted in the dirac _ central task queue_. the presence of workloads in the dirac _ central task queue _ triggers pilot jobs submission .
pilot jobs are regular grid jobs that are submitted through the standard glite wms .
once the pilot job gets on the worker node , it performs some environment checks and only in case these tests succeed , the workload is pulled from the dirac _ central task queue _ and executed . in case the environment checks fail or the pilot job gets aborted ( for example because of a mis - configured site ) , only the pilot job is affected .
the net result of this mechanism is a significant improvement on the workload success rate . also , since the resources are pre - reserved by pilot jobs , the waiting time to get the workload execution started is reduced .
while dirac has specifically been designed to tackle intrinsic grid inefficiencies through its pilot job concept , it can not solve some of the initial issues when being connected to the pipeline system .
in particular : * the grid uses personalized certificates . when submitting jobs through the pipeline web interface , we submit them with a generic user i d , which is most feasible as the number of authorized pipeline users is small .
* some of our sites , in particular those from infn , can not directly send emails to the pipeline server , thus rendering the standard communication of job status and task logic de - facto unusable . * in general large data on the grid
would be stored on grid storage elements ( grid se ) and require user intervention to download it once finished .
this makes the automatic copy to our central slac xrootd space difficult to be used , in particular using the tight integration with the data catalog .
we decided to implement an interface to the dirac system for several reasons : * independence of grid middleware : currently the vo comprises only glite sites but dirac is able to communicate with all common grid middleware platforms , thus making it less difficult to connect to other grid initiatives , such as the open science grid in the us @xcite . *
additional monitoring functionality : dirac introduces a more detailed job monitor that in addition reports minor and major application statuses together with the overall job status , thus allowing easier debugging of tasks without the definite need to inspect log files manually .
this task has always been among the most time consuming .
we hope to achieve an improvement by using the new monitor , albeit as read - only system .
the reason for that is that the pipeline i d and the grid i d are generally not the same and a re - submitted job on the pipeline keeps its i d while on the grid it is relaunched and assigned a new unique grid i d . * while is entering its 4th mission year , the number of developers for software has begun to dwindle .
thus it is important to keep the required manpower as low as possible while maintaining full performance and possibly improving the capabilities of the pipeline . by using dirac
we can build our interface on a system in active development with many possibilities to influence the development process to meet our needs . implementing the dirac interface comes in two steps , first the job control daemon as described in the previous section that wraps the basic pipeline commands through the dirac api and secondly the configuration of a dedicated dirac server .
the implementation is shown in fig .
[ [ fig : p2dirac ] ] .
we make use of two of diracs core technologies : the pilot factory mechanism is used to renew proxies and authenticate the job control daemon to the grid allowing us to use one certificate retaining our previous user scheme .
the dirac notification service provides means for the dirac server to communicate with grid worker nodes .
this service is both safe to inception and can be modified to relay the content of our status emails to the dirac server that itself implements the email communication with the pipeline server .
one typical by - product of running our code are more or less extensive log files .
usually we have several hundreds of small log - files that each do not exceed a few mb . in the past
other experiments had to artificially enlarge their log files to ensure the stability of the grid ses due to journalling of small files .
we can effectively circumvent this by declaring a local storage element at the computing center in lyon as a dedicated dirac se .
these storage elements are addressed like normal grid elements but they do not need journaling to function . since this is a local se , we can view its content e.g. through a web server providing access to log - files that can directly be linked from the pipeline server . in conclusion
we believe that this solution provides an easy to implement and maintain interface to the pipeline system used for -lat .
the existing vo resources of glast.org suggest the possibility to establish a new connection to grid services with our existing pipeline architecture .
we mitigate issues such as the transfer and subsequent registration of data products at the fermi data catalog by using existing pipeline technologies .
grid inefficiencies are handled by the dirac system that acts as a broker providing asynchronous communication with grid worker nodes and a closed mechanism to automatically renew proxies to use for grid operations .
we leave handling of meta data to the existing pipeline technologies . as such
, the pipeline software itself was designed in a manner to not contain any fermi specific functionality . through the plug - in feature of the middleware it is successfully used for data handling for the enriched xenon observatory or mc simulations for the cryogenic dark matter search experiment ( cdms @xcite ) , the upcoming cherenkov telescope array ( cta @xcite ) and the large synoptic survey telescope ( lsst @xcite ) , and
is also being considered for use by other nasa missions .
we acknowledge the ongoing generous support of slac ( us ) and in2p3 ( france ) as well as we thank them for increasing allocations at both sites .
furthermore we are thankful for grid resources provided by infn and in2p3 .
sz likes to thank the organizers for a stimulating meeting and particularly helpful discussions with stphane guillaume poss .
the -lat collaboration acknowledges generous ongoing support from a number of agencies and institutes that have supported both the development and the operation of the lat as well as scientific data analysis .
these include the national aeronautics and space administration and the department of energy in the united states , the commissariat lenergie atomique and the centre national de la recherche scientifique / institut national de physique nuclaire et de physique des particules in france , the agenzia spaziale italiana and the istituto nazionale di fisica nucleare in italy , the ministry of education , culture , sports , science and technology ( mext ) , high energy accelerator research organization ( kek ) and japan aerospace exploration agency ( jaxa ) in japan , and the k. a. wallenberg foundation , the swedish research council and the swedish national space board in sweden .
additional support for science analysis during the operations phase is gratefully acknowledged from the istituto nazionale di astrofisica in italy and the centre national dtudes spatiales in france .
9 atwood et al
. _ apj _ * 697 * ( 2009 ) http://cdms.berkeley.edu/ http://www.cta-observatory.org/ dubois , _ asp conf .
ser . _ * 411 * , eds .
d bohlender , d durand , & p dowler ( 2009 ) http://www.egi.eu/ http://www-project.slac.stanford.edu/exo/ flath et al . , _ asp conf .
ser . _ * 411 * , eds .
d bohlender , d durand , & p dowler , slac - pub-13549 ( 2009 ) see http://gammaray.msfc.nasa.gov/gbm/ for details on the gbm http://glite.cern.ch/ focke , `` implementation and performance of the fermi lat level 1 pipeline '' , ii .
fermi symposium ( 2009 ) http://www.platform.com/workload-management/high-performance-computing http://james.apache.org http://www.lsst.org/lsst/ http://www.jython.org/ https://www.opensciencegrid.org http://java.sun.com/javase/technologies/core/basic/rmi/index.jsp http://www.oracle.com/us/sun/index.htm tsaregorodtsev at al . , _ j. phys .
_ * 119 * ( 2008 ) | the data handling pipeline ( `` pipeline '' ) has been developed for the gamma - ray space telescope ( ) large area telescope ( lat ) which launched in june 2008 .
since then it has been in use to completely automate the production of data quality monitoring quantities , reconstruction and routine analysis of all data received from the satellite and to deliver science products to the collaboration and the fermi science support center . aside from the reconstruction of raw data from the satellite ( _ level 1 _ ) , data reprocessing and various event - level analyses are also reasonably heavy loads on the pipeline and computing resources .
these other loads , unlike level 1 , can run continuously for weeks or months at a time .
in addition it receives heavy use in performing production monte carlo tasks . in daily use
it receives a new data download every 3 hours and launches about 2000 jobs to process each download , typically completing the processing of the data before the next download arrives .
the need for manual intervention has been reduced to less than 0.01% of submitted jobs .
the pipeline software is written almost entirely in java and comprises several modules .
the software comprises web - services that allow online monitoring and provides charts summarizing work flow aspects and performance information .
the server supports communication with several batch systems such as lsf and bqs and recently also sun grid engine and condor .
this is accomplished through dedicated job control services that for are running at slac and the other computing site involved in this large scale framework , the lyon computing center of in2p3 . while being different in the logic of a task , we evaluate a separate interface to the dirac system in order to communicate with egi sites to utilize grid resources , using dedicated grid optimized systems rather than developing our own .
more recently the pipeline and its associated data catalog have been generalized for use by other experiments , and are currently being used by the enriched xenon observatory ( exo ) , cryogenic dark matter search ( cdms ) experiments as well as for monte carlo simulations for the future cherenkov telescope array ( cta ) . |
a real - valued polynomial , or a real - analytic function , @xmath3 on @xmath4 can be regarded as a hermitian form by considering the matrix of coefficients of the series .
that is , in multi - index notation , write @xmath5 the matrix @xmath6}_{\alpha \beta}$ ] is hermitian symmetric if and only if @xmath7 is real - valued ( @xmath8 is uniquely determined by @xmath7 ) .
when we apply linear algebra terminology ( such as _ rank _ , _ eigenvalues _ , _ signature _ , or _ positive semidefinite _ ) to @xmath7 , we simply refer to the underlying matrix @xmath8 .
when @xmath7 is a polynomial , the terminology is obvious .
when @xmath7 is real - analytic , then @xmath8 is an infinite matrix .
after possibly rescaling so that the series converges in a neighborhood of the closed unit polydisc , @xmath8 defines a hermitian trace - class operator , and the terminology easily extends to the real - analytic case ; see section [ section : racase ] . while the matrix depends on the point where we expand the series ,
we show that the rank does not change under a biholomorphic change of coordinates .
in particular the rank does not depend on the point where we expand the series . in order to state the first result
, we need to define what we mean by the rank of the restriction to an affine manifold .
the grassmannian @xmath9 is the manifold of all @xmath10-dimensional linear subspaces of @xmath4 , and the affine grassmannian @xmath11 is the manifold of all affine @xmath10-dimensional subspaces in @xmath4 .
notice that @xmath11 is an open set in @xmath12 .
let @xmath13 be a convex neighborhood of the origin , and define @xmath14 .
suppose that the series converges in @xmath15 .
let @xmath16 be an affine @xmath10-plane in @xmath4 ( i.e. an element of @xmath11 ) .
let @xmath17 be an affine embedding of @xmath16 in @xmath4 .
if @xmath16 intersects @xmath18 , then define @xmath19 it is not hard to show that the rank does not depend on the particular embedding @xmath20 , and therefore the notation makes sense .
furthermore as @xmath18 is convex @xmath21 is always connected .
if @xmath16 does not intersect @xmath18 , then define @xmath22 let @xmath23 .
[ thm : rankres ] let @xmath24 and let @xmath25 .
let @xmath3 be a real - analytic function defined in a convex neighborhood @xmath13 of the origin such that the complexified series converges in @xmath15 . if @xmath26 then @xmath27 .
moreover , there exists a function @xmath28 such that for any such real - analytic @xmath7 , @xmath29 it is worthwhile to state the theorem for bihomogeneous polynomials .
a polynomial @xmath3 is said to be _ bihomogeneous _ of _ bidegree _
@xmath30 if @xmath31 .
when @xmath7 is a polynomial then the coefficient matrix is simply a finite matrix .
the function @xmath32 in the following theorem is the same as above .
[ finiterankres ] let @xmath33 , @xmath34 , and @xmath25 .
then there exists a function @xmath28 such that for any bihomogeneous polynomial @xmath3 , @xmath35 as an application of theorem [ thm : rankres ] , we prove a rigidity result for mappings of hyperquadrics .
the hyperquadric @xmath36 is the set defined by @xmath37 when @xmath38 , then @xmath39 is simply the sphere .
hyperquadrics and spheres are the cr analogues of flat euclidean space from riemannian geometry . in cr geometry
, however , there is no analogue of the nash embedding theorem .
therefore , a natural question in cr geometry is to study the cr mappings @xmath40 for a cr manifold @xmath18 . in this paper
we take an open subset @xmath41 .
we study hyperquadrics @xmath42 that are not equivalent to the sphere . by a theorem of lewy @xcite
a cr function on @xmath42 extends to a holomorphic function of both sides of @xmath42 .
it is therefore enough to consider real - analytic cr mappings or in other words , restrictions of holomorphic mappings
. therefore we study holomorphic mappings of a neighborhood of @xmath18 in @xmath43 to @xmath44 that take @xmath41 to @xmath0 .
see the books @xcite for more background information .
when considering mappings between hyperquadrics , we need to consider problems of positivity rather than just rank .
see @xcite for more on positivity conditions .
after a linear fractional change of coordinates we can assume that @xmath45 and @xmath46 .
it is possible that the change of coordinates has a pole on @xmath18 , but as @xmath42 is a real hypersurface , there still has to exist a perhaps smaller dense open set @xmath47 such that @xmath48 is a cr mapping .
the study of cr mappings between spheres has a long history .
when the target is also a sphere ( @xmath49 ) and the codimension @xmath50 is small , then certain strong rigidity results can be proved ; see for example @xcite and the references within .
forstneri proved that sufficiently smooth cr mappings of spheres must be rational of degree bounded by a constant depending only on the dimensions involved @xcite .
when the target dimension is large , there is less rigidity .
when both the source and target are spheres , increasing the dimension of the target always adds new cr mappings as long as the dimension is large enough ; see @xcite .
furthermore , if the source is a sphere and the target is a hyperquadric not equivalent to a sphere , then for large target dimension not only do we always get new mappings , but we get new rational mappings of arbitrarily large degree ; see @xcite .
we therefore concentrate on the case when @xmath51 .
baouendi and huang @xcite proved that if @xmath52 , then @xmath53 must be a linear embedding .
later , baouendi , ebenfelt , and huang @xcite proved that if @xmath54 , then after a change of coordinates @xmath53 can be written as @xmath55 where an arbitrary cr mapping @xmath56 and @xmath57 are vector valued functions with the right number of components . in particular , unless @xmath58 and @xmath52 , @xmath59 is contained in a complex hyperplane .
the mapping can be written , after an affine change of coordinates , as the identity mapping direct sum an arbitrary cr mapping going into a lower dimensional ambient space .
for any mapping @xmath53 of hyperquadrics , if @xmath59 lies in a complex hyperplane , then we apply an affine change of coordinates and obtain a mapping @xmath60 from @xmath18 to @xmath61 for some @xmath62 and @xmath63 .
we can study the first @xmath64 components of @xmath60 as a mapping from @xmath42 to @xmath65 , as the last @xmath66 components of @xmath60 are arbitrary .
therefore , it is natural to study those mappings where @xmath59 is not contained in a hyperplane .
[ thm : rigidity ] let @xmath67 , @xmath41 be a connected open set , and @xmath40 be a real - analytic cr mapping such that @xmath59 does not lie in a complex hyperplane , then @xmath68 where @xmath69 is a constant depending only on @xmath70 , @xmath71 and @xmath2 .
note that the hypotheses on @xmath42 mean that @xmath42 is not equivalent to a sphere .
if @xmath42 were equivalent to a sphere , there is no rigidity . in this sense
, the theorem is optimal .
we are interested in what happens when @xmath72 . in this situation
, there do exist nontrivial mappings that do not map to a hyperplane .
see sections [ section : construct ] and [ section : stability ] for a general method of constructing mappings .
the bound that we obtain on @xmath1 is not sharp .
for example when @xmath73 and @xmath74 , it is possible to use the method in this paper to obtain an explicit bound @xmath75 see section [ section : greens ] for the definition of @xmath76 .
when @xmath77 we obtain @xmath78 , though we know that @xmath79 by the result of baouendi - huang . therefore , we dispense with finding explicit formulas for @xmath80 . unlike in the sphere to sphere case , there do exist non - rational mappings , and rational mappings of arbitrarily high degree . in section [ section : stability ]
we prove a stability result about existence of hyperquadric mappings that do not map into hyperplanes as long as both @xmath1 and @xmath2 are sufficiently large and comparable .
the proof is constructive and all constructed mappings are monomial ; that is , all components are single monomials .
the stability result discussed above shows that theorem [ thm : rigidity ] is optimal in the sense that super - rigidity as studied by baouendi - huang and baouendi - ebenfelt - huang only appears in small codimension . in general the best constant @xmath81
must grow asymptotically at least as a linear function of @xmath2 .
the methods explored in this paper do not readily give the best bound for @xmath81 as we have mentioned above , but we do obtain that @xmath81 asymptotically grows no faster than a polynomial in @xmath2 .
finally , we also note that the rigidity result extends when the target is an infinite dimensional hyperquadric @xmath82 , where @xmath83 ; see section [ section : hilbert ] .
lempert @xcite shows that any strictly pseudoconvex real - analytic compact hypersurface can be mapped into a sphere in a possibly infinite dimensional hilbert space .
we show that if @xmath42 is not equivalent to a sphere , then for a cr mapping @xmath84 where the image is not contained in a hyperplane we must have @xmath85 .
however , we also construct mappings whose image is not contained in a hyperplane from an @xmath86 with an indefinite nondegenerate levi - form to @xmath82 for finite @xmath2 .
the authors would like to acknowledge john dangelo for many useful conversations on the subject and suggestions relating to this project .
the second author would also like to thank peter ebenfelt for useful discussions on this subject .
finally the authors would like to acknowledge msri and aim for holding workshops on the subject of cr complexity , which the authors attended and which led to the present project .
we begin by recalling the relevant definitions and proving some basic results about restrictions of real - valued polynomials .
let @xmath87 denote the standard hermitian inner product .
let @xmath1 be a hermitian symmetric matrix , then we call @xmath88 a _ hermitian form _ where @xmath34 .
let @xmath89 be the mapping whose components are all the degree @xmath90 monomials in @xmath91 variables @xmath92 .
that is , @xmath93 is the mapping given by @xmath94 the mapping @xmath93 is called the _
veronese mapping in @xmath91 variables of degree @xmath90_. let @xmath16 be an @xmath95-dimensional subspace of @xmath96 .
let @xmath97 be a linear embedding of @xmath16 into @xmath96 .
define @xmath98 to be the linear mapping that makes the following diagram commute : @xmath99 define @xmath100 to be the matrix associated to the linear mapping @xmath101 , then @xmath102 we often abbreviate @xmath103 for @xmath100 when the particular choice of embedding is not crucial .
note that @xmath100 is not unique ; it depends on the particular embedding @xmath97
. however if we have @xmath97 and @xmath104 two embeddings of @xmath16 then we have @xmath105 where @xmath56 is an isomorphism .
[ lemma : changecoord ] let @xmath3 be a bihomogeneous polynomial defined for @xmath106 .
if @xmath107 is a linear change of coordinates of @xmath96 , then @xmath108 we prove a more general fact in lemma [ lemma : rankchangecoord ] .
the lemma shows that rank of @xmath109 does not depend on the choice of embedding @xmath97 .
we state this observation in the following lemma .
[ lemma : welldefined ] let @xmath110 where @xmath8 is a hermitian matrix , and suppose @xmath111 .
let @xmath16 be an @xmath95-dimensional subspace of @xmath96 .
let @xmath97 and @xmath104 be two embeddings of @xmath16 into @xmath96 . then @xmath112 [ defoftnd ] let @xmath113 denote the hyperplane in @xmath96 given by the following defining equation : @xmath114 where @xmath115 .
we embed @xmath113 using @xmath116 .
we abbreviate the notation for the restriction matrix @xmath117 by @xmath118 .
the restriction matrix @xmath118 is of size @xmath119 .
to study the maximum rank of restrictions of bihomogeneous polynomials to hyperplanes , it suffices to study the hyperplanes of the form @xmath113 .
[ lemma : restrictgraphs ] let @xmath3 be a bihomogeneous polynomial defined for @xmath120 and suppose @xmath111 .
then @xmath121 where @xmath122 ranges over all complex hyperplanes through the origin .
the set of hyperplanes of the form @xmath113 is dense in the grassmannian @xmath123 .
the rank of the restriction being bounded by a fixed integer is a holomorphic condition .
therefore @xmath124 achieves the maximum on an open dense set in @xmath123 , and the conclusion follows .
the main technical tool in this paper is green s hyperplane restriction theorem from @xcite .
we begin this section by recalling macaulay representations and their basic properties .
then we give a precise statement of green s theorem and translate the result into the language of this paper . given a non - negative integer @xmath125 and a positive integer @xmath90 ,
then @xmath125 can be uniquely expressed in the following form : @xmath126 where @xmath127 .
this expression is known as the @xmath90-th macaulay representation of @xmath125 .
following @xcite , we introduce the following notation : @xmath128 where we define @xmath129 for @xmath130 .
notice that @xmath131 is monotone increasing .
[ thm : green ] let @xmath20 be a linear system in @xmath132 and @xmath133 be the restriction of @xmath20 to a general hyperplane @xmath134 . if @xmath125 is the codimension of @xmath20 in @xmath132 and @xmath135 is the codimension of @xmath133 in @xmath136 , then @xmath137 we seek a bound of the dimension of @xmath20 in terms of the generic dimension of the restriction . thus we restate the result in this way .
first we restate the theorem in terms of matrices .
if @xmath20 is a linear system in @xmath132 then we can write : @xmath138 where @xmath139 . since each @xmath140 is a degree @xmath90 homogeneous polynomial in @xmath92 , then we can rewrite as @xmath141 where we think of each @xmath142 as a vector of the right dimension
. then we have @xmath143 where @xmath1 is the matrix whose rows are @xmath142 .
if we restrict @xmath20 to a hyperplane @xmath122 we obtain @xmath144 then @xmath145 , and therefore @xmath146 if we refer to a generic choice of @xmath122 then equivalently , we refer to a generic set of @xmath147 s , where we have : @xmath148 } at_{\vec{c}}^d$ ] . that is , the rank of @xmath149 over @xmath150 is equal to the rank of @xmath151 over the polynomial ring @xmath152 $ ] .
green s theorem can now be restated as follows : let @xmath1 be any matrix of size @xmath153 , and @xmath154 . recall the matrix @xmath118 is of size @xmath155 .
let @xmath156 } at_{\vec{c}}^d$ ] .
then we have the following relationship : @xmath157 where @xmath158 in the following lemma , we show that the lower bound @xmath159 is constant in degree .
[ lemma : gconstantdegree ] if @xmath160 , @xmath161 , and @xmath33 , then @xmath162 .
the lemma follows immediately from @xmath163_{<d+1 > } = \binom{n+d-1}{d+1 } + \left [ \binom{n+d}{d}-n \right]_{<d>}.\ ] ] by the lemma , we know that @xmath159 is constant in degree when it is defined .
we also have @xmath159 is monotone in @xmath81 and @xmath164 , hence the following bound is finite and well - defined for fixed @xmath66 . fix @xmath165 and @xmath66 .
define @xmath166 the function @xmath76 gives the sharp dimension bound from green s theorem .
the bound is achieved by restricting all the monomials in @xmath165 variables of degree @xmath90 . in two dimensions
, @xmath167 has a particularly simple form : @xmath168 one can show that @xmath169 , but this bound is not sharp in general . the version of green s theorem that we use in this paper is recorded in the following lemma .
the important point is that the bound @xmath76 does not depend on the degree @xmath90 .
[ lemma : green ] let @xmath24 .
if @xmath170 is a linear system in @xmath132 with rank @xmath81 and @xmath66 is the rank of the restriction to a general hyperplane ; namely , @xmath171 } a t^d_{\vec{c}}$ ] , then @xmath172 the lemma gives an upper bound for @xmath81 that is independent of @xmath90 for fixed @xmath66 and @xmath24 .
no such upper bound is possible if @xmath173 .
in this section , we prove theorem [ finiterankres ] . to illustrate the main idea of the proof we prove a weaker version of the main result for hyperplanes .
[ lemma : hyperplanesalg ] let @xmath3 be a real - valued bihomogeneous polynomial defined for @xmath120 , and suppose @xmath174 .
then there exists a function @xmath175 such that @xmath176 where @xmath122 ranges over all complex hyperplanes through the origin .
suppose @xmath7 is a nonzero bihomogeneous polynomial of bidegree @xmath30 .
we write @xmath110 , where @xmath8 is the coefficient matrix . restricting @xmath7 to a hyperplane @xmath122 ,
we have @xmath177 let @xmath178 where @xmath122 ranges over all complex hyperplanes , then by lemma [ lemma : restrictgraphs ] , @xmath179 } { ( t^d_{\vec{c}})^*ct^d_{\vec{c}}}.\ ] ] the strategy is to polarize and use lemma [ lemma : green ] twice .
let @xmath180 , @xmath181 , and let @xmath182 } d$ ] . by lemma [ lemma : green ] , we have @xmath183 for a generic choice of @xmath184 , we have @xmath185 likewise , polarizing and taking generic choices of @xmath186 , we have @xmath187 } { ( t^d_{\vec{c}})^*ct^d_{\vec{c } } } = { \operatorname{rank}}_{{{\mathbb{c}}}[c_1,\ldots , c_n ] } { d^{*}t^d_{\vec{c}}}\ ] ] applying lemma [ lemma : green ] again , we get @xmath188 thus @xmath189 , and we have an upper bound for @xmath190 that is independent of degree .
we complete the proof by defining @xmath191 to be the maximum @xmath190 for any bihomogeneous polynomial @xmath7 with @xmath192 . in the proof of lemma [ lemma : hyperplanesalg ]
, we proved that @xmath193 in fact , we believe that @xmath194 .
we are now ready to prove the main result , theorem [ finiterankres ] . for reader convenience
we restate the theorem .
let @xmath33 , @xmath34 , and @xmath25 .
then there exists a function @xmath28 such that for any bihomogeneous polynomial @xmath3 , @xmath35 suppose @xmath7 is a nonzero bihomogeneous polynomial of bidegree @xmath30 .
we again write @xmath110 . after applying a linear change of coordinates
we can assume without loss of generality that all @xmath195 are given as @xmath196 for @xmath197 and @xmath198 is a matrix .
using the embedding @xmath199 we obtain a restriction matrix @xmath200 .
we notice that @xmath200 depends holomorphically on the entries of @xmath201 .
we write @xmath202 and again note that @xmath203 if @xmath204 we are done .
we let @xmath205 .
otherwise , @xmath200 is a composition @xmath206 where @xmath122 is a hyperplane through the origin in @xmath207 , where @xmath207 is an @xmath208 dimensional plane through the origin in @xmath96 . as @xmath209 is of rank less than @xmath66
, @xmath210 is of rank less than @xmath211 by lemma [ lemma : hyperplanesalg ] .
the theorem follows by applying lemma [ lemma : hyperplanesalg ] @xmath212 times .
the function @xmath213 is then defined recursively by composition .
we say a set @xmath214 is _ generic _ if @xmath215 is not contained in any proper complex algebraic subvariety of @xmath12 .
that is , if no nontrivial polynomial vanishes on @xmath215 . with this terminology
, we have the following corollary : [ cor : rankpospoly ] let @xmath3 be a positive semidefinite bihomogeneous polynomial defined for @xmath34 , @xmath33 .
let @xmath25 . then there exists a function @xmath216 such that @xmath217 where @xmath214 is a generic subset .
note that @xmath218 for all @xmath120 is not enough to conclude that the matrix of coefficients is positive semidefinite . on the other hand ,
if the matrix is positive semidefinite , then @xmath218 is true for all @xmath120 .
see @xcite for more on positivity conditions .
the proof as before carries on after we make the following observation
. let @xmath110 .
if @xmath8 is positive semidefinite then we can write @xmath219 . in the restriction : @xmath202 and again note that @xmath220 let @xmath221 .
the condition @xmath222 is defined by the vanishing of certain polynomials in the entries of @xmath223 . defining @xmath16 with a matrix @xmath198 as in
, we notice that as @xmath1 is fixed , then is defined by vanishing of certain polynomials in the entries of @xmath201 .
we know that is true for all @xmath224 , where @xmath215 is not contained in any complex algebraic subvariety of @xmath12 .
the set of corresponding matrices @xmath201 is not contained in any complex algebraic subvariety of @xmath225 .
thus holds for all @xmath198 , and so for all @xmath195 .
the conclusion follows by applying lemma [ lemma : green ] @xmath212 times .
the function @xmath226 is then defined recursively by composition as in the proof of theorem [ finiterankres ] .
we wish to decompose real - analytic functions as a difference of squared norms of hilbert space valued holomorphic functions .
this line of reasoning follows the ideas pioneered by dangelo @xcite , although we introduce a rescaling that allows us to work with bounded operators .
first let us prove that the matrix of coefficients is a trace - class operator if a series converges in the right neighborhood . see the book @xcite for more information on matrix operators on @xmath227 .
let @xmath228 be the mapping of @xmath229 to the space of infinite sequences , where components of the mapping are all possible monomials . by using the geometric series we note that @xmath230 maps the unit polydisc @xmath231 to @xmath227 .
[ lemma : compactop ] let @xmath232 be the unit polydisc .
suppose that @xmath3 is a real - analytic function whose complexified power series converges in a neighborhood of @xmath233 .
then the matrix of coefficients of @xmath7 defines a trace - class operator on @xmath227 .
write @xmath7 as @xmath234 where @xmath235_{\alpha\beta}$ ] is the matrix of coefficients .
the hypothesis says that the interior of the domain of convergence includes the point @xmath236 .
as the series converges absolutely at @xmath237 we get @xmath238 . a matrix
whose entries are summable defines a compact operator .
moreover , if @xmath239 is the standard basis corresponding to the monomials , then @xmath240 here @xmath241 is the unique positive hermitian square root of @xmath242 .
so @xmath8 is trace - class . as @xmath8 is a hermitian trace - class operator ,
we apply the spectral theorem .
we write @xmath8 as a sum of rank one matrices as follows .
let @xmath243 be the @xmath66-th nonzero eigenvalue .
let @xmath244 be the sign of the @xmath66-th nonzero eigenvalue .
we ignore all zero eigenvalues .
let @xmath245 be an orthonormal set of corresponding eigenvectors .
let @xmath246 be the rank of @xmath8 , then we write @xmath247 the sum converges as the eigenvalues of a trace - class operator are absolutely summable .
we define holomorphic functions @xmath248 . as @xmath249
, we note that each @xmath250 converges in the unit polydisc .
we see that @xmath251 suppose that for some @xmath252 , @xmath253 , and some @xmath254 holomorphic in the unit polydisc we write @xmath255 if @xmath256 , then obviously @xmath257 .
if @xmath252 is finite then we have written @xmath8 as a sum of @xmath252 rank - one operators , and hence @xmath257 .
next we have to show that the rank of the matrix @xmath8 is unchanged under changes of coordinates .
we define the rank of @xmath7 at the origin to be the rank of the matrix @xmath8 .
[ lemma : rankchangecoord ] let @xmath232 be the unit polydisc .
suppose that @xmath3 is a real - analytic function whose complexified power series converges in a neighborhood of @xmath233 .
suppose that @xmath258 is a biholomorphic change of coordinates that takes a neighborhood of @xmath259 to a neighborhood of @xmath259
. then @xmath260 ( which also converges on a neighborhood of @xmath233 ) has the same rank as @xmath7 at the origin .
suppose that @xmath7 is of finite rank @xmath66 .
then there exist @xmath66 eigenvectors of @xmath8 and therefore we can write @xmath261 for @xmath262 and @xmath250 holomorphic functions .
therefore @xmath263 so @xmath264 . we are finished by symmetry
. we can also define the number of positive and negative eigenvalues at a point ( the rank of the positive or negative definite part ) .
using exactly the same argument as above we can show that the rank , the number of positive and the number of negative eigenvalues are constant on connected sets .
[ lemma : rankchangept ] let @xmath13 be a connected open set .
let @xmath265 be a real - analytic function .
let @xmath266 and let @xmath267 denote the rank , the number of positive , and the number of negative eigenvalues respectively at @xmath268 .
then @xmath269 , @xmath270 , and @xmath271 . with the aid of the lemmas above
we can define the rank of @xmath7 regardless of where it converges .
we simply translate and dilate @xmath7 so that it converges in the unit polydisc and then take the rank of the resulting compact operator .
if the domain @xmath18 of @xmath7 is connected , we define the _ rank _ of @xmath7 as the rank at some fixed point of @xmath18 .
similarly define the _ signature pair _ to be the pair @xmath272 if @xmath7 has @xmath70 positive and @xmath71 negative eigenvalues .
of course we allow @xmath70 and @xmath71 be in @xmath273 .
[ lemma : restcompact ] let @xmath16 be an affine complex submanifold of @xmath4 of dimension @xmath10 that intersects the unit polydisc @xmath274 and @xmath275 an embedding such that @xmath20 takes the closed unit polydisc @xmath276 to a subset of @xmath277 .
the restriction matrix @xmath200 defined by @xmath278 is a hilbert - schmidt operator .
note that @xmath279 is the absolute value squared of the geometric series and hence converges in the unit polydisc @xmath280 .
as @xmath20 takes the closed unit polydisc @xmath276 to @xmath281 , we have that @xmath282 converges absolutely for @xmath283 , therefore @xmath284 where @xmath200 is the restriction matrix induced by the embedding @xmath20 . by lemma [ lemma : compactop ] , @xmath285 is trace - class and so @xmath200 is a hilbert - schmidt operator . when @xmath286 is positive semi - definite ,
that is , has no negative eigenvalues , we can always write @xmath287 when everything is scaled appropriately so that @xmath8 is trace - class , then @xmath1 is hilbert - schmidt . we now prove the real - analytic version of the main theorem on rank of hermitian forms for positive semi - definite forms .
the proof reduces to the algebraic version .
however , note that a key point here is that the rank of @xmath7 is not required to be finite . as before a generic subset
@xmath288 is a subset not contained in any proper complex algebraic subvariety of @xmath11 , that is , no nontrivial polynomial vanishes on @xmath215 .
[ lemma : finiterankrespos ] let @xmath3 be a positive semi - definite real - analytic function whose complexified power series converges in a neighborhood of @xmath233 .
let @xmath25 , and let @xmath288 be a generic subset such that all the corresponding manifolds intersect @xmath231 .
further suppose that @xmath289 then @xmath290 is finite and moreover there exists a function @xmath216 such that @xmath217 for an affine manifold @xmath16 that passes through the unit polydisc , we can find an embedding @xmath291 such that @xmath20 takes the closed unit polydisc @xmath276 to a subset of @xmath292 .
the induced restriction matrix @xmath200 is a hilbert - schmidt operator by lemma [ lemma : restcompact ] . as before
, we assume that @xmath16 is a graph over the first @xmath10 variables and we assume that it contains a point of the set @xmath293 .
so assume that @xmath215 only contains such @xmath16 .
as in the proof of theorem [ finiterankres ] we identify @xmath215 with a subset @xmath294 . for such @xmath16 we can simply take a dilation of the first @xmath10 variables to obtain an embedding @xmath20 that maps the closed unit polydisc @xmath295 into @xmath281 , to ensure that @xmath200 is a hilbert - schmidt operator .
without loss of generality , we can also assume that the dilation is the same for all @xmath224 .
therefore we can again assume that entries of @xmath200 depend holomorphically on the entries of @xmath296 .
let @xmath297 .
write @xmath298 thus for @xmath224 we have @xmath299 as for the finite case , this condition is defined by the vanishing of polynomials in the entries of @xmath300 , these depend holomorphically on the entries of @xmath201 . thus is true for all affine manifolds of dimension @xmath10 that intersect @xmath231 . unless @xmath301 we again write @xmath206 for a hyperplane @xmath122 in @xmath207 , where @xmath207 is an affine manifold of dimension @xmath302 .
hence if we prove the result of the theorem for hyperplanes , the general case follows by induction as before .
therefore assume that @xmath303 for all hyperplanes @xmath122 that intersect @xmath231 .
then we have that @xmath304 for all such hyperplanes @xmath122 .
let us suppose that @xmath305 is the hyperplane given by the vector @xmath306 as before . we take @xmath307 and look at the submatrix of rows and columns corresponding to monomials of degree @xmath90 or less , for some integer @xmath90 .
then we see that @xmath308 where @xmath309 , where @xmath118 is given by definition [ defoftnd ] and @xmath310 is an invertible diagonal matrix ( coming from the dilation used above ) .
let us first suppose that @xmath1 is of finite rank @xmath81 .
we can therefore assume that @xmath1 is a @xmath311 matrix .
we have that @xmath304 for all @xmath122 .
write @xmath312 $ ] , then @xmath313 .\ ] ] therefore , @xmath314 for all @xmath306 . as @xmath1 is of finite rank then for some large @xmath90
, @xmath315 contains enough columns such that @xmath316 .
we have therefore reduced the proof of the analytic case to the same matrix @xmath118 and to finish the proof for @xmath1 of finite rank @xmath81 , it is enough to apply lemma [ lemma : green ] .
that is , the rank of @xmath1 and therefore @xmath7 are bounded by @xmath317 .
now let us drop the assumption that the rank of @xmath1 is finite .
let @xmath318 be an infinite by infinite matrix consisting of the first @xmath319 rows of @xmath1 followed by all zero rows .
obviously @xmath318 is of finite rank .
we have that @xmath320 it follows that @xmath321 for all @xmath319 . from the proof of lemma [ lemma : compactop ] we note that the entries of @xmath322 are absolutely summable and hence the entries of @xmath1 are square summable .
therefore @xmath318 converges to @xmath1 in the hilbert - schmidt norm .
then as the rank of @xmath318 is uniformly bounded by @xmath317 , we obtain @xmath323 to see this fact , simply pick a finite orthonormal set @xmath324 in the range of @xmath1 and then notice that @xmath325 must converge to @xmath326 as @xmath327 .
thus for large enough @xmath319 , the set @xmath328 must be linearly independent ( and hence of cardinality less than @xmath317 ) .
[ example : posnes ] let us show that positivity of @xmath7 is necessary .
first note that the set @xmath215 of affine complex manifolds of dimension 1 that lie in @xmath329 forms a generic set in @xmath330 .
see section [ section : crmaps ] for a proof of this fact .
let @xmath331 then for @xmath332 @xmath333 therefore the rank of @xmath334 for @xmath224 is always 1 .
however the rank of @xmath7 is @xmath335 . by adding a function of infinite rank to @xmath7 that vanishes on @xmath336
we can obtain an example of @xmath7 with rank of @xmath334 bounded but rank of @xmath7 infinite .
we can now prove theorem [ thm : rankres ] .
that is , we can drop positivity of @xmath7 if we assume that the rank is bounded on all hyperplanes .
the proof is very similar to the proof of lemma [ lemma : hyperplanesalg ] .
suppose @xmath7 is a nonzero hermitian form .
we can assume that @xmath7 converges in a neighbourhood of the closed unit polydisc as above .
write @xmath337 .
it is again enough to prove the theorem for hyperplanes . restricting @xmath7 to a hyperplane @xmath122 that intersects the closed unit polydisc ,
we have @xmath338 let @xmath178 where @xmath122 ranges over all complex hyperplanes .
as before , the set of hyperplanes @xmath113 is open and dense in the set of all hyperplanes , so @xmath179 } { ( t^d_{\vec{c}})^*ct^d_{\vec{c}}}.\ ] ] again we polarize and apply the semi - definite result twice .
let @xmath180 , @xmath181 , and let @xmath339 } d$ ] .
note that we still allow the ranks to be infinite . applying the bound from the proof of lemma [ lemma : finiterankrespos ]
we obtain @xmath340 for a generic choice of @xmath184 , we have @xmath341 likewise , polarizing and taking generic choices of @xmath186 , we have @xmath187 } { ( t_{\vec{c}})^*ct_{\vec{c } } } = { \operatorname{rank}}_{{{\mathbb{c}}}[c_1,\ldots , c_n ] } { d^{*}t_{\vec{c}}}\ ] ] again applying the bound we get @xmath188 now we note that as @xmath66 is finite , then @xmath342 is finite and hence @xmath343 . thus defining @xmath191 as in the proof of lemma [
lemma : hyperplanesalg ] completes the proof .
in this section we apply theorem [ thm : rankres ] on bounding the rank of hermitian forms by the rank of restrictions to prove theorem [ thm : rigidity ] on rigidity of cr mappings between hyperquadrics .
first let us note how to construct mappings from a real - analytic function @xmath3 vanishing on the hyperquadric .
let @xmath344 be a point and let @xmath345 be a connected open set with @xmath346 .
let @xmath347 be nonzero real - analytic function of finite rank with @xmath1 positive and @xmath2 negative eigenvalues such that @xmath7 vanishes on @xmath348 .
then there exists a point @xmath349 arbitrarily close to @xmath350 and a neighbourhood @xmath41 of @xmath349 , and a real - analytic cr mapping @xmath351 whose image is not contained in a hyperplane .
write @xmath352 with @xmath250 and @xmath254 linearly independent .
the mapping @xmath353 is a meromorphic mapping taking @xmath42 to @xmath354 . if @xmath355 we are finished .
otherwise note that the set @xmath356 is nowhere dense in @xmath42 and hence there is a point @xmath357 arbitrarily close where @xmath358 and a holomorphic mapping defined near @xmath349 taking @xmath42 to @xmath354 . therefore if @xmath7 vanishes on @xmath42 and has signature pair @xmath359 then @xmath7 induces a mapping from @xmath42 to @xmath354 .
let us prove a lemma about signature pairs of functions vanishing on @xmath42 before proving theorem [ thm : rigidity ] .
[ lemma : rigidsigpair ] let @xmath360 , @xmath361 , and @xmath45 .
let @xmath345 be a connected open set such that @xmath362 is nonempty .
let @xmath347 be a real - analytic function vanishing on @xmath362 with signature pair @xmath359 . then if @xmath363 we have @xmath364 where @xmath365 is a constant depending only @xmath70 , @xmath71 , and @xmath2 .
write @xmath366 such that the components of @xmath53 and @xmath367 are linearly independent .
let us suppose that @xmath2 is finite , but for now we allow @xmath53 to map into an infinite dimensional hilbert space .
after an affine change of coordinates we can assume that we are working in a neighborhood of the origin and that @xmath53 and @xmath367 converge in a neighborhood of a closed unit polydisc @xmath259 .
let @xmath368 and @xmath369 correspond to @xmath42 and @xmath370 after this linear change of coordinates .
let @xmath230 be the infinite vector of all monomials in @xmath371 .
we can then find an @xmath372 matrix @xmath373 such that @xmath374 .
similarly we have a @xmath375 matrix @xmath159 such that @xmath376 . we know
that @xmath377 converges in the closed unit polydisc , and so by lemma [ lemma : compactop ] @xmath373 defines a hilbert - schmidt operator ( @xmath378 is trace - class ) on @xmath227 .
similarly @xmath159 also defines a hilbert - schmidt operator . for @xmath379
we have @xmath380 as @xmath53 and @xmath367 are linearly independent we have @xmath381 let @xmath16 be an affine complex submanifold of @xmath368 .
taking the right embedding of @xmath16 and applying lemma [ lemma : restcompact ] , we find that the restriction matrix @xmath200 is a hilbert - schmidt operator . restricted to @xmath16 we have
that @xmath382 thus have that @xmath383 .
in other words , rank of @xmath384 is equal to the rank of @xmath385 .
the rank of @xmath384 is at most @xmath386 .
that means that the rank of @xmath385 is at most @xmath386 for all @xmath16 that lie in @xmath368 .
let @xmath387 be coordinates such that @xmath42 is defined by @xmath388 .
we assume that @xmath45 .
the defining function for @xmath16 that lie in @xmath42 is @xmath389 where @xmath390 ( an @xmath391 matrix ) has orthonormal rows .
let us denote by @xmath392 the affine manifold defined by such a @xmath201 .
we claim that the set @xmath393 of matrices with orthonormal columns is not contained in any proper complex algebraic subvariety of @xmath394 .
suppose that there was a polynomial @xmath350 in @xmath395 variables that vanishes on @xmath396 .
if we multiply the @xmath66-th column by @xmath397 we note that @xmath350 still has to vanish . by the uniqueness theorem for fourier series we note that there has to be a polynomial that is independently homogeneous in entries of the @xmath66-th column ( for all columns ) .
that is , we can multiply each column independently by any complex number and still stay in the zero locus of @xmath350 .
the claim follows by working on each column independently now .
the claim shows that the set @xmath215 of affine manifolds @xmath16 corresponding to @xmath296 is a generic set in @xmath11 . if we let @xmath398 correspond to the set of affine manifolds that lie in @xmath368 .
we see that @xmath399 must be generic as well .
we have all the ingredients to apply lemma [ lemma : finiterankrespos ] to obtain that @xmath400 let us restate theorem [ thm : rigidity ] for reader convenience before proving it .
let @xmath401 , @xmath41 be a connected open set , and @xmath40 be a real - analytic cr mapping such that @xmath59 does not lie in a complex hyperplane then @xmath68 where @xmath69 is a constant depending only @xmath70 , @xmath71 , and @xmath2 .
let @xmath402 denote a real - analytic cr mapping of @xmath41 to @xmath0 , where @xmath403 and @xmath404 . plugging into the defining equation of the target hyperquadric
we obtain that for @xmath405 @xmath406 real - analytic cr mappings extend to holomorphic mappings on a neighborhood of @xmath18 .
therefore we can assume that @xmath53 and @xmath367 are holomorphic in a neighborhood of @xmath18 .
we let @xmath407 .
the condition that @xmath408 does not lie in a complex hyperplane is simply stating that components of the mapping @xmath409 are linearly independent .
in particular we see that @xmath7 has rank @xmath410 , and signature pair @xmath411 .
we can now apply lemma [ lemma : rigidsigpair ] to @xmath7 to obtain @xmath412
forstneri @xcite has shown that there exist strictly pseudoconvex real - analytic compact hypersurfaces that can not be embedded into a sphere of any finite dimension . on the other hand
, lempert @xcite has shown that a mapping exists if one takes the sphere in the infinite dimensional hilbert space @xmath227 .
let us make the following natural definitions to extend hyperquadrics to @xmath227 : @xmath413 we write @xmath414 for the unit sphere in @xmath227 . by a real - analytic cr mapping from a cr manifold @xmath415 to @xmath416 we mean a holomorphic mapping to @xmath227 defined on a neighborhood of @xmath415 taking @xmath415 to @xmath416 .
we can use the ideas from the previous section to construct mappings .
for example , consider the hermitian form @xmath417 the function @xmath7 vanishes precisely on the unit sphere in @xmath418 .
in fact @xmath419 is 1 on the unit sphere and is positive semi - definite .
hence @xmath7 has signature pair @xmath420 ; that is , @xmath7 has infinitely positive eigenvalues and exactly one negative eigenvalue corresponding to the @xmath421 .
therefore the form induces a real - analytic cr mapping @xmath422 in particular we expand the series : @xmath423 therefore @xmath424 note that the image is not contained in a hyperplane ; this fact can even be seen directly by uniqueness of power series . by taking @xmath425 for an arbitrary holomorphic function @xmath426
, we obtain a new form with signature pair @xmath427 , and therefore a cr mapping @xmath428 , whose image is not contained in a hyperplane .
in fact , by repeating the procedure we construct mappings @xmath429 for any finite @xmath71 . finally , by considering the form @xmath430 we obtain a mapping from @xmath431 to @xmath432 . as in the finite dimensional case ,
the situation changes when the source is a hyperquadric not equivalent to a sphere . if we now apply lemma [ lemma : rigidsigpair ] , we notice the following immediate consequence .
let @xmath360 , @xmath361 .
let @xmath41 be a connected open set and @xmath433 , where @xmath83 , be a real - analytic cr mapping such that @xmath59 is not contained in any complex hyperplane of @xmath227 .
then @xmath85 .
therefore , there is no mapping from @xmath42 for finite @xmath70 and @xmath71 into a @xmath82 for finite @xmath2 .
this rigidity result does not generalize to other hypersurfaces with indefinite nondegenerate levi - form . in this case
we can in fact construct mappings to @xmath82 for finite @xmath2 .
for example , define @xmath7 as @xmath434 by expanding the series we see that the signature pair of @xmath7 is @xmath427 .
furthermore , we look at the series up to the quadratic terms @xmath435 we see that the levi form of @xmath436 at the origin is indefinite ; that is , it has one positive and one negative eigenvalue . as the signature pair is @xmath427
we obtain a real - analytic cr mapping @xmath437 whose image is not contained in a hyperplane .
similarly we can easily obtain a submanifold with @xmath71 negative eigenvalues and arbitrary number of positive eigenvalues in the levi - form that admits a real - analytic cr mapping to @xmath416 whose image is not contained in a hyperplane .
the same construction can be used in the finite dimensional case . by truncating the series expansion of @xmath7
, we obtain a polynomial @xmath438 and a manifold @xmath439 with an indefinite levi - form at @xmath57 , and such that @xmath438 has signature pair @xmath440 for an arbitrarily high finite @xmath1 .
so we obtain a mapping from @xmath441 to @xmath442 whose image is not contained in a hyperplane .
let us give some explicit examples in nonhomogeneous coordinates before we give a general method for constructing mappings .
let @xmath443 .
we tensor with the identity ( which we write as @xmath444 ) on the @xmath445 variable .
that is , we construct the mapping @xmath446 we reorder the components to obtain a mapping taking @xmath42 to @xmath447 .
that is , @xmath448 therefore the super - rigidity result of baouendi , ebenfelt , and huang @xcite is sharp . in the same language as in the introduction we have that @xmath449 and @xmath450 . on the other hand , by theorem [ thm : rigidity ] we are not able to make @xmath1 arbitrarily large while keeping @xmath450 .
we could repeat the procedure of the first example any number of times .
we take a mapping @xmath451 and construct @xmath452 again we reorder the components to obtain @xmath453 taking @xmath42 to @xmath454 . for the constructions , it is easier to work in homogeneous coordinates .
we define the homogeneous hyperquadric @xmath455 the set @xmath456 defines a real hypersurface in the projective space @xmath457 . by setting @xmath458
we obtain @xmath459 in @xmath460 .
the homogeneous setting is more symmetric , although one has to be careful when converting back to the nonhomogeneous situation .
suppose that @xmath3 is a bihomogeneous polynomial and vanishes on @xmath456 .
if @xmath7 has signature pair @xmath359 ( that is @xmath1 positive and @xmath2 negative eigenvalues ) , then we find homogeneous polynomials @xmath461 and @xmath462 with linearly independent components such that @xmath463 therefore the mapping @xmath464 takes @xmath456 to @xmath465 . by dividing by @xmath466 and setting @xmath458
we obtain a rational cr mapping @xmath467 .
if the components of @xmath468 are linearly independent then the image of @xmath373 does not lie in a complex hyperplane .
therefore when constructing examples we will consider homogenous polynomial mappings from @xmath456 to @xmath465 with linearly independent components .
we wish to show that as long as the number of positive and negative eigenvalues of the target hyperquadric grow in a comparable way , then nontrivial hyperquadric mappings always exist .
we have the following stability result .
we make the normalization @xmath469 for @xmath456 .
[ thm : stability ] fix integers @xmath470 .
there exists an @xmath415 ( @xmath471 suffices ) such that when @xmath472 , where @xmath473 and @xmath474 then there exists a homogeneous polynomial mapping @xmath475 with linearly independent components .
that is , there exists a rational cr mapping @xmath476 whose image is not contained in a complex hyperplane .
we note that in @xcite it is proved that when the source is a sphere ( that is , @xmath477 ) then such a rational mapping always exists for large @xmath478 provided only that @xmath479 . in the hyperquadric case , this sort of stability result holds in a sector , where the size of the sector depends on @xmath70 and @xmath71
. there may exist mappings outside of the sector given above ; however , by theorem [ thm : rankres ] it is clear that any sector in which existence holds must make an acute angle . for the source hyperquadric @xmath480 .
the lattice represents the positive quadrant of the @xmath359 plane .
a black dot represents that a mapping to @xmath465 exists by applying the theorem .
a black square represents that a mapping was also constructed in the proof .
a circle represents a position in the lattice for which the theorem tells us nothing .
the three lines represent the limits of the stability region from the theorem .
the dashed squares represent one succession of the `` squares '' constructed in the proof .
[ fig : stability ] ] suppose that we can find a homogeneous polynomial @xmath481 in the variables @xmath482 with @xmath1 positive and @xmath2 negative coefficients such that @xmath483 when @xmath484 .
we call such polynomials admissible for @xmath359
. we define @xmath485 if @xmath350 is admissible for @xmath359 then we let @xmath486 in @xmath350 .
we obtain a hermitian symmetric polynomial @xmath487 as distinct monomials are linearly independent , the rank of @xmath7 is equal to the number of distinct monomials of @xmath350 , that is @xmath478 and the signature pair is @xmath359 . as @xmath7 vanishes on @xmath488 , we have that @xmath7 induces a cr mapping of @xmath456 to @xmath465 such that the image is not contained in a complex hyperplane .
we therefore work with admissible polynomials @xmath350 as above .
we fix @xmath70 and @xmath71 and we work in the plane @xmath489 and simply say @xmath490 whenever an admissible @xmath350 exists for those @xmath359 ( a mapping to @xmath465 exists ) .
we note that @xmath491 and @xmath492 . if @xmath350 is admissible for @xmath359 of degree @xmath10 then @xmath493 vanishes when @xmath494 vanishes .
if we pick @xmath66 large enough then the monomials of @xmath495 and @xmath496 are distinct and @xmath497 is admissible for @xmath498 . by using @xmath499 instead of @xmath494
we build an admissible polynomial for @xmath500 .
therefore for any @xmath501 we have that @xmath502 therefore we get a lattice of admissible points .
we claim that for each point @xmath503 we also obtain that the `` square '' @xmath504 is a subset of @xmath20 .
once we have the claim , it is a simple matter of noticing that the squares must overlap for large enough @xmath505 . the inequalities from the theorem follow .
see figure [ fig : stability ] for illustration of the proof .
the first point we have already seen belongs to @xmath20 .
now take @xmath350 that is admissible for @xmath359 . by dividing through by @xmath508
we can suppose that @xmath350 contains a monomial @xmath10 that does not depend on @xmath508 .
furthermore we can assume that @xmath10 has a positive coefficient , since if no such monomial existed , setting @xmath509 would obtain a contradiction .
write @xmath510 and construct @xmath511 where we note that @xmath512 .
the polynomial @xmath497 vanishes when @xmath494 vanishes .
as the coefficient of @xmath10 is positive then @xmath497 is admissible for have @xmath513 . | using green s hyperplane restriction theorem , we prove that the rank of a hermitian form on the space of holomorphic polynomials can be bounded by a constant depending only on the maximum rank of the form restricted to affine manifolds . as an application
we prove a rigidity theorem for cr mappings between hyperquadrics in the spirit of the results of baouendi - huang and baouendi - ebenfelt - huang .
if we have a real - analytic cr mapping of a hyperquadric not equivalent to a sphere to another hyperquadric @xmath0 , then either the image of the mapping is contained in a complex affine subspace or @xmath1 is bounded by a constant depending only on @xmath2 . finally , we prove a stability result about existence of nontrivial cr mappings of hyperquadrics .
that is , as long as both @xmath1 and @xmath2 are sufficiently large and comparable , then there exist cr mappings whose image is not contained in a hyperplane .
the rigidity result also extends when mapping to hyperquadrics in infinite dimensional hilbert - space . |
the study of vertex separators and associated combinatorial problems has been a fascinating research in the field of combinatorial computing .
one such classical problem , namely _ connectivity augmentation _ focuses on increasing the vertex connectivity by one by augmenting a minimum number of edges .
this study was initiated by eswaran et al @xcite and they studied the fundamental problem _ bi - connectivity augmentation _ : given a 1-connected graph @xmath2 , find a minimum number of edges whose augmentation to @xmath2 makes it 2-vertex connected ( bi - vertex connected ) . subsequently , hsu @xcite studied the tri - connectivity augmentation of bi - vertex connected graphs , making a bi - vertex connected graph 3-vertex connected by augmenting a minimum number of edges .
the complexity of general vertex connectivity augmentation , i.e. , given a @xmath0-vertex connected graph , find a minimum number of edges whose augmentation to the given graph makes it @xmath3-vertex connected , was settled by vegh @xcite and this result has been a breakthrough result as the complexity of which was open for al most three decades .
+ there are some note - worthy results as far as bi ( tri)-connectivity augmentation problems are concerned . to solve bi - connectivity augmentation , in @xcite , the given 1-vertex connected graph is transformed into a _ block tree _ and the augmentation is done with the help of the block tree . in @xcite ,
a bi - connected component tree transforms the 1-connected graph and helps in obtaining a biconnectivity augmentation set .
interestingly , in both the approaches a minimum bi - vertex connectivity augmentation can be obtained with the help of proposed tree - like graphs .
recent work due to surabhi et .
al @xcite developed a strategy using which one can augment many edges in parallel , thus obtaining a simpler approach for sequential and parallel bi - vertex connectivity augmentation .
however , their work is restricted to the class of trees which are 1-vertex connected . similarly , for tri - connectivity augmentation of 2-vertex connected graphs ,
a 3-block tree of 2-vertex connected graphs was used to obtain a minimum tri - connectivity augmentation set @xcite .
+ in this article , we initiate the study of @xmath4-vertex connectivity augmentation which is to find a @xmath5-vertex connected graph from a @xmath0-vertex connected graph by augmenting a minimum number of edges .
having known the results of @xcite , it is natural to know whether iterative application of algorithm mentioned in @xcite @xmath4 times will yield a @xmath5-vertex connected graph
. it is important to highlight the fact that even for tri - connectivity augmentation of 1-connected graphs the chain approach fails to produce a minimum connectivity augmentation set , i.e , the approach of making a 1-vertex connected graph 2-vertex connected and making a 2-vertex connected graph 3-vertex connected does not yield an optimum augmentation set .
this calls for a good understanding of minimum vertex separators of @xmath0-vertex connected graphs and their role in @xmath4-vertex connectivity augmentation . towards this attempt
, we shall explore the study of tri - connectivity augmentation in trees .
in particular , we present the following results in this paper ; * given a tree @xmath11 , any minimum tri - connectivity augmentation set has at least @xmath12 edges , where @xmath9 and @xmath10 denote the number of degree one vertices and degree two vertices , respectively . * a polynomial - time algorithm to compute a minimum tri - connectivity augmentation set meeting the above bound .
we believe that the results presented in this paper can be extended to tri - connectivity augmentation of 1-connected graphs .
+ * roadmap : * in the next section , we present lower bound results for tri - connectivity augmentation followed by an algorithm which will yield the minimum tri - connectivity augmentation set in polynomial time .
we conclude this paper with some directions for @xmath4-connectivity augmentation of 1-connected graphs , @xmath13 .
notation and definitions are as per @xcite .
let @xmath14 be an undirected connected graph where @xmath15 is the set of vertices and @xmath16 , @xmath17
. for @xmath18 , @xmath19 and @xmath20 refers to the degree of @xmath21 in @xmath2 .
let @xmath18 is said to be a _ leaf _ if @xmath22 .
@xmath23 and @xmath24 refers to the minimum and maximum degree of @xmath2 , respectively . for simplicity , we use @xmath25 and @xmath26 when the associated graph is clear from the context .
@xmath27 is a _ path _ defined on @xmath28 such that @xmath29 . for a graph @xmath2 , we define @xmath30 such that @xmath31 refers to the number of vertices in @xmath32 and we define @xmath33 such that @xmath34 refers to the number of vertices in @xmath35 . for @xmath36 ,
@xmath37 $ ] denotes the graph induced on the set @xmath38 and @xmath39 is the induced graph on the vertex set @xmath40 .
vertex separator _ of a graph @xmath2 is a set @xmath41 such that @xmath39 has more than one connected component . a vertex separator @xmath38 is said to be _ minimal _ if there no proper subset @xmath42 of @xmath38 such that @xmath42 is a vertex separator .
a _ minimum vertex separator _
@xmath38 is a vertex separator of least size and the cardinality of such @xmath38 is the vertex connectivity of a graph @xmath2 , written @xmath43 .
a graph is @xmath0-_vertex connected _ if @xmath44 .
if @xmath45 then the graph is 1-connected ( also known as singly connected ) and in such a graph a minimum vertex separator @xmath38 is a singleton set and the vertex @xmath46 is a _ cut - vertex _ of @xmath2 .
cycle _ is a connected graph in which the degree of each vertex is two .
a _ tree _ is a connected and an acyclic graph .
for a graph @xmath2 with @xmath44 , a minimum connectivity augmentation set @xmath47 is such that the graph obtained from @xmath2 by augmenting @xmath48 is of vertex connectivity @xmath49 , @xmath50 .
this paper is written in the context of augmenting @xmath48 edges to a tree such that the obtained graph is @xmath51-connected .
in this section , we shall first present the lower bound analysis which is a number representing the number of edges to be augmented in any minimum connectivity augmentation set to make a tree 3-vertex connected . in the subsequent sections , we first give an sketch of the algorithm and
then we shall present an algorithm with analysis which will output a connectivity augmentation set meeting the lower bound .
our approach finds a minimum tri - connectivity augmentation set for trees with @xmath52 ( which are called paths ) and for trees with @xmath53 ( called non - path trees ) separately .
let @xmath11 be a tree and @xmath9 and @xmath10 denote the number of degree one and degree two vertices , respectively .
then , any tri - connectivity augmentation set @xmath48 is such that @xmath54 .
it is well - known that for any @xmath51-connected graph @xmath2 , @xmath55 .
therefore to make @xmath11 a @xmath51-connected graph , we must increase the degree of elements in @xmath32 by at least two and the degree of elements in @xmath35 by at least one . since an edge joins a pair of vertices , any augmentation set @xmath48 has at least @xmath56 edges .
this completes the proof of the lemma . @xmath57
our approach varies for the path and the non - path , for the input tree on @xmath58 vertices .
if the input is a path , the algorithm converts the path to a cycle and then augments edges in such a way that every edge creates a cycle of length @xmath59 .
if the input is a non - path tree , @xmath11 : first root the tree @xmath11 at the maximum degree vertex , @xmath4 .
let @xmath60 denotes the set of leaves in @xmath11 .
now , we group the vertex set into branches , namely @xmath61 .
the branch @xmath61 contains the vertices in the path from the root @xmath4 to the leaf @xmath62 .
thus , the number of branches in the input tree is the number of degree one vertices .
next we perform a level ordering and label the degree two vertices as per the ordering as @xmath63 .
let @xmath64 .
now we initialize every vertex in @xmath65 as unmarked .
as we iterate , we augment edges as follows : for every unmarked vertex @xmath66 find the least unmarked vertex @xmath67 of different branch and if such @xmath67 exists for @xmath68 , mark the vertices @xmath67 and @xmath68 and augment an edge between @xmath67 and @xmath68 .
once this process is done , group the unmarked vertices in @xmath65 .
if there are no unmarked vertices then we form a cycle among the degree one vertices , if there are odd number of unmarked vertices we form a cycle among the degree one vertices and then we augment edges between the remaining degree two vertices using the ordering of vertices . if there are even number of unmarked vertices we form a path among the leaves and then we augment edges between the remaining degree two vertices using the ordering of vertices .
interestingly , this new approach guarantees that the algorithm augments exactly @xmath56 edges .
we now present an algorithm for tri - connectivity augmentation of trees .
further , we show that our algorithm is optimal followed by the proof of correctness .
let t be a tree with @xmath69 vertices .
algorithm @xmath70 yields a graph @xmath71 , where @xmath72 .
the algorithm , first converts the path @xmath73 to a cycle @xmath74 , by adding an edge between two end vertices , i.e. , the algorithm augments an edge @xmath75 in _
step 2_. now , the degree of each vertex in the resultant graph is two . in
_ steps 3 - 8 _ , for each vertex @xmath62 of degree two in the set @xmath76 , we identify a vertex @xmath77 such that the length of the path @xmath78 and further , we augment an edge between @xmath62 and @xmath77 . thus , in the resulting graph , degree of every vertex is three if @xmath58 is even and degree of all vertices other than the vertex @xmath79 is three , if @xmath58 is odd .
so , if @xmath58 is odd , the algorithm augments an edge @xmath80 in _ step 9 - 11_. this completes the path augmentation and in the resulting graph @xmath71 degree of each vertex is at least three .
@xmath57 let t be a tree with @xmath69 vertices .
algorithm @xmath81-@xmath82 yields a graph @xmath71 , where @xmath72 .
the algorithm , collects all degree two vertices and augment edges between those vertices which satisfies the condition in _ step 11 _ and marks the end vertices of the augmented edges .
now , the marked vertices are of degree three .
collect the unmarked vertices ( remaining vertices of degree 2 ) into the set @xmath83 .
we shall now analyze the _ steps 18 - 37 _ of the algorithm by considering the following cases .
@xmath84 + i.e. , all the degree two vertices in the given tree have become the degree three vertices in @xmath71 .
we now augment edges among the leaves such that there is a cycle @xmath85 .
we can easily see that in the resultant graph @xmath71 , for every vertex @xmath86 , @xmath87 .
@xmath88 and @xmath89 is even , say @xmath90 .
+ we first form a path among leaves from @xmath91 to @xmath92 such that all the degree one vertices are converted to degree three vertices except @xmath91 and @xmath92 , which is of degree two .
now , augment the edges @xmath93 , @xmath94
, @xmath95,@xmath96 .
thus , the only remaining degree two vertices are @xmath97 , @xmath98 , @xmath91 and @xmath92 .
therefore , if @xmath99 , then augment @xmath100 and @xmath101 and if @xmath102 , then augment @xmath103 and @xmath104 .
hence in the resultant graph @xmath71 every vertex is of degree three .
@xmath88 and @xmath89 is odd , say @xmath90 .
+ we first form a cycle among leaves such that all the degree one vertices are converted to degree three vertices .
now , augment the edges @xmath105 , @xmath106 ,
thus , the only remaining degree two vertex is @xmath109 .
therefore , if @xmath110 , then augment @xmath111 such that @xmath112 and @xmath113 and if @xmath114 , then augment @xmath115 such that @xmath112 and @xmath116 . hence in the resultant graph @xmath71
every vertex is of degree at least three .
@xmath57 let t be a tree with @xmath69 vertices .
algorithm @xmath70 precisely augments @xmath117 edges .
_ step 2 _ of algorithm augments an edge between two leaves and this increases @xmath10 by two .
thus there are @xmath118 degree two vertices and _ steps 3 - 8 _ augments @xmath119 new edges , if @xmath58 is even and @xmath120 new edges , if @xmath58 is odd . if @xmath58 is odd , _ steps 9 - 12 _ augments an edge .
thus , if @xmath58 is odd , we have augmented @xmath121 ( since , @xmath10 is odd ) .
if @xmath58 is even , we have augmented @xmath122 edges . in total , since @xmath123 , the algorithm augments @xmath124 .
therefore , the algorithm augments @xmath56 edges in total .
@xmath57 let t be a tree with @xmath69 vertices . algorithm @xmath81-@xmath70 precisely augments @xmath125 edges .
we present a proof by case analysis based on the cardinality of the set @xmath83 generated by _
algorithm 3 _ in _
step 17_. case 1 : : : @xmath126 + @xmath9 edges are augmented in _
step 19 _ by forming a cycle among leaves . in _
steps 10 - 16 _ , we augment edges between the degree two vertices and since @xmath127 , @xmath10 is even and @xmath128 edges are augmented . in total , we have augmented @xmath129 edges , i.e. , we have augmented @xmath130 edges .
case 2 : : : @xmath131 .
+ let @xmath132 . among @xmath10 degree
two vertices , degree of @xmath133 vertices increases by one in _ steps 10 - 16 _ , i.e. , @xmath134 edges are augmented .
note that @xmath135 edges are augmented in _ step 21 _ , @xmath136 edges are augmented in _
, @xmath137 edges are augmented in _ steps 23 - 27_. in total , we have augmented @xmath138 edges . since @xmath10 is even , @xmath139 . therefore , @xmath140 edges are augmented .
case 3 : : : @xmath141 .
+ let @xmath132 . among @xmath10 degree
two vertices , degree of @xmath133 vertices increases by one in _ steps 10 - 16 _ i.e. , @xmath134 edges are augmented in _ steps 10 - 16_. @xmath142 edges are augmented in _
step 29 _ , @xmath9 edges are augmented in _
step 30 - 31 _ and a edge is augmented in _
steps 32 - 36_. in total , we have augmented @xmath143 edges .
since @xmath144 is odd and @xmath133 is even , @xmath145 edges are augmented in total .
thus , the algorithm augments @xmath56 edges .
@xmath57 for a tree @xmath11 , the graph obtained from the algorithm @xmath70 is 3-connected .
our claim is to prove that every minimal vertex separator is of size at least @xmath51 and there exist at least one vertex of degree @xmath51 . on the contrary ,
assume that there exist at least one minimal vertex separator of size at most 2 , say @xmath146 .
let @xmath147 be the cycle formed in _
step 2 _ of _ algorithm 2 _ , where @xmath148 .
case 1 : : : @xmath149 .
+ for every vertex @xmath150 , @xmath151 .
hence , @xmath152 is connected .
case 2 : : : @xmath153 .
let @xmath154 and @xmath155 .
+ case 2.1 : ; ; @xmath156 . since @xmath157 is a cycle ,
@xmath152 is connected .
case 2.2 : ; ; @xmath158 or @xmath159 + for every internal vertex @xmath160 in the path @xmath161 there exist a vertex @xmath162 such that @xmath163 and @xmath164 by _ steps 3 - 12 _ of _ algorithm 2_. thus , @xmath152 is connected . in all the above cases , the graph @xmath152 is connected , which is a contradiction to the assumption that @xmath38 is a vertex separator .
therefore , every minimal vertex separator of @xmath71 is of size at least 3 .
note that by our augmentation procedure , @xmath165 . clearly , @xmath166 is a minimal vertex separator of size three .
thus , the graph @xmath71 is 3-connected .
@xmath57 for a tree @xmath11 , the graph obtained from the algorithm @xmath81-@xmath70 is 3-connected .
it is enough to prove that the size of every minimal vertex separator is at least 3 and there exists at least one minimal vertex separator of size 3 . on the contrary ,
assume that there exist at least one minimal vertex separator @xmath38 such that @xmath167 .
let @xmath168 be the path formed in _
steps 18 - 37 _ of _ algorithm 3 _ , where @xmath169 .
case 1 : : : @xmath170 .
let @xmath171 .
the vertex @xmath172 can be a root node , @xmath4 , or a node in the path , @xmath173 , or neither . by case analysis
, we prove that the graph @xmath152 is connected , which forms a contradiction to the definition of @xmath38 .
+ + case 1.1 : ; ; if @xmath174 , then since @xmath173 is the path @xmath175 is connected .
case 1.2 : ; ; if @xmath176 , then , since @xmath177 , @xmath178 is connected , hence @xmath175 is also connected . case 1.3 : ; ; if @xmath179 , then every vertex @xmath180 has a path @xmath181 such that @xmath182 and every vertex @xmath183 , where @xmath184 for some @xmath185 such that @xmath186 exist in @xmath11 , has a path @xmath187 such that @xmath188 . thus , the graph @xmath152 is connected .
case 2 : : : if @xmath189 , say @xmath190 .
@xmath38 can either be a clique or an independent set .
if @xmath38 is a clique then either the edge is from the tree @xmath11 or from the augmentation set @xmath48 . in each case
, we prove that the graph @xmath152 is connected , which is a contradiction to the definition of @xmath38 .
+ case 2.1 : ; ; @xmath191 .
+ * if either @xmath172 or @xmath21 is a root node . without loss of generality ,
let @xmath172 be the root node . by the path @xmath173 in @xmath71 ,
every pair of vertex @xmath192 has a path connecting them in @xmath152 .
thus , the graph @xmath152 is connected . * if neither @xmath172 nor @xmath21 is a root node . in @xmath152 ,
for every vertex @xmath193 , there exists a path @xmath181 such that @xmath194 and for every vertex @xmath195 , where @xmath184 for some @xmath185 such that @xmath186 exist in @xmath11 , there exists a path from @xmath187 such that @xmath196 .
since @xmath197 and by the path @xmath173 in @xmath71 , the graph @xmath152 is connected .
case 2.2 : ; ; @xmath198 .
+ * if @xmath199 and @xmath200 then , the graph @xmath201 is connected .
thus , the graph @xmath152 is connected . * if either @xmath199 or @xmath200 .
@xmath202 , assume that @xmath200 . * * if @xmath203 for some @xmath204 .
+ since @xmath205 , every internal vertex of @xmath206 in @xmath11 contributes degree 2 to the path @xmath206 and the remaining degree to the vertices which does not belong to @xmath207 ( by _ steps 10 - 37 _ ) . for every internal vertex @xmath208 of @xmath11 ,
there exists a path from @xmath181 such that @xmath194 . since @xmath197 and by the path @xmath173 , the graph @xmath152 is connected .
* * if @xmath209 and @xmath210 for some @xmath211 and @xmath212 .
+ since @xmath200 , @xmath213 is connected .
for every internal vertex @xmath208 of @xmath11 , there exists a path from @xmath181 such that @xmath194 and for every internal vertex @xmath214 of @xmath11 , where @xmath184 for some @xmath185 such that @xmath186 exist in @xmath11 , there exists a path from @xmath187 .
since @xmath197 and by the path @xmath173 , the graph @xmath152 is connected . * if neither @xmath215 nor @xmath216 . * * if @xmath203 for some @xmath204 .
+ i.e. , the edge @xmath217 is augmented by the _ step 21 _ or the _ step 29_. thus , there exists an internal vertex @xmath218 in @xmath11 and one of the internal vertex in the path @xmath206 is adjacent to a vertex in different branch , by the _
steps 23 - 27 _ or by the _ steps 32 - 36_. for every internal vertex @xmath208 of @xmath11 , there exists a path @xmath181 such that @xmath194 , since if @xmath219 then the vertex @xmath220 is augmented to a vertex in different branch by the least degree two vertex condition in _ step 11 _ and if @xmath221 then there exists a path from @xmath220 to a vertex in different branch in @xmath11 . for every internal vertex @xmath214 of @xmath11 , where @xmath184 for some @xmath185 such that @xmath186 exist in @xmath11 , there exists a path @xmath187 . since @xmath197 and by the path @xmath173 , the graph @xmath152 is connected . * * if @xmath209 and @xmath210 for some @xmath211 and @xmath212 .
+ i.e. , the edge @xmath217 is augmented by the _ steps 10 - 16_. for every internal vertex @xmath208 ( @xmath222 ) , there exists a path @xmath181 ( @xmath223 ) such that @xmath194 ( @xmath224 ) and for every internal vertex @xmath214 ( @xmath225 ) , where @xmath184 for some @xmath185 such that @xmath186 exist in @xmath11 , there exists a path @xmath187 ( @xmath226 ) . since @xmath197 and by the path @xmath173 , the graph @xmath152 is connected .
case 2.3 : ; ; @xmath227 .
+ * if @xmath174 or @xmath228 without loss of generality , assume that @xmath174 . then @xmath229 , @xmath230 .
since , all the degree one vertices are connected by a path @xmath173 , the graph @xmath231 is connected .
every internal vertex @xmath232 of @xmath11 contributes degree 2 to the path @xmath206 and remaining to the vertices which are not in the path @xmath206 .
thus , the graph @xmath152 is connected . * if @xmath203 for some @xmath204 .
similar argument as in _ case 2.2_. * if @xmath209 and @xmath210 for some @xmath211 and @xmath212 . similar argument as in _ case 2.2_. in all the above cases , the graph @xmath152 is connected , which is a contradiction to the fact that @xmath38 is a vertex separator .
therefore , our assumption that there exists at least one minimal vertex separator in @xmath71 of size at most two is wrong .
hence , every minimal vertex separator of @xmath71 is of size at least 3 .
+ now , our claim is to prove that there exists at least one minimal vertex separator of size 3 . by _
lemma 2 _ , the degree of every vertex in @xmath233 is 3 .
the graph @xmath71 is 3-connected .
@xmath57 for a tree @xmath11 , the graph @xmath71 obtained from _ algorithm 1 _ is 3-connected .
further , @xmath71 is obtained from @xmath11 by augmenting a minimum set of edges .
the lower bound for the tri - connectivity augmentation of trees is @xmath56 by _
lemma 1_. if the tree @xmath11 is a path , _ algorithm 1 _ calls _ algorithm 2 _ , which converts the tree to a 3-connected graph @xmath71 by augmenting exactly @xmath56 edges ( by _ lemma 4 _ and _ lemma 6 _ ) .
if the tree @xmath11 is a non - path tree , the _ algorithm 1 _ calls _ algorithm 3 _ , which converts the tree to a 3-connected graph @xmath71 by augmenting exactly @xmath56 edges ( by _ lemma 5 _ and _ lemma 7 _ ) . thus , for a tree @xmath11 the graph obtained from _
algorithm 1 _ is 3-connected .
further , @xmath71 is obtained by using a minimum connectivity augmentation set .
therefore , the claim follows .
@xmath57 let @xmath11 be a tree with the vertex set @xmath234 such that @xmath235 , with the edge set @xmath236 such that @xmath237 and @xmath238 be the number of leaves .
the algorithm @xmath70 takes @xmath239 time in _ step 2 _ , @xmath240 time for _ steps 3 - 8 _ and @xmath239 time for _ steps 9 - 12 _ of _ algorithm 2_. thus , the algorithm @xmath70 takes @xmath240 time . the algorithm @xmath81-@xmath70 : since , the level ordering can be implemented by the data structure @xmath241 , the _ step 3 _ takes @xmath240 time , takes @xmath242 time for _ steps 4 - 7 _ of _ algorithm 3 _ and @xmath240 time for the _ step 8_. we implement the data structure queue for the _ steps 8 - 16 _ , which is used to keep track of marked and unmarked vertices in @xmath65 .
this process ends after visiting all vertices in @xmath11 and it takes @xmath243 time for _ steps 10 - 16 _ and the algorithm takes @xmath240 time for _ steps 18 - 38_. in total , the algorithm @xmath81-@xmath70 takes @xmath243 time .
therefore , _ algorithm 1 _ takes @xmath243 time .
thus , for a given tree , a minimum tri - connectivity augmentation set can be found in @xmath243 time . ] , title="fig : " ] + , title="fig : " ]
in this paper , we have presented an algorithm for finding a minimum tri - connectivity augmentation set in trees .
we believe that the approach can be extended to @xmath244-connected graph with the help of block trees , biconnected component trees proposed in @xcite .
a logical extension of this work would be to look at @xmath4-connectivity augmentation of trees for any @xmath50 . | for a connected graph , a _ minimum vertex separator _ is a minimum set of vertices whose removal creates at least two connected components .
the vertex connectivity of the graph refers to the size of the minimum vertex separator and a graph is @xmath0-vertex connected if its vertex connectivity is @xmath0 , @xmath1 .
given a @xmath0-vertex connected graph @xmath2 , the combinatorial problem _ vertex connectivity augmentation _ asks for a minimum number of edges whose augmentation to @xmath2 makes the resulting graph @xmath3-vertex connected . in this paper
, we initiate the study of @xmath4-vertex connectivity augmentation whose objective is to find a @xmath5-vertex connected graph by augmenting a minimum number of edges to a @xmath0-vertex connected graph , @xmath6 .
we shall investigate this question for the special case when @xmath2 is a tree and @xmath7 .
in particular , we present a polynomial - time algorithm to find a minimum set of edges whose augmentation to a tree makes it 3-vertex connected . using lower bound arguments , we show that any tri - vertex connectivity augmentation of trees requires at least @xmath8 edges , where @xmath9 and @xmath10 denote the number of degree one vertices and degree two vertices , respectively .
further , we establish that our algorithm indeed augments this number , thus yielding an optimum algorithm . |
the night sky brightness , the number of clear nights , the seeing , transparency and photometric stability are some of the most important parameters that qualify a site for front - line ground - based astronomy .
there is limited control over all these parameters , and only in the case of the sky brightness is it possible to keep it at its natural level by preventing light pollution from the immediate vicinity of the observatory .
previous to the installation of any observatory , extensive tests of these parameters are carried out in order to find the best locations , maximizing then the efficiency of these expensive infraestructures . however , most of these parameters are not constant along the time .
an example of this can be seen in the seeing evolution of the paranal observatory , which is worse now than when the decision to built it at its current location was taken .
this is not an untypical situation .
we have started a program to determine the actual values of the main observing conditions for the calar alto observatory .
the calar alto observatory is located at 2168 m height above the sealevel , in the sierra de los filabres ( almeria - spain ) at @xmath045 km from the mediterranean sea .
it is the second largest european astronomical site in the north hemisphere , just behind the observatorio del roque de los muchachos ( la palma ) , and the most important in the continental europe .
currently there are six telescopes located in the complex , three of them operated by the centro astronomico hispano aleman a.i.e .
( csic - mpg ) , including the 3.5 m , the largest telescope in the continental europe .
along its 26 years of operations there has been different attempts to characterize some of the main properties described before : ( i ) leinert et al .
( 1995 ) determined the sky brightness corresponding to the year 1990 ; ( ii ) hopp & fernandez ( 2002 ) studied the extinction curve corresponding to the years 1986 - 2000 ; ( iii ) ziad et al .
( 2005 ) estimated the median seeing in the observatory from a single campain in may 2002 .
however , there is a need for a consistent study of all these properties , spanning over a similar time period . in this article
we study the main characteristics of the night - sky at the observatory including : ( i ) the night - sky spectrum , identifing the natural and light pollution emission lines and their strength , ( ii ) the moonless night - sky brightness in different bands , ( iv ) the extinction and its yearly evolution and ( v ) the atmospheric seeing and its yearly evolution .
the study is limited to the last four years , with mostly corresponds to a period of minimun solar activity ( which strongly affect several sky properties , like night - sky brightness and airglow ) .
the derived main properties have been compared with similar properties at other observatories .
the structure of this article is as follows : in section [ data ] we describe the dataset collected for the current study , including a description of data and the data reduction ; in section [ ana ] we show the analysis performed over the different types of data and the results derived for each one ; in section [ conc ] we summarize the main results and present the conclusions .
in order to understand the properties of the night - sky emission at the calar alto observatory we collected different observational data , including both imaging and spectroscopic data . since none of the data were obtained directly for this study , we scanned thoroughly the arquived data to acquire a data set with a sufficient degree of homogeneity .
spectroscopic data were obtained to determine the mean properties of the night - sky spectrum of moonless nights at the observatory . in calar alto spectrographs
are normally mounted in bright and gray nights , while dark nights are more frequently allocated for deep imaging programs .
thus , it is somehow difficult to find spectrocopic data taken during dark nights .
the most frequently mounted spectrographs at the calar alto observatory are cafos at the 2.2 m ( @xmath050% of the allocated time ) and pmas @xcite at the 3.5 m telescope ( @xmath030% of the allocated time ) .
pmas is an integral field unit with two different setups , a lensarray with a reduced field - of - view ( 16@xmath8@xmath916@xmath8 in its largest configuration ) , and a wide fiber - bundle that covers a field - of - view of 72@xmath8@xmath964@xmath8 with 331 individual fibers of 2.7@xmath8 diameter each one ( ppak , kelz et al .
this latter configuration is particularly interesting to study the properties of the sky emission , since in a single shot centred on a calibration star ( or a science target ) a substantial fraction of the field - of - view samples the sky .
its large aperture , and the possibility of performing self - calibration , allows one to obtain spectrophotometric calibrated and high signal - to - noise spectra of the night - sky emission even with reduced exposure times .
evenmore , this instrument is frequently mounted with a low - resolution grating ( v300 ) , which covers a wavelength range of @xmath03500 with a spectral resolution of @xmath010 ( fwhm ) .
this is also very convenient to obtain spectra of the sky emission in all the optical wavelength range . restricting ourselves to the same instrument and configuration
ensures the homogeneity of the data .
there were 14 clear moonless nights when the instrument was mounted using this configuration in the period between january 2005 and december 2006 .
a sample of 23 observations , including night - sky emission spectra , were selected from the data taken that nights .
the sample was selected by including only observations of calibration stars and/or small - size and faint targets ( eg . , high - z ly-@xmath10 emitters , snchez et al .
2007c ) , with most of the ppak field - of - view sampling sky emission .
in addition , only observations near the zenith , with an airmass lower than 1.5 , were included in the sample . in all the cases
the observations were taken far away the ecliptic and the galactic plane .
the data were then reduced using r3d ( snchez et al .
2006 ) , following the standard steps for fiber - based ifs data .
once reduced , the sky spectra were extracted from the frames using e3d @xcite , by selecting areas clean of objects within the field - of - view .
although the signal - to - noise level of each individual sky spectrum is somehow different , depending on the exposure time and the number of selected fibers to extract the spectra , in all the cases is good enough for the purposes of this study .
not all the observations during a moonless night are equally dark .
the darkness of an observation depends strongly on the time distance of the corresponding night from the full moon , the presence of cirrus , dust and local contamination when pointing towards highly populated areas ( like almeria , towards the south of the observatory ) , the zenith distance , and even more the time distance from the twilight .
thus , from the original dataset we did not consider those spectra which intensity at 5200 was larger than 0.3 10@xmath11 erg s@xmath12 cm@xmath4 @xmath12([email protected] mags ) . we were then left with a final sample of 10 individual spectra obtained in 7 different nights , representative of the typical dark observation in a moonless dark night at calar alto .
this sample still comprises a wide range of brightness , and only one ( 02/06/2005 ) can be considered completely dark , corresponding to a new moon night .
table [ tab_data ] shows the final list of nights , together with the wavelength range covered by the spectra each night .
multiband imaging data are collected to characterize the sky - brightness of moonless nights at the observatory .
the search was restricted to cafos covering a similar period of time of the selected spectroscopic dataset , to preserve the homogeneity of the data .
only exposures on calibration fields were selected to perform a self flux - calibration , avoiding the possible errors due to changes in the atmospheric condition between calibration and measurement . although these exposures tend to have short exposure times , from 20s to 200s , the large field - of - view of cafos ccd ( @xmath015@xmath14@xmath915@xmath14 ) and pixel size ( 0.53@[email protected]@xmath8 ) , allow a good estimation of the sky background .
the use of calibration fields , with more than 4 calibration stars observed per field , guarantees a good estimation of the magnitude zero - point per filter and exposure ( photometric calibration error @xmath150.1 mags ) .
we focused our search in nights dedicated to a single project , aimed to study the time evolution of the multiband photometry of supernovae ( pi : dr .
w. hillebrandt ; pastorello et al .
2007 ) , observed in service mode .
the reason for that was that this program uses the same instrument setup , filters , calibration fields and covers a large periods of time .
the number of nights totally or partially dedicated to this project were @xmath021 along the considered time period .
finally we selected the darkest possible nights , ie .
, moonless nights that distance at least 13 days from the full moon .
we were then left with 5 nights .
however , only in two of them the calibration fields were observed at least one hour after the begining of the astronomical night . in the remaining 3 nights
there was still substantial contamination from the twilight , and they are therefore excluded from any further analysis .
table [ tab_data ] shows the final list of selected nights , including the broad - band filters observed each night .
the landolt calibration fields pg1633 and pg0918 ( landlot 1992 ) were observed each night , respectively , in the listed filters .
the images were reduced following the standard steps , using iraf routines .
first a master bias frame was created for each night by averaging all the bias frames obtained along the night and smoothing it .
all images were then bias corrected by substracting the corresponding master bias . for each band
a master flat - field frame was obtained by averaging the bias corrected domeflat images , and normalizing to the median intensity value .
images were then corrected for pixel - to - pixel response variations dividing by their corresponding flat - field frames .
no sky subtraction was performed since the aim of this study is to determine its intensity .
the calar alto extinction monitor ( cavex ) , is an inhouse developed instrument ( pi : u.thiele ) , which estimates the monocromatic extinction in the v - band continously along each night .
the system is fully automatic , opening half an hour after the beggining of the astronomical night and closing half an hour before its end .
it continously points towards the north ( the polar star ) , taking images of @xmath020s exposure time every @xmath076s covering a field - of - view of @xmath055 degrees with a resolution of 2.3@xmath14/pixel . by traking the location of 15 - 20 stars in the field ,
it estimates the extinction by measuring their apparent magnitudes across a range of 1.1 - 2.4 airmasses .
it shares the same limits of humidity and wind speed than the telescopes at the obsevatory , producing a measurement of the extinction every @xmath02 minutes if the night is clear .
when the night is cloudy or the extinction has strong fluctuations , the instrument does not produce a reliable data , flagging it .
therefore , the fraction of time without precise extinction measurements from the cavex is a good estimation of the amount of time lost due non astronomical weather conditions in the observatory .
we have collected all the available data from the extinction monitor during the last 4 years , from may 2003 to may 2007 .
the data comprises 214193 individual measurements , corresponding to 1044 nights from a total of 1478 nights included in this period .
cavex estimates the total extinction in a single band . to characterize the extinction curve it is needed to measure the extinction at different wavelengths . in the summer of 2006 ( 27/07/06 - 15/08/06 ) and the winter of 2006 - 2007 ( 17/12/06 - 13/01/07 ) , an instrument built with the sole purpose of estimating this extinction curve was installed , named the extinction camera and luminance background register ( excalibur , pi : j.aceituno ) .
excalibur is a robotic extinction monitor able to do cuasi - simultaneous photometric observations in 8 narrow bands covering the wavelength range between 3400 and 10200 , characterizing the extinction curve in this wavelength range ( although only 6 of the bands were operative when installed at calar alto ) .
the instrument was built to estimate the aerosol abundance based on the shape of the extinction curve ( prez - rmirez 2007a , b ) .
it estimates @xmath016 extinction coefficients for each of the sampled bands per hour , and an average of @xmath0160 extinction coefficients per band for each night .
it was operative for a total of 6 nights in the summer season and 14 nights in the winter season .
we collected all these data for the current study .
the night - sky seeing is measured by a differential image motion monitor ( dimm , aceituno 2004 ) at the calar alto observatory since august 2004 , and it is nowadays a fully automatic instrument .
it measure the seeing at the wavelengths corresponding to the johnson @xmath2-band .
this dimm was calibrated with a previous one installed in the observatory @xcite , which , at the same time , was also calibrated with a generalized seeing monitor in may 2002 @xcite , when they both estimated a median seeing of 0.90@xmath8 for that time period . to study the possible evolution of the seeing at the observatory and its possible dependencies with another night - sky parameters we collected all the available seeing measurements on the time period between january 2005 and december 2006 .
contrary to the cavex , the dimm has more severe limitations to be operative , and it closes automatically when the humidity is larger than an 80% or the wind speed is 12 m s@xmath12 , which comprises @xmath050% of the time .
the dimm produces an estimation of the transversal and horizontal seeing in each measurements .
whenever there is an occasional large difference between both estimations ( not due to a dominant wind component ) they most probably are not due to atmospheric effects , but rather to mechanical oscilations .
due to inherent stability of a dimm system to this kind of oscilations , these cases comprise a relatively small number ( @xmath05% ) .
they have been excluded from the final dataset .
we finally collected a total of 213622 seeing measurements distributed along 335 nights during the considered time period .
we describe here the analysis performed over each of the different collected data .
the data included in the final sample of night - sky spectra for moonless dark nights described in section [ spec_data ] were combined to create a typical moonless night - sky spectrum at the observatory .
for doing so each spectrum was normalized to the mean intensity ( of the sample ) at 5200 ( 0.13 10@xmath11 erg s@xmath12 cm@xmath4 @xmath12 ) .
then the final spectrum was produced by averaging the flux of these normalized spectra at each sampled wavelength .
figure [ spec ] shows the resulting spectrum , covering a wavelength range between 3700 and 7933 .
this is the first time that a typical night - sky spectrum is published for the calar alto observatory .
the emission lines clearly identified in the spectrum have been labeled with its corresponding atomic name and wavelength .
several of the distinctive features of the night - sky spectrum are due to airglow , although a substantial fraction is due to light - pollution .
the airglow is emitted by atoms and molecules in the upper atmosphere which are excited by solar uv radiation during the day and twilight ( ingham 1972 ) .
airglow is the most important component of the light of night - sky .
it produces the 5577 and 6300 lines from oi ( which are stronger near the twilight ) .
it contributes to the ubiquitious 5890 - 6 nad doublet , although this line is heavely contaminated by light - pollution from low and high preassure street - lamps .
airglow is also responsible for the oh rotation / vibration bands in the red and ir , known as the meinel bands ( meinel 1950 ) , visible at the redder wavelengths of the spectrum ( see fig .
[ spec ] ) .
in addition it produces a pseudo - continuum in the blue from overlapping o@xmath16 bands ( 2600 - 3800 ) and in the green from no@xmath16 bands ( 5000 - 6000 ) .
a more detailed description of the effects of the airglow on the night - sky emission can be found in benn & ellison ( 1998a ) .
an atlas of the airglow from 3100 to 10000 was presented by ingham ( 1962 ) and broadfoot & kendell ( 1968 ) .
other contributions to the night - sky spectrum are the zodiacal light , the starlight , and the extragalactic light , which increases the background continuum emission ( see benn & ellison 1998a and references therein for a more detailed explanation on their effects ) . all of them , including the airglow , comprises the natural processes that produce the night - sky spectrum in any astronomical site .
in addition to them , the night - sky spectrum can be affected by the light - pollution due mostly to the street - lights of populated areas near the observatories .
light pollution arises principally from troposheric scattering of light emitted by sodium and mercury - vapour and incandescent street lamps ( mcnally 1994 ; holmes 1997 ) .
typical spectra of the three more used types of street - lamps were presented by osterbrock et al .
( 1976 ) : the sodium low - pressure lamps , the sodium high - pressure lamps and the mercury lamps .
the first of the three is the one with less impact in astronomical observations , since its produces most of its light concentrated in the 5890 - 6 nad and 8183 - 95 nad emission lines .
therefore only a very reduced number of astronomical observations are strongly affected by them . on the other hand
the high - pressure sodium lamps emitt most of its light in a broad nad line centred in @xmath05890 , with a fwhm of @xmath0400 , that shows a central reversal .
they also show a strong 8183 - 95 emission line , and fainter emission lines at 4494 - 8 , 4665 - 9 , 4758 - 52 , 4979 - 83 , 5149 - 53 , 5683 - 88 , 6154 - 61 and 7665,7699 .
the high - pressure sodium lamps may affect strongly the quality of any astronomical observation .
finally the mercury lamps produce narrow lines at 3651/63 , 4047 , 4358 , 5461 , 5770 and 5791 , together with broad features at 6200 and 7200 of fwhm@xmath0100 , from the phosphor used to convert uv to visible light .
they also produce a weak continuum emission over the whole visible range .
some mercury pollution lines can strongly affect certain astronomical studies : ( i ) the 4358 hg line strongly affect any attempt to measure the emission of the 4364 [ oiii ] line in any object in the galaxy .
this line is fundamental for the estimation of the electron temperature of galactic nebulae .
( ii ) the 5460 hg line lies in the centre of the @xmath17-band of the @xmath18 strmgren photometric system , which may affect the programs devoted to the study of stellar populations that use this filter set .
other sources of light pollution are the incandescent lamps and the high - pressure metal halide lamps . the spectrum of the former consists of continuum emission only , and it is difficult to identify in a night - spectrum .
the latter are nowadays frequently used in the illumination of sport stadiums and arquitectonical monuments ( which can be considerable , since they are normally oriented towards the sky ) .
these high - pressure metal halide lamps exhibit some scandium , titanium and litium emission lines , that are charactezed by a blue edge due to molecular bands ( general electric 1975 ; lane & garrison 1978 ; osterbrock et al .
1976 ) .
the night - sky spectrum shown in fig.[spec ] shows clear evidence of strong light pollution from all the street - lamps described before .
it shows strong mercury lines all over the spectrum , the typical emission lines and features of high - pressure sodium lamps , with a well detected broad emission at @xmath05900 , and a strong nai emission line at 5893 indicative of low - pressure sodium lamps .
evenmore , it also presents the typical emission lines corresponding to high - pressure metal halide lamps .
all this pollution comes from the populations nearby the observatory , in particular from almeria ( @xmath0250000 habitants ) , at 40 km towards the south , and smaller towns at the north ( like baza , @xmath021000 habitants , and macael , @xmath06000 habitants ) .
there are well established estimations of the contribution of city lighting to dark - sky brightness ( e.g. , treanor 1973 , walker 1973 , yocke et al .
1986 , garstang 1991 , and reference therein ) . however , its contribution to the spectrum is more complex , since it depends on the particular kind of lamps used for street illumination .
most major observatories nearby populated areas have some kind of night - sky protection laws with the aim of reducing the effects of light pollution by controlling the fraction of light that escapes towards the sky and the kind of lamps used ( normally they promote the use of low - pressure sodium lamps ) .
the calar alto observatory does not benefit yet from any local sky - protection law which regulates the street illumination , with the corresponding effects that can be appreciated in the night - sky spectrum . to estimate the contribution of light pollution to the night - sky spectrum
we derive the flux intensity corresponding to each of the detected emission lines .
each line was fitted with a single gaussian function using fit3d @xcite , using the same procedure described in @xcite .
table [ tab_line ] lists the integrated flux for each of the detected emission lines shown in fig.[spec ] , together with the identification of the line and the nominal wavelength .
in addition to the emission lines , the flux intensity of the sodium broad - band emission at @xmath05900 was also estimated by fitting the feature with a single broad gaussian function .
the result is also listed in table [ tab_line ] .
all the fluxes were converted to rayleighs following the conversion formulae by benn & ellison ( 1998a ) .
these values can be compared to those ones obtained for another astronomical sites .
e.g. , pedani ( 2005 ) presented a study of the night - sky spectrum at the observatory of the roque de los muchachos , la palma .
the intensity of the lines produced by the mercury and low - pressure sodium lamps found in our spectrum are only comparable to those ones in la palma when pointing in the directions towards the most polluting towns in this island . on the other hand ,
the contribution of the lines produced by the high - pressure sodium lamps is much lower , being comparable to that of the less polluted areas of the sky in la palma . this may indicate that most of the street lamps used in almeria are mercury and low - pressure sodium lamps , rather than high - pressure sodium ones , and therefore they only affect specific wavelength ranges ( and science programs ) . finally we clearly see lines produced by high pressure metal halide lamps , only marginally ( and recently ) detected in the sky spectrum at la palma ( pedani 2005 ) . once determined the contribution of each emission line to the night - sky spectrum it is possible to decontaminate it by the effect of the pollution lines ( ie .
, all but the oi ones ) , creating a _
clean _ night - sky spectrum .
this can be done directly for the mercury , scandium , titanium and litium lines , since all its emission is produced by light pollution . however , in the case of the sodium lines there is a natural contribution due to the airglow .
we lack a direct measurement of this natural contribution at calar alto , that can only be achived nowadays with a general blackout .
however its natural contribution must not be significantly larger than in other astronomical sites .
benn & ellison ( 1998a ) estimated the natural contribution by the broad sodium emission band in @xmath00.03 mag in v and @xmath00.02 mag in r for la palma observatory .
similar contributions are expected by the nai 5893,6 emission lines .
therefore , the natural contribution by the sodium due to the airglow is expected to be of the order of @xmath00.04 mag in both bands .
thus , as a first order we can assume that all the detected emission by sodium in our sky - emission spectrum is due to pollution , and once determined its contribution we can correct it by the expected contribution of natural emission , if needed .
table [ tab_cont ] lists the estimated contribution of the light pollution to the sky background in different bands .
it includes the @xmath1 , @xmath2 and @xmath3-johnson filters , and a set of medium band filters selected from those ones used by the alhambra survey ( moles et al .
2007 ) , a major imaging survey currently on going at the calar alto observatory .
they are included to illustrate the effects of the pollution lines in the background sky when using median / narrow - band filters affected by these lines .
the contamination from light pollution is clearly stronger in the @xmath1-band than in other astronomical sites : e.g , @xmath00.02 mag at la palma ( benn & ellison 1998a ) , 0.02 - 0.04 mag at kitt peak ( massey et al .
although there are astronomical sites with similar or stronger contaminations : e.g. mount hopkins ( see fig .
2 , from massey & foltz 2000 ) .
this contamination can be reduced by a proper light pollution law that limits the use of the mercury street lamps .
strong benefits from such laws has been experienced in different sites ( e.g. , benn & ellison 1998a ; massey & foltz 2000 ) .
the contamination in the @xmath2 and @xmath3-bands is slightly stronger than in some other places , like la palma : 0.05 - 0.10 mag at the @xmath2-band and 0.07 - 0.12 mag at the @xmath3-band ( pedani et al .
2005 ) , but it is similar or smaller than in other astronomical sites , like kitt peak : 0.19 - 0.33 mag in the @xmath2-band ( massey et al .
1990 ) or mount hopkings : 0.17 mag in the @xmath2-band ( massey & foltz 2000 ) .
one of the two recomendations of the iau for a dark place is that the contribution of the pollution to the sodium emission should not exceed in intensity the natural airglow one ( smith 1979 ) .
if we consider that the airglow emission is similar in calar alto than in la palma ( or at least of the same order ) , it is clear that the contribution from the light pollution is much stronger . in these regards
calar alto does not fulfill the iau recomendations for a dark place . a proper light - pollution law , that increases the use of low - preasure sodium lamps rather than high - presure ones , will not reduce the net effect of the light pollution to the sky - background in these bands , but it will concentrate it in a more reduced wavelength range , affecting less observing programs .
most of the lamps that cause the light pollution also produce a certain level of continuum emission . in particular mercury and high pressure metal halide ones .
however its contribution is difficult to estimate .
although we donecannot quantify its effect , the reduction of the use of mercury , high - pressure sodium lamps and high pressure metal halide lamps in the vicinity of the observatory will certainly also reduce the sky - background continuum .
the lack of previous similar studies of the night sky spectrum at the calar alto observatory ( to our knowledge ) does not let us to analyze the evolution of the light pollution along the time .
the night sky brightness was determined by using both the imaging and spectrocopic data described before . in the case of the imaging data ,
the calibration fields contain at least 5 calibration starts per frame .
the magnitude zeropoint for each image was determined by measuring the counts intensity of each of these stars within a fixed aperture of 8@xmath8 radius , using imexam task of iraf package , and applying the formula : @xmath19 where @xmath20 is the magnitude zeropoint , @xmath21 is the apparent magnitude of the calibration star in the corresponding band , @xmath22 is the extinction , derived from the correponding value measured by the cavex this night and the airmass of the image , @xmath23 are the measured counts within the indicated aperture , and @xmath24 is the exposure time .
since the measured seeing of the images ( fwhm of the field stars ) , ranges between 0.9@xmath8 and 1.3@xmath8 , this aperture ensures that most of the flux is contained within it , and no aperture correction was applied .
the average of the values derived for each calibration star in the field is considered the zeropoint of the image , and the standard deviation from this mean value is considered as the photometric error .
this standard deviation ranges between 0.02 mag and 0.06 mag for each band and each night .
the sky brightness was then determined in each image by measuring the mean counts level in several ( @xmath25 ) square apertures of @xmath050@xmath8@xmath950@xmath8 , in areas free of targets , using imexam task of iraf package .
the mean value of these measurements is considered as the count level of the sky brightness , and the standard deviation with respect to this mean value the error in the count level estimation .
finally , the sky surface brightness was determined using the formula : @xmath26 where @xmath27 is the surface brightness in magnitudes per square arcsec , @xmath20 is the zeropoint described before , @xmath28 is the sky counts level estimated and @xmath29 is the pixel scale in arcseconds ( ie .
, 0.53@xmath8 for cafos ) .
it is noticed that the magnitude zeropoint was corrected by the extinction , but the sky brightness was not , following the convention adopted in most of the recent studies of sky brightness ( e.g. , walker 1988b ; krisciunas 1990 ; lockwood et al 1990 ; leinert et al .
1995 ; mattila et al . 1996
; benn & ellison 1998a , b ) .
correcting the sky brightness for extinction would be appropiate only if the extinguishing layer were known to be below all sources of sky brightness , which is not the case ( benn & ellison 1998a , b ) .
in addition to the estimations of the sky surface brighness obtained using direct imaging we also derived the sky surface brightness by using the ppak spectroscopic information for the only full moonless night of our sample ( 02/06/2005 ) .
the two night - sky spectra obtained this night cover the wavelength range of the @xmath1 , @xmath2 and @xmath3-band filters .
we derived the sky flux intensity for each of these filters by convolving each spectrum with the corresponding filter tranmission curves listed in the asiago database on photometric systems ( moro & manuri 2000 ) .
then the fluxes were transformed to magnitudes by using the zero - pointings listed in fukugita et al .
the mean value of the two derived magnitudes in each filter is adopted as the sky surface brightness of that night , and the absolute difference between both magnitudes as the error .
table [ tab_mag ] lists the sky - surface brightness obtained from both the imaging and spectroscopic data at the different bands for each moonless night .
in addition , it lists previous results on the sky - brightness at different astronomical sites , including the results presented by leinert et al .
( 1995 ) for calar alto , derived from broad - band photometry obtained along three nights in 1990 .
there is a wide range of sky surface brightness values for the different bands , considering that all the measurements were obtained in full moonless nights , without twilight contamination .
the data from the first night , derived using ppak data , are very similar to those ones from the third night , derived using cafos data , being the sky in the former slightly brigher in the @xmath1-band .
the sky brightness of the second night is brighter in all the bands .
looking back to the data , we realize that the airmass of the images of the 1st night ( @xmath01.7 ) , corresponds to an elevetion much lower than that of the 2nd and 3rd night data ( @xmath01.3 and @xmath01.2 , respectively ) .
the sky brightness increases with the airmass for two different reasons .
one is a natural effect of the airglow , which is brighter at low elevations cause the line of sight intercepts a larger number of atoms in the airglow layer ( garstang 1989 ; benn & ellison 1998a , and references therein ) .
a second effect is the increase of light pollution when pointing towards high populated areas at low airmass .
walker ( 1971,1991 ) and garstang ( 1989 ) estimate that the increase in the brightness at an air mass of @xmath01.4 , in the direction of a populated area of p inhabitants at a distance d km to be @xmath30 mag .
that would correspond to @xmath00.3 mag when pointing directly towards the south , where is the largest populated city nearby ( almeria ) . however since our observed fields are mostly pointing towards the east its contribution is more difficult to estimate .
it is not possible to know which is the actual contribution of the light pollution to the continuum brightness at the zenith .
therefore we do not know if calar alto fulfill ( or not ) the other iau recomendation for a dark site , that is that this contribution must be lower than @xmath00.1 mag ( smith 1979 ) .
it is possible to derive an approximate expression of the sky brightness dependecy on the zenith distance due to natural effects based on the results by garstang ( 1989 ) , as already pointed out by krisciunas & schaefer ( 1991 ) .
this expression can be used to correct the measured values and derive a much appropiate estimation of the zenithal sky brightness ( benn & ellison 1998a ; patat 2003 ) .
patat ( 2003 ) derive the following formula ( appendix c of that article ) for this correction : @xmath31+\kappa ( x-1)\ ] ] where @xmath32 is the increase in sky brightness at a certain band and airmass ( @xmath33 ) , @xmath34 is the fraction of the total sky brightness generated by airglow , being ( 1-@xmath34 ) the fraction produced outside the atmosphere ( hence including zodiacal light , faint stars and galaxies ) and @xmath35 is the extinction coefficient at the corresponding wavelength .
we applied this correction to our data , using the typical extinction curve at calar alto ( following sections ) normalized to the corresponding @xmath36 extinction coefficient of each night .
a typical value of @xmath370.6 was used for this correction ( patat 2003 ) .
once applied there is a significant reduction of the dispersion between the values obtained for each night .
this indicates that the correction works pretty well , despite the fact that it does not take into account the effects of light pollution .
the mean values of the sky brightness at the zenit after correction for each band are also listed in table [ tab_mag ] .
the night sky - brightness at calar alto shows no significant change in the last 15 years , when comparing with the results by leinert et al .
they did not applied any correction for the dependecy on the zenith distance to their data ( table 6 of that article ) .
therefore we must compare with the mean values without correction .
the only band where it is appreciated an increase of the sky - brightness is the @xmath5-band ( @xmath00.4 mag brigther ) .
however , if we take into account that we only have data for this band at low elevation , we can not consider these results conclusive .
for the remaining bands the sky seems to be @xmath00.2 mag brighter in the @xmath1-band , @xmath00.2 mag fainter in the @xmath3-band and without changes in the @xmath2- and @xmath6-bands , when comparing with the mean values derived for the three nights of our sample
. however none of these differences seems to be significative , lying withing the errors of our measurements . when comparing with the two darkest nights with data obtained at high elevation the differences ( if any ) disappear for the @xmath1 and @xmath2-bands , and the sky seems to be even darker in the @xmath3 and @xmath6-band .
a possible caveat to this comparison is that the results listed in leinert et al .
( 1995 ) were obtained not exactly in the solar minimun , which may produce an increase of the sky - brightness . however , although their broad - band data were obtained in the 1990 , they also obtained intermediate - band data in the 1993 and their sinthetized broad - band surface sky - brightness are similar to those obtained in the nights of the 1990 .
the sky surface brightnesses at the calar alto observatory at different bands , listed in table [ tab_mag ] , are remarkable similar to those ones at many other different astronomical sites . in the optical wavelength range ( @xmath5,@xmath1,@xmath2 and @xmath3-bands ) ,
calar alto seems to be a particular dark site , comparible to mauna kea .
the fact that both places seem darker than paranal may be an artifact since the data presented by patat et al .
( 2003 ) were taken during the maximun of solar activity .
this result is in anycase remarkable , considering the strong light pollution present in the calar alto spectra , which effect is particularly strong in the @xmath1 , @xmath2 and @xmath3-bands ( tab .
[ tab_cont ] ) .
most of the listed observatories have little light pollution or they benefit of specific protection laws against it .
this has been demonstrated as a tremendous useful tool to preserve or increase the quality of the night - sky ( eg . ,
benn & ellison 1998a , b ; massey & foltz 2000 ; walker & schwarz 2007 ) .
if the effects of light pollution could be reduced in the vicinity of calar alto it would become a particularly dark site for optical observations .
on the other hand , the sky is clearly brighter in the @xmath6-band than in any other astronomical site listed in this table . despite the fact that the observatory is located in the most arid place in europe , in the vecinity of a desert ( the tabernas desert ) ,
the water vapor meinel bands are particularly strong . the humidity at calar alto is higher than in other astronomical sites , like paranal or mauna kea , although there are frequent epochs of low humidity ( @xmath1520% ) in the summer .
the height of the observatory , @xmath02200 m over the sea level , normally places it under the inversion layer , which has a particular strong impact in the strength of the water vapor emission lines .
both combined effects can explain the rise of the sky - brightness in the @xmath6-band .
it is important to note here that this effect has a relatively reduced effect over near - infrared observations in the @xmath38 , @xmath39 and @xmath40-band .
the median @xmath2-band extinction at the calar alto observatory for the time period covered by the cavex data ( section [ ext_data ] ) was @xmath00.18 mag , with a mean value of @[email protected] mag .
this value is slightly smaller than the previously reported by hopp & fernandez ( 2002 ) , which was based on a much smaller sample of data ( comprising 74 nights spanned between 1986 and 2000 ) .
they found that there was an increase of the extinction at the summer season , that was most probably associated with an increase of the aerosols ( ie .
dust ) in this period of the year .
similar seasonal pattern has been appreciated in another major observatories : e.g. , la palma observatory is strongly affected by dust extinction in the summer when dust from the sahara desert ( @xmath0400 km away ) blows over the canary islands ( benn & ellison 1998a ) .
although the calar alto observatory is nearer to the sahara desert ( @xmath0250 km away ) than la palma , it is normally out of its main wind streams , being shielded by the altas mountains . on the other hand , it is located in an arid region nearby a much smaller desert , the tabernas desert ( @xmath015 km away ) .
figure [ cavex ] shows the evolution of the average @xmath2-band extinction for each night along the period of time sampled by the dataset . as already suspected by hopp & fernandez ( 2002 ) , there is a clear seasonal pattern .
the typical extinction in the winter nights is @xmath70.15 mag , being mostly restricted to values lower than @xmath420.2 mag . in summer time there is a wider range of extinctions , although in most of the cases the extinction is lower than @xmath420.4 mag .
as indicated before this increase of the extinction is most probably associated with a rise of the aerosols ( dust ) in the atmosphere .
we will explore that possibility latter .
this seasonal pattern is somehow similar to the one seen in la palma . indeed
the fraction of nights with @xmath430.25 mag is similar in both observatories , @xmath020% of the nights .
however there is a major difference : the fraction of nights with high extinction , @xmath430.4 mag , at calar alto is very reduced , @xmath03% , while at la palma this fraction is @xmath010% of the nights , with frequent peaks of extinction over @xmath430.6 mag ( benn & ellison 1998a , figure 3 ) .
based on the fraction of nights that the cavex was operative and derived realiable measurements of the @xmath2-band extinction ( @xmath36 ) , it is estimated that @xmath070% of the nights were astronomically useful in the period covered by these data ( 4 complete years ) .
this fraction is remarkable similar to that one in many other astronomical sites ( eg . , la palma , benn & ellison 1998a ) .
the fraction of fully photometric nights , defined as nights where the @xmath2-band extinction varies less than a 20% along all the night , was @xmath030% .
the excalibur data described in section [ ext_data ] were used to determine the typical extinction curve at calar alto .
this curve was previously studied by hopp & fernandez ( 2002 ) , using an inhomogenous dataset .
the extinction coefficients were analyzed separately for the summer and winter seasons due to the observed seasonal pattern .
first , it was derived the mean extinction coefficients per night by averaging all the measured extinction coefficients per band obtained along each night ( @xmath0160 values ) .
then , the mean extinction coefficients per season were determined by averaging all the measured extinction coefficients per band obtained along each season nights .
table [ tab_ext ] lists the average extinctions coefficients obtained for each season for each of the 6 bands sampled by the instrument , including their central wavelengths and the standard deviation with respect to these mean values .
as expected the standard deviation in the extinction coefficients is larger for the summer data than for the winter ones , in agreement with the results shown in the previous section .
the excalibur results are consistent with the cavex ones for the wavelength covered by both instruments , ie .
, @xmath0500 nm . this indicates that the extinction coefficients listed in table [ tab_ext ] are a good representation of the typical values for each season .
the total extinction is mostly due to three contributions , the rayleigh scattering at the atmospheric atoms and molecules , the extinction due to aerosol particles ( mostly dust ) , and extinction due to ozone ( walker 1987b ) .
the rayleigh scattering can be described by @xmath44 where @xmath45 is the wavelength and @xmath1 is a constant that mostly depends on the pressure , the temperature and the normalized refractive index of the air ( @xmath46 , slightly wavelength dependent ) . in general @xmath1
can be replaced by its average value for the mean conditions in a certain astronomical site ( e.g. , rufener 1986 ) .
the aerosol particles produce a similar absorption , that can be described by @xmath47 where @xmath48 is a parameter that depends mostly on the height of the observatory ( @xmath49 ) , and @xmath10 is a power law index that depends on the size of the aerosol grains .
although an @xmath501.3 , derived by siedentopf ( 1948 ) , is widely used , we adopted a value of @xmath510.8 for consistency with the previous study of the extinction curve at calar alto ( hopp & fernandez 2002 ) . the ozone extinction is a selective absorption by molecular bands .
it can be approximately described by a broad gaussian function centred in @xmath526000 ( matching the shape shown in runefer 1986 and hopp & fernandez 2002 ) : @xmath53\ ] ] each single contribution to the extinction depends on the wavelength and a particular constant that , in the case of the two first , depends on the height and the average atmospheric conditions in the observatory .
we adopted the values listed for @xmath545400 in rufener ( 1986 ) , consitent with those derived by tg ( 1977 ) , obtained for la silla observatory .
a similar approach was followed by hopp & fernandez ( 2002 ) .
the height and average weather conditions in this observatory are very similar to those of calar alto , which justifies the use of these constants .
therefore the total extinction curve is a linear combination of these three contributions : @xmath55 the extinction coefficients for the two seasons were fitted to this linear combination , deriving the relative contribution of each one ( @xmath56 ) to the total extinction .
table [ tab_ecurve ] lists the results from this fitting analysis , including the relative contribution ( @xmath56 ) derived for each component for each season . for comparison it also includes the same relative contributions derived for calar alto by hopp & fernandez ( 2002 ) , and their compilation of similar results for different astronomical sites .
the contribution of the rayleigh scattering seems to be rather constant for the two considered seasons , like the ozone absorption . on the other hand ,
the aerosol contribution rises considerable in the summer time , being responsible of the increase of the extinction in this season , as it was thought .
the estimated contributions of the rayleigh scattering and the aerosol extinction are very similar to the values reported by hopp & fernandez ( 2002 ) for the winter season , which may indicate that both contributions have not changed considerably with time ( their data corresponds to 1986 - 2000 ) .
both contributions are also similar to the ones derived for other major astronomical sites . on the other hand
the contribution of the ozone absorption seems to be stronger than in previous measurements and other astronomical sites .
unfortunally the coverage of excalibur when mounted at calar alto did not allow to perform an accurate sampling of the wavelength range affected more strongly by the ozone absorption , which did not let us to be conclusive on this respect .
figure [ ext_curve ] shows the distribution of the extinction coefficients along the wavelength for the two season datasets .
it also includes the best fitted linear combination of the three components to the extinction and each of these components scaled to its relative contribution to the total extinction . despite the apparent increase of the ozone absorption its actual contribution to the total extinction at any wavelength
is very limited , being neglectible in comparison with the other two contributions .
indeed the fitting process derives equally good results when this contribution is removed . as already indicated
the rayleigh scattering contribution is very similar for both datasets , being the dominant contribution in the winter season , ie .
, in conditions of low extinction . in winter time is responsible of the @xmath085% of the extinction in the @xmath2-band , only @xmath011% is due to aerosol extinction and @xmath04% to ozone absorption . on the other hand , in summer its contributions
drops to @xmath063% , with @xmath035% due to aerosol extinction and @xmath02% due to ozone absorption .
curiously , the contribution of aerosols to the extinction in the conditions of minimun extinction are much reduced than the one found in other astronomical sites , like la silla ( burki et al .
1995 ) .
as we shown in the previous section the extinction curve depends strongly on the relative contribution of each of the three major components to the extinction .
each of them has a different dependecy with the wavelength , and two of them depends also on the particular atmospheric conditions at the observatory .
therefore , it is difficult to derive a precise extinction curve valid for every night that depends only in a reduced number of parameters .
however it is still possible to look for an approximate expression for the extinction curve that provide an useful estimation of the extinction at any wavelength and that depends only in a single parameter : the @xmath2-band extinction , measured each night by the cavex .
based on the results of the previous section it is known that the ozone absorption have a marginal effect in the total extintion at any wavelength .
therefore we have not considered it for our approximate expression .
it is also known that the contribution of the rayleigh scattering is almost contant along the year , and therefore the variations in the extinction are controlled by the amount of aerosol particles ( ie . , dust ) . based on this assumption an approximate expression for the extinction curve
can be derived by considering that the extinction in the @xmath2-band is due to a fix contribution of rayleigh scattering ( the average for both seasons ) and a variable contribution due to aerosol extinction .
the derived expression is : @xmath57 the typical differences found in the extinction coefficients derived using this formula and the more precise decomposition presented in the previous section are of the order of @xmath010% .
the seeing data described in section [ seeing_data ] were used to determine an average seeing for each night comprised in the dataset ( spanned along @xmath02 years ) .
the median atmospheric seeing for all the time period was @xmath00.90@xmath8 , with a @xmath070% of the nights under subarcsecond seeing ( @xmath151@xmath8 ) .
figure [ seeing1 ] shows the nightly averaged seeing distribution along the time .
the epochs without data in april 2005 and march 2006 were due to reparations in the hut of the dimm .
there is a mild seasonal pattern in the seeing distribution , with the best seeing concentrated in the summer season . to further investigate this possibility we created the seeing histogram for all the data comprised in the dataset and for two different subsets of data corresponding to the summer season ( may - september ) and the winter season ( the rest of the months ) .
figure [ seeing2 ] shows the three histograms .
the differences in the seeing distribution for the summer ( median seeing @xmath00.87@xmath8 ) and the winter seasons ( median seeing @xmath00.96@xmath8 ) are clearly appreciated . not only the median seeing is better in the summer season , the chances of having better seeing in a summer night are larger . table [ tab_seeing ] lists the median seeing estimated for both the total sample and the two season subsamples . for comparison purposes it also lists the atmospheric seeing measured at different astronomical sites world - wide , ordered by increasing seeing .
although the median seeing at calar alto is larger than that of some major astronomical observatories ( mauna kea , la palma ) , it is actually better than in many other astronomical sites ( eg . , mtgraham , paranal ) .
we have characterized the main properties of the night - sky at the calar alto observatory , comparing them , when possible , with similar properties of other different astronomical sites .
the main results of this article can be summarized in the following points : * an average night - sky spectrum for the moonless dark - time at the observatory has been presented for the first time .
this spectrum , which covers the optical wavelength range ( 3700 - 7933 ) , is distributed freely to the community .
airglow and light - pollution emission lines are detected in this spectrum .
the strength of the light - pollution lines has been measured , estimating their contribution to the emission in different bands . in comparison with other sites the mercury lines are particularly strong .
the contribution of the light pollution to the sodium emission is far stronger than its natural emission . in this
regards calar alto does not fulfill the iau recomendations for a dark astronomical site ( smith 1979 ) , like other major astronomical sites ( eg . , la palma , pedani 2005 ) . *
the moonless night - sky brightness at the zenith has been determined for the @xmath5 , @xmath1 , @xmath2 , @xmath3 and @xmath6-bands .
there was no appreciable change in the sky - brightness over the last 15 years . in comparison with other astronomical sites ,
calar alto shows a particularly dark sky in the optical bands , similar to that of mtgraham or mauna kea
. the sky brightness could be even darker if it was possible to reduce the light pollution in the optical bands , which would place calar alto as a very dark astronomical site .
on the other hand , the sky is brighter in the @xmath6-band , mostly due to the strength of the water - vapor meinel bands . *
the extinction , measured along the last 4 years , shows a seasonal dependency with a typical value of @xmath580.15 mag in winter time and a wide range of values in summer , most of them restricted to @xmath590.4 mag .
this seasonal pattern , caused by saharan dust , is similar to the one found in la palma , but with a smaller range of values in summer time .
the analysis of the typical extinction coefficients at different wavelenghts for each season indicates that the rise of the extinction in summer is due to an increase of aerosols ( dust ) in this period of the year . due to the reduced contribution of the ozone absorption to the extinction , and the stability of the contribution of the rayleigh scattering along the year it was possible to derive an aproximate expression for the extinction curve parametrized only by the @xmath2-band extinction . * the fraction of astronomical useful nights , when the weather was good enough to allow an acurate measurement of the extinction , was @xmath070% of the nights in the last 4 years .
this fraction is similar to the one found in la palma ( benn & ellison 1998 ) .
the fraction of these nights that were photometric was a @xmath030% .
* the median seeing along the last 2 years was 0.90@xmath8 , being slightly better in summer ( 0.87@xmath8 ) than in winter ( 0.96@xmath8 ) .
the seeing was better than 1@xmath8 in a @xmath070% of the nights .
although this seeing is slightly worse than in some astronomical sites ( eg .
mauna kea , la palma ) , it is better than the currently seeing at paranal or mtgraham , two astronomical sites where 10m - like telescope are currently in operation or under construction .
we conclude that calar alto remains a good astronomical site , similar in many aspects to places where there are 10m - like telescopes under operation or construction .
it will strongly benefit from a sky protection law that would reduce the light pollution , particularly due to mercury and high - pressure sodium street lamps .
such a law has been under discussion by the local andalusian goverment during the last few years and we hope it will be soon operative . the fact that calar alto is placed in continental europe is a major advantage in comparison with other european observatories away from the continent , since both the operational and development costs are significantly smaller . for both reasons
we consider that this observatory is a good candidate for the location of future large aperture optical telescopes .
sfs thanks the spanish plan nacional de astronoma program aya2005 - 09413-c02 - 02 , of the spanish ministery of education and science and the plan andaluz de investigacin of junta de andaluca as research group fqm322 .
prez - rmirez , d. , ruiz , b. , aceituno , j. , olmo , f.j . , & alados - arboledas , l. , 2007 .
`` application of sun / star photometry to derive the aerosol optical depth '' .
international journal of remote sensing . in press .
llrdate & instrument & wavelegth range + 25/05/05&ppak&3700 - 7000 + 26/05/05&ppak&3700 - 7000 + 02/06/05&ppak&3700 - 6800 + 30/06/05&ppak&4700 - 7940 + 03/07/05&ppak&3700 - 7100 + 05/07/05&ppak&3700 - 7100 + 23/08/06&ppak&3700 - 7000 + date & instrument & filter bands + 28/08/05&cafos&u , b , v , r & i - bands + 28/03/06&cafos&b , v , r & i - bands + llrrline i d & wavelength & flux@xmath60 & flux@xmath61 + & ( ) & & r + hgi & 4047 & 3.5 & 9.6 + hgi & 4078 & 0.7 & 2.0 + nai & 4165,8 & 1.2 & 3.6 + hgi & 4358 & 6.7 & 22.9 + nai & 4420,3 & 0.3 & 1.1 + tii & 4511.6 & 3.6 & 13.7 + nai & 4665,9 & 1.4 & 5.9 + hgi & 4827,32 & 0.5 & 2.3 + nai & 4978,83 & 1.5 & 7.7 + nai & 5149,53 & 0.5 & 2.8 + ni & 5199 & 0.7 & 4.1 + sci & 5351.1 & 3.0 & 19.0 + hgi & 5461 & 6.4 & 43.1 + oi & 5577 & 32.7 & 234.5 + nai & 5683,88 & 2.9 & 22.0 + hgi & 5770,91 & 5.0 & 39.7 + nai & 5893,6 & 4.0 & 33.8 + broad nad&5893 & 14.9 & 126.0 + nai & 6154,61 & 0.9 & 8.7 + oi & 6300 & 18.5 & 191.2 + oi & 6364 & 6.5 & 69.3 + li & 6708 & 0.3 & 3.7 + lllllllsite & date & u & b & v & r & i & reference + 02/06/2005@xmath60 & - & [email protected] & [email protected] & [email protected] & & + 28/08/2005 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & + 28/03/2006 & - & [email protected] & [email protected] & [email protected] & [email protected] & + mean & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & + air - mass corrected & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & + la silla 1978 & & 22.8 & 21.7 & 20.8 & 19.5 & mattila et al .
( 1996 ) + kitt peak 1987 & & 22.9 & 21.9 & & & pilachowski et al .
( 1989 ) + cerro tololo 1987 & 22.0 & 22.7 & 21.8 & 20.9 & 19.9 & walker ( 1987a,88a ) + calar alto 1990 & 22.2 & 22.6 & 21.5 & 20.6 & 18.7 & leinert et al . ( 1995 ) + la palma 1990 - 92 & & 22.5 & 21.5 & & & benn & ellison ( 1998a , b ) + la palma 1994 - 96 & 22.0 & 22.7 & 21.9 & 21.0 & 20.0 & benn & ellison ( 1998a , b ) + mauna kea 1995 - 06 & & 22.8 & 21.9 & & & krisciunas ( 1997 ) + paranal 2000 - 01 & 22.3 & 22.6 & 21.6 & 20.9 & 19.7 & patat et al .
( 2003 ) + mtgraham 2000 - 01 & 22.38 & 22.86 & 21.72 & 21.19 & & taylor et al .
( 2004 ) + cerro pachon 2005 & 22.1 & 22.43 & 21.63 & & 20.3 & walker & schwarz ( 2007 ) + cllwavelength & @xmath65 & @xmath65 + ( nm ) & summer & winter + 380 & [email protected] & [email protected] + 436 & [email protected] & [email protected] + 500 & [email protected] & [email protected] + 671 & [email protected] & [email protected] + 880 & [email protected] & [email protected] + 1020 & [email protected] & [email protected] + night - sky spectrum at the calar alto observatory in the optical wavelength range ( 3700 - 7950 ) , obtained after averaging 10 spectra of 6 moonless nights pointing near the zenit ( black solid - line ) .
the intensity has been scaled to that of the darkest moonless night in the @xmath2-band .
several emission lines are indentified in the spectrum .
the most relevant ones have been labeled with its corresponding name and wavelength .
in addition , the broad - emission band of nai centred at @xmath05900 , and the water vapor meinel bands are clearly identified in the spectrum . for comparison purposes we included the night sky spectrum at the kitt peak observatory derived by massey & foltz ( 2000 ) , obtained from their webpage : _ http://www.lowell.edu/users/massey/nightsky.html _ ( orange dotted - line ) .
it is appreciated how strong are the pollution lines at calar alto , in comparison with that observatory . , title="fig : " ] distribution of the v - band extinction along 1044 astronomical useful nights in the period between march 2003 and may 2007 ( @xmath070% of the total nights ) .
the extinction shows a clear seasonal pattern , with a maximum in summer and a minimun in winter .
the median value of the extinction is @xmath36=0.18 mags , with a typical value of @xmath00.15 mags in winter .
the fraction of nights with @xmath430.25 mags , @xmath020% , is similar to other observatories ( e.g. , la palma ) , although the peaks of extinction are much lower .
, title="fig : " ] extinction curves derived for a set of winter ( left panel ) and summer nights ( right panel ) , using the excalibur extintion curve monitor ( red solid circles ) .
the error bars indicate the standard deviation over the mean values .
the data were fitted to a linear combination of the three different contributions to the extinction described in walker ( 1987 ) : the rayleigh scattering ( dashed line ) , the aerosol or dust extinction ( dotted - dashed line ) , and the ozone extinction ( dotted line ) , each one scaled to their relative contribution to the extinction .
the blue solid line shows the best fit to the data .
, title="fig : " ] extinction curves derived for a set of winter ( left panel ) and summer nights ( right panel ) , using the excalibur extintion curve monitor ( red solid circles ) .
the error bars indicate the standard deviation over the mean values .
the data were fitted to a linear combination of the three different contributions to the extinction described in walker ( 1987 ) : the rayleigh scattering ( dashed line ) , the aerosol or dust extinction ( dotted - dashed line ) , and the ozone extinction ( dotted line ) , each one scaled to their relative contribution to the extinction .
the blue solid line shows the best fit to the data .
, title="fig : " ] distribution of the seeing along the time for the 335 nights with measurements from the dimm monitor in the period between january 2005 and december 2006 .
there is a suggestion for a seasonal pattern , with a minimun during summer and a maximum during winter .
, title="fig : " ] normalized histogram of the seeing distribution shown in fig.[seeing1 ] ( solid line ) .
the median value of the seeing is @xmath00.90@xmath8 , with a @xmath046% of the nights with a seeing better than this value .
the red dashed line shows a similar histogram for the summer values ( may to september ) , when the median seeing drops to @xmath00.87@xmath8 .
the blue dashed - dot line shows a similar histogram for the winter values , comprising the remaining data , when the median seeing rise to @xmath00.96@xmath8 .
the seeing distribution is clearly different in both seasons .
, title="fig : " ] | we present a characterization of the main properties of the night - sky at the calar alto observatory for the time period between 2004 and 2007 .
we use optical spectrophotometric data , photometric calibrated images taken in moonless observing periods , together with the observing conditions regularly monitored at the observatory , such as atmospheric extinction and seeing .
we derive , for the first time , the typical moonless night - sky optical spectrum for the observatory .
the spectrum shows a strong contamination by different pollution lines , in particular from mercury lines , which contribution to the sky - brightness in the different bands is of the order of @xmath00.09 mag , @xmath00.16 mag and @xmath00.10 mag in @xmath1 , @xmath2 and @xmath3 respectively .
regarding the strength of the sodium pollution line in comparison with the airglow emission , the observatory does not fulfill the iau recomendations of a dark site .
the zenith - corrected values of the moonless night - sky surface brightness are 22.39 , 22.86 , 22.01 , 21.36 and 19.25 mag arcsec@xmath4 in @xmath5 , @xmath1 , @xmath2 , @xmath3 and @xmath6 , which indicates that calar alto is a particularly dark site for optical observations up to the @xmath6-band .
the fraction of astronomical useful nights at the observatory is @xmath070% , with a @xmath030% of photometric nights .
the typical extinction at the observatory is @xmath70.15 mag in the winter season , with little dispersion . in summer
the extinction has a wider range of values , although it does not reach the extreme peaks observed at other sites .
the analysis of the winter and summer extinction curves indicates that the rayleigh scattering is almost constant along the year .
the rise of the extinction in the summer season is due to an enhance of the aerosol extinction , most probably associated with an increase of dust in the atmosphere .
the median seeing for the last two years ( 2005 - 6 ) was @xmath00.90@xmath8 , being smaller in the summer ( @xmath00.87@xmath8 ) than in the winter ( @xmath00.96@xmath8 ) .
we conclude in general that after 26 years of operations calar alto is still a good astronomical site .
its main properties are similar in many aspects to those of other major observatories where 10m - like telescopes are under operation or construction , being a natural candidate for future large aperture optical telescopes . |
we are grateful to goffredo chirco , tommaso de lorenzo , alejandro perez , and carlo rovelli , for interesting discussions and useful comments . | in quantum statistical mechanics , equilibrium states have been shown to be the typical states for a system that is entangled with its environment , suggesting a possible identification between thermodynamic and von neumann entropies . in this paper
, we investigate how the relaxation toward equilibrium is made possible through interactions that do not lead to significant exchange of energy , and argue for the validity of the second law of thermodynamics at the microscopic scale .
the vast majority of phenomena we witness in every day life are irreversible . from the erosion of cliffs under the repeated onslaughts of the ocean to the erasure of our memories , we experience the unyielding flow of time .
thermodynamics , originally the science of heat engines , allows us to predict simple macroscopic processes such as the melting down of an ice cube forgotten on the kitchen table , by the mean of two laws : the conservation of energy and the growth of entropy . among the wealth of things we are given to perceive ,
there are also a few phenomena showing a high degree of regularity . through the trajectories of planets and stars in the sky , through the swinging of a pendulum clock ,
we encounter the immutability of time .
these led to the development of mechanics and time - reversible dynamical laws .
this description of nature extends below atomic scale provided that quantum variables , represented by non - commuting operators , are introduced .
the revolution initiated in the second half of the @xmath0 century by the founders of statistical mechanics was to understand that thermodynamics emerges from the microscopic structure of matter ; this idea had actually been suggested by d. bernoulli in _ hydrodynamica _ ( 1738 ) .
+ recently , a new approach to the foundations of statistical mechanics , sometimes called typicality , has been proposed @xcite . rather than the usual time or ensemble
averages , this alternative viewpoint finds its roots in the genuinely quantum notion of entanglement . for a composite system in a pure state ,
each component , when taken apart , appears in a probabilistic mixture of quantum states , that is , even individual states can exhibit statistical properties .
for instance , in the case of a system in contact with a heat bath , it has been shown that its density matrix is generically very close to a thermal state @xcite .
this result has been extended to more complicated couplings between the system and its surrounding ( see @xcite for details and proofs ) .
concretely , one considers a system @xmath1 and its environment , subjected to a constraint @xmath2 ( e.g. fixing the total energy ) . in terms of hilbert spaces
, that means @xmath3 .
if the number of degrees of freedom composing the system is small compared to the one of the environment , then for most @xmath4 ( in the sense of the haar measure on the unit sphere ) , one gets @xmath5 in other words , in the framework of typicality , the actual state of the system at some instant of time is very likely to resemble the equilibrium state . for that reason ,
it is legitimate to use the latter to estimate standard statistical quantities like expectation values of observables or von neumann entropy @xmath6 .
however , from this perspective , the formulation of a second law of thermodynamics seems at first sight problematic .
indeed , it is usually taken for granted that fine - grained entropy has to be conserved under the microscopic dynamics imposed by schrdinger s equation , and that thermalization may only be read from a growth in a coarse - grained entropy , such as in boltzmann s @xmath7-theorem . in this work ,
we propose a very simple and practical resolution of this issue , based on the possibility of creating entanglement without exchanging energy ( cf .
@xcite for alternative outlooks on this problem ) . for clarity
, we will focus on a single example , prototype of all irreversible transformations , that is the joule - gay - lussac expansion .
the solution suggested here is nonetheless much more general , and is thought to be relevant for all ordinary thermodynamic systems .
+ the celebrated experiment , initially studied by gay - lussac and used later by joule , has played a central role in the history of thermodynamics .
it consists in the adiabatic free expansion of a gas , initially kept in one side of a thermally isolated container while the other side is empty ( see fig .
[ joule - gay - lussac - exp ] ) .
when the tap separating the two flasks is opened , the gas starts spreading out until homogeneity is achieved .
if the internal energy functional does not depend on the volume , then the initial and final temperatures are found to be equal .
the treatment of the joule - gay - lussac expansion for an ideal gas can be found in all textbooks of thermodynamics and is as follows : there is no exchange of heat nor work between the gas and the outside , and , while the accessible volume is doubled , the entropy of the gas , and therefore of the whole universe , increases by @xmath8 indicating the irreversibility of the process . on the other hand , it is also instructive to have a mechanical description of the joule - gay - lussac expansion .
assuming for simplicity a monoatomic ideal gas , the dynamics of the @xmath9 particles is governed by the hamiltonian @xmath10 where @xmath11 is the potential keeping the particles inside the whole container .
liouville s theorem then ensures the conservation of the volume - form induced by the symplectic structure , that is to say , there can be no loss of information whatsoever at the microscopic scale . shifting to quantum mechanics ,
the evolution is given by the unitary operator @xmath12 where @xmath13 is simply obtained by canonical quantization of the classical hamiltonian .
assuming thermal equilibrium , the initial state for the gas reads @xmath14 where @xmath15 is the potential constraining the particles in one flask only , and @xmath16 is the inverse temperature at which the gas has been prepared .
while the von neumann entropy of the gas initially matches with the thermodynamic one , it is inevitably preserved under the unitary evolution , namely @xmath17 the tension between the irreversible macroscopic thermodynamics and the reversible microscopic mechanics is manifest .
our proposal is that an increase of von neumann entropy is actually allowed by a more careful analysis of what a thermally isolated system is .
+ consider a classical particle arriving with a momentum @xmath18 perpendicularly to a wall initially at rest .
assuming an elastic collision , and given the large ratio between the masses of the wall and the particle , one obtains that after the bounce , the particle has a momentum @xmath19 , while the wall has acquired a momentum @xmath20 . the kinetic energy transferred to the wall being negligible
, the energy of the particle is conserved over the bounce with a very good approximation .
consequently , the classical dynamics of the gas is well described by the effective hamiltonian . from the point of view of quantum mechanics ,
the situation is more delicate .
the total energy and momentum are also conserved during a rebound , so that the state for the composite system made of the particle and the wall evolves from @xmath21 to @xmath22 . by linearity , if the particle started in a superposition @xmath23 , then the unitary transformation associated to the bounce reads @xmath24 and , after tracing out the wall s degrees of freedom , the evolution of the state of the particle becomes @xmath25 which is , in general , not unitary .
hence , the bounce is accompanied with a rise of von neumann entropy .
the model presented above is of course very schematic , the physics of the interaction between the gas and the container , and the initial state of the latter , are in fact much more complicated .
however , it already shows that such an interaction must affect their quantum correlations , and can lead to an increase of von neumann entropy for the gas ( see @xcite for a general result ) . in other words , at the quantum level , the energy distribution of the gas is conserved yet the evolution is not unitary , and the naive canonical quantization of does not strictly apply .
it is now possible to derive the final state for the gas from its quantum mechanical description alone .
the space @xmath26 of accessible states is such that the energy distribution for the gas remains unchanged ( for instance , @xmath27 ) . after a transient phase , characterized by a relaxation time @xmath28 , during which the gas and its environment get more entangled ,
the state of the universe @xmath29 can be seen as a generic unit vector in @xmath26 . following the conclusion of typicality
, we find that the gas reaches a new equilibrium state , namely @xmath30 as a consequence , after a long - enough time , the variation of von neumann entropy of the gas matches the increase of entropy expected from thermodynamics .
+ let us now have a quick look at a slightly different experiment : the adiabatic reversible expansion of a gas pushing a piston .
contrary to the joule - gay - lussac expansion , the temperature of the gas is decreasing as work is performed on the piston . during this process
, the space @xmath26 of accessible states is changing , but its dimension remains constant since the size of the container and the thermal de broglie wavelength are increasing in the same way .
later , if the piston is slowly pushed back , the gas reverts to its initial temperature . in this transformation
, reversibility can not be understood as the possibility to return to the initial microstate , which is practically impossible as soon as a few particles are involved , but only as conservation of von neumann entropy . on the contrary ,
when a constraint initially applied to the system is suddenly relaxed , like in joule - gay - lussac s experiment , the space of accessible states is enlarged . the new states , of possibly higher entropy , can only be reached through interaction with the environment .
however , this interaction does not need to involve significant energy transfer , the quantum system remains thermally isolated , and the growth of entanglement is responsible for the irreversibility of its evolution .
+ let us mention that the idea of varying entropy without exchanging energy has recently been suggested as a solution to the information loss paradox in black holes evaporation @xcite .
more precisely , at the very last stage of its life , a black hole is believed to have lost almost all its mass in the form of hawking s radiation , leaving a planckian size object and a matter field in a thermal state , with an entropy proportional to the initial black hole area .
hence , it has been argued that a huge amount of information should be encoded adiabatically into correlations with geometric degrees of freedom , in order to ensure purity of the final quantum state .
+ to summarize , the second law of thermodynamics , characterizing irreversible transformations , is usually considered at a mesoscopic level , by the mean of a coarse - grained entropy , while no information loss nor thermalization is expected to occur for the fundamental degrees of freedom .
on the other hand , the framework of typicality attempts to make use of properties of quantum systems , specifically entanglement , as foundations for equilibrium statistical mechanics .
in particular , it identifies thermodynamic and von neumann entropies . in that context , it is legitimate to raise the issue of thermalization , namely how an isolated system can reach the maximal entropy state .
the observation that , for quantum systems , being thermodynamically isolated is qualitatively very different from being exactly isolated , opens the possibility of nontrivial quantum interactions without significant exchange of energy , and provides a physical ground for the second law at the microscopic scale .
it is worth noting that recent work involving heat baths in squeezed states also points out the role of quantum correlations in thermodynamic processes @xcite .
finally , this approach offers a fresh look at irreversibility .
if the entropy of a system evolving unitarily has to be constant , it is not the case for those of its components .
indeed , the nonextensivity of von neumann entropy allows for subsystems to see their entropy growing , due to a stronger entanglement between them , without any contradiction with the conservation of total information .
that might be of relevance for explaining the arrow of time in cosmology : any region of space is constantly interacting with new degrees of freedom entering its past light cone , and , assuming a separable initial state , new quantum correlations can be established . this viewpoint may help us to understand the emergence , locally , of classical mixtures rather than highly - quantum states , and the increase of entropy for the observable universe . |
understanding when and how the most massive galaxies in the universe formed is one of the most outstanding problems in cosmology and galaxy formation ( conselice 2007 ) .
a number of observations suggest that the era of @xmath3 is important in galaxy formation and evolution for various reasons : the cosmic star formation rate ( sfr ) density begins to drop at @xmath3 from a flat plateau at higher redshifts ; the morphological type mix of field galaxies changes remarkably at @xmath3 ; the number density of qsos has a peak at @xmath3 ; @xmath12 of the stellar mass assembly of galaxies took place in the redshift range @xmath13 ( dickinson 2003 ; fontana 2003 ; steidel 2004 ; kong 2006 ; richards 2006 ; arnouts 2007 ; pozzetti 2007 ; noeske 2007 ) .
however , in the redshift range @xmath14 , identifiable spectral features such as h&k and the 4000 break ( for passive galaxies ) and [ ] @xmath15 , [ ] @xmath16 , h@xmath17 , and h@xmath18 ( for star - forming ones ) move out of the optical bands , thus near - infrared imaging and spectroscopy become essential .
on the other hand , near - infrared galaxy samples offer several advantages compared to purely optical selections ( cowie 1994 ) .
they allow us to select galaxies at @xmath19 in the rest - frame optical , correspond more closely to a stellar - mass - selected sample , and are less prone to dust extinction .
there are many ways of using optical and infrared ( ir ) colors to select the sample of galaxies at @xmath3 .
a sample of ir - luminous dust - obscured galaxies ( dogs ) , with a median @xmath20 , selected to have very red @xmath21 $ ] color , was recently presented by dey ( 2008 ) . from spectroscopic observations dogs
were found to have a tight redshift distribution around @xmath3 , very similar to that of the submillimeter - selected galaxies ( chapman 2005 ) .
another different sample , based on near - infrared color selection , is that distant red galaxies with @xmath22 ( drgs ; franx 2003 ; van dokkum 2004 ) .
these sources are redder and strong clustering and are believed to be more massive than @xmath23 ( labbe 2005 ; papovich 2006 ; van dokkum 2006 ; quadri 2008 ) . a third sample uses _
spitzer_/irac color to select galaxies at @xmath24 . in the redshift range @xmath25 , the four irac bands
probe the rest - frame near - infrared bands where galaxy spectral energy distributions ( seds ) have similar shape , thus the irac colors are very robust in determining redshift in this range ( farrah 2008 ; desai 2009 ; huang 2009 ; fadda 2010 ; fiolet 2010 ; fang 2011 ) .
most sources of these samples are starburst - dominated ultraluminous infrared galaxies ( ulirgs ; @xmath26 ) , with an apparent 1.6 stellar bump in the irac channels . an independent way to select galaxies at @xmath3 is to use @xmath27 colors .
daddi ( 2004 ) introduced a new photometric technique , @xmath28 and @xmath29 , for obtaining a virtually complete sample of star - forming ( @xmath30 ) and passively evolving galaxies ( @xmath31 and @xmath32 ) at @xmath33 .
these criteria are reddening independent for star - forming galaxies in the selected redshift range , thus can be used to select the reddest dust - extinguished galaxies .
this should allow a relatively unbiased selection of star - forming ( ) and passively evolving galaxies ( ) at @xmath3 within the magnitude limit of the sample studied . based on @xmath34 photometry from subaru and new technology telescope ( ntt ) over two separate fields , using the @xmath27 technique , kong ( 2006 )
have obtained a sample of @xmath35 and @xmath36 , which were identified over an area of @xmath37 arcmin@xmath38 to @xmath39 , of which 320 arcmin@xmath38 are complete to @xmath40 .
they found that the log of the number counts of flattens out by @xmath41 .
similar conclusions were also found by lane ( 2007 ) and mccracken ( 2010 ) , with much larger samples .
investigations of the physical properties ( i.e. , stellar mass , luminosity , size , clustering , sfr , morphology , etc . ) of various types of galaxies from high to low redshift are required for better understanding the history of the galaxy formation and evolution .
dunne ( 2009 ) studied the star formation history of _ k_-selected galaxies in the ukidss ultra - deep survey .
they found that the specific star formation rate ( ssfr ) for _ k_-selected sources rises strongly with redshift at all stellar masses . in the same field , williams ( 2009 , 2010 ) investigated the physical properties ( including clustering , size , stellar mass , surface density , sfr , color , etc . ) of quiescent and star - forming galaxy populations to @xmath4202 with pure photometric data , employing a novel rest - frame @xmath43 versus @xmath44 technique ( wuyts 2007 ; wolf 2009 ; balogh 2009 ; whitaker 2010 ; patel 2011 ) .
as shown in figure 8 of williams ( 2009 ) and figure 2 of brammer ( 2011 ) , the red quiescent and star - forming galaxies are found to occupy two distinct population in the rest - frame @xmath43 versus @xmath44 color space .
this bimodal behavior is still seen up to @xmath3 .
at all redshifts , furthermore , massive quiescent galaxies occupy the extreme high end of the surface density distribution and a tight mass size correlation , while star - forming show a broad range of both densities and sizes . on the other hand , the sizes and surface densities of massive quiescent and star - forming galaxies evolve as simple power - law behavior in ( @xmath45 ) , with more massive galaxies exhibiting faster evolution . in the meantime , brammer ( 2009 , 2011 ) , van dokkum ( 2010 ) , and whitaker ( 2011 ) provided further evidence for the results above , using the sample from the newfirm medium band survey ( nmbs ; van dokkum 2009 ) .
more studies for the _ k_-selected galaxies have been performed up to @xmath46 ( cimatti 2002 ; abraham 2004 ; drory 2005 , 2009 ; fontana 2006 ; pozzetti 2007 ; prez - gonzlez 2008 ; furusawa 2011 ) .
morphologies are also essential in studying galaxy mass assembly history and evolution . for the morphologies of galaxies at @xmath3 , the traditional hubble sequence of regular spirals and elliptical galaxies
has not settled into place by @xmath3 , and a much higher frequency of diffuse , clumpy , and irregular structures is observed among star - forming systems ( lotz 2006 ; szomoru 2011 ) . with the advent of _ hubble space telescope _
_ hst_/wide field camera 3 ( wfc3 ) , it is now possible to obtain a complete rest - frame optical morphological census for massive galaxies at @xmath3 .
understanding the nature of active galactic nuclei ( agns ) and the galaxies that host them is important for such diverse goals as pinpointing the sources of the cosmic x - ray and ir backgrounds and deriving the star formation history of the universe .
multi - wavelength surveys are particularly important for the study of agns because their appearance in different wavelength regimes can be quite different . in this work
, we will discuss the fraction of agns in our sample by using the irac band color criteria of stern ( 2005 ) and mid - ir spectral index ( barmby 2006 ; park 2010 ) .
to study the physical properties of galaxies in the redshift range @xmath47 , in this paper , we select a sample of passively evolving and star - forming galaxies to @xmath48 in the all - wavelength extended groth strip international survey ( aegis ) data set .
the paper is organized as follows .
we describe the multi - band spatial- and ground - based observations of the aegis field , and introduce data reduction , photometric redshifts , and method for galaxy sample selection in section 2 .
section 3 presents source number counts ( and ) .
section 4 presents the physical properties of ( including the estimate sfr and stellar mass of star - forming galaxies and the stellar mass versus sfr relation ) .
section 5 describes the morphologies of both and in the aegis .
section 6 discusses the fraction of agns in our galaxy sample .
finally , a brief summary is presented in section 7 .
all magnitudes and colors are in the ab system unless stated otherwise . throughout the paper
we adopt the following cosmology : @xmath49[km s@xmath50 mpc@xmath50]@xmath52 , @xmath53 , @xmath54 .
the aegis is a collaborative effort to obtain both deep imaging covering all major wavebands from x - ray to radio and optical spectroscopy over a large area of sky ( @xmath55deg@xmath38 ) with the aim of studying the panchromatic properties of galaxies over the last half of the hubble time .
aegis is targeted on a special area of the sky , called the extended groth strip ( egs ; centered at @xmath17(j2000)@xmath56 , @xmath57(j2000)@xmath58 ) , which has low - extinction , low - galactic , and zodiacal infrared emission , has good schedulability by space - based observations , and has therefore attracted a wide range of deep observations at essentially every accessible wavelength over this comparatively wide field ( davis 2007 ) .
the egs is a public deep multi - wavelength survey ; this field provides an unique combination of area and depth at almost every waveband observable . in this paper , _ chandra_/acis x - ray , cfht /
megacam legacy survey optical , cfht / cfh12k optical , palomar / wirc near - infrared , _
spitzer_/irac mid - infrared , and _ spitzer_/mips far - infrared imaging data or catalog of the egs is used .
davis ( 2007 ) provided more details on the data reduction and its performance . for our analysis ,
specially , the _ chandra_/acis dr2 catalog presented by laird et al.(2009 ) is used .
the x - ray data reduction was performed using the ciao data analysis software version 3.3 .
basic data reduction proceeded in a manner similar to that described by nandra ( 2005 ) .
the vast majority of x - ray sources detected in the egs field are agns . in the meantime , the egs is also one of four @xmath59deg@xmath38 fields covered by the cfhtls deep survey ( labeled cfhtls - d3 ) .
the data covers the observed wavelength range 3500@xmath609400 in the @xmath61 , @xmath62 , @xmath63 , @xmath64 , @xmath65 filters ( hereafter labeled @xmath66 , @xmath0 , @xmath67 , @xmath68 , @xmath1 ) .
the limiting ab magnitudes is @xmath69 , @xmath70 , @xmath71 , @xmath72 , @xmath73 in the current data set ( corresponding to the magnitude limit at @xmath74 point - source detection ) .
near - infrared data catalog in @xmath75 and @xmath2 bands are available over @xmath76 deg@xmath38 ( _ k_-band ) with the @xmath74 magnitudes limits of @xmath77 and @xmath78 ( bundy 2006 ) .
deep observations of the egs with _ spitzer_/infrared array camera ( irac ) cover an area of @xmath79 deg@xmath38 to a 50% completeness limit of @xmath80 at @xmath81 ( barmby 2008 ) .
the catalog comprises 57,434 objects detected at @xmath81 , with 84% , 28% , and 24% also detected at @xmath82 , @xmath83 , @xmath84 .
as for far - infrared data sets , we also use the _ spitzer_/multiband imaging photometer for _ spitzer _ ( mips ) @xmath85 imaging data with the @xmath74 flux limit of @xmath86 over @xmath87 deg@xmath38 .
sources are identified and photometry extracted with point - spread function ( psf ) fitting using the daophot software ( stetson 1987 ) .
a catalog of objects with redshifts known from spectroscopic observations is very important to the construction of a photometric redshift code .
for this reason , we use the deep2 ( deep extragalactic evolutionary probe ) spectroscopic data obtained with the deep imaging multi - object spectrograph boarded on the keck 10 m telescope .
the deep2-dr3 redshift catalog structures contain 24 tags for each object with 50,386 entries in the entire catalog .
for the egs field ( deep2-field1 ) , we only consider objects with redshift quality greater than or equal to 3 .
to specifically study the accuracy of the photometric redshift , we also adopt spectroscopic redshifts drawn from the lyman break galaxy ( lbg ; @xmath88 ) catalog of steidel ( 2003 ) and from a small sample of ulirgs ( @xmath3 ) presented in huang ( 2009 ) and fang ( 2011 ) .
the nmbs employs a new technique of using medium - bandwidth near - infrared filters to sample the balmer/4000 break from @xmath89 at a higher resolution than the standard broadband near - infrared filters ( van dokkum 2009 ) , thereby improving the accuracy of photometric redshifts .
a custom set of five medium - bandwidth filters in the wavelength range of @xmath90 were fabricated for the newfirm camera on the mayall 4 m telescope on kitt peak for the nmbs ( see figure 1 of whitaker 2011 ) .
the survey targets two fields within the cosmos ( scoville 2007 ) and aegis ( davis 2007 ) surveys , chosen to take advantage of the wealth of publicly available ancillary data over a broad wavelength range .
the more details of the observations , data reduction , and photometry for the survey can be found in whitaker ( 2011 ) .
the nmbs catalogs contain @xmath91 galaxies at @xmath92 with accurate photometric redshifts and rest - frame colors . in this paper
, we will use their photometric redshifts in the aegis field .
source extraction for the cfhtls optical images is performed with ( bertin & arnouts 1996 ) in the dual image mode ( @xmath68 band is used as the reference image ) ; a 2@xmath93diameter aperture is used for aperture magnitudes .
the edges of the resulting final images , where the signal to noise of cfhtls / megacam images is very low , are trimmed .
we finally cross - correlate the @xmath68-band catalog ( that included @xmath66 , @xmath0 , @xmath67 , and @xmath1 photometry ) with the @xmath2-band catalog , with a matching radius 2@xmath93 .
the final overlapping area of all our data sets is @xmath94deg@xmath38 , and 16,228 objects are included in our final catalog ( @xmath95 mag ) . comparing to the nmbs in the aegis field ( @xmath96deg@xmath38 )
, we have larger area and more faint galaxies ( see section 2.4 ) .
stellar object are isolated with the color criterion @xmath97 , similar as daddi ( 2004 ) .
a total of 1847 objects are classified as stars , and the number of the final galaxy sample is 14,381 .
all magnitudes are corrected for galactic extinction . an sed fitting technique based on an updated version of the hyperz code ( bolzonella et al .
2000 ) is used to calculate the photometric redshifts ( hereafter @xmath98 ) for all the @xmath48 galaxies in the egs .
a set of templates generated by brammer & van dokkum ( 2007 ) for their eazy photometric redshift code are used . for each object , the sed derived from the observed magnitudes is compared to each template spectrum in turn , and the weighted mean redshift , computed in the confidence intervals at 99% around the main solution , from hyperz , is calculated . to check the accuracy of the photometric redshift we derived , we first compare the @xmath98 with the spectroscopic redshifts . for the 14,381 galaxies in our sample , 2940 of them
have spectroscopic redshifts ( @xmath99 ) from the deep2-dr3 redshift catalog , which have a confidence level greater or equal to 97% . in the meantime ,
spectroscopic redshifts are also available for 35 lbgs at @xmath88 and 14 ulirgs at @xmath3 within the egs field ( 14 ulirgs presented in huang 2009 and fang 2011 , 35 lbgs drawn from the catalog of steidel 2003 ) . in figure [ fig : zphot](a ) , we show a comparison of the photometric redshift with the spectroscopic redshift for those 2989 galaxies .
filled triangles and stars correspond to ulirgs and lbgs with spectroscopic redshifts , respectively . from this figure
, we find that our photometric redshifts are in good agreement with the spectroscopic redshifts , with an average @xmath100 .
only 128 of 2989 galaxies have @xmath101 ; the percentage of catastrophic failures is about 4.3% .
the normalized median absolute deviation , @xmath102 , for our galaxy sample is @xmath103 . on the other hand , whitaker ( 2011 )
described the full details of the observations , data reduction , and photometry from the nmbs in the aegis field ( @xmath96deg@xmath38 ) , also presented a public _ k_-selected photometric catalog along with accurate photometric redshifts and rest - frame colors . for this catalog
, we will use their photometric redshifts ( @xmath104 ) .
we finally cross - correlate our catalog with the nmbs catalog in the aegis field , with a matching radius 1@xmath93 .
a total of 5533 galaxies are located in the overlapping area between nmbs and aegis . in figure
[ fig : zphot](b ) , our photometric redshifts ( @xmath105 ) are in good agreement with nmbs photometric redshifts ( @xmath104 ) in the aegis field , although there are some large dispersion .
to select a catalog of objects at @xmath3 , we use the @xmath27 selection technique which was introduced by daddi ( 2004 ) . this technique has been adopted and tested in several subsequent studies ( e.g. , kong 2006 ; lane 2007 ; blanc 2008 ; dunne 2009 ; mccracken 2010 ; onodera 2010 ; yoshikawa 2010 ) . to make the comparison possible with previous studies , we wanted our photometric selection criterion to match as closely as possible as the original @xmath27 selection introduced in daddi ( 2004 ) .
figure [ fig : samp](a ) shows the transmission profiles of the vlt-@xmath27 filters , which were used in daddi ( 2004 ) , the transmission profiles of the cfht-@xmath106 , and the palomar-@xmath2 filters of aegis , which are used by this paper . from this figure
, we find that the shapes of those filters are different .
to account for the differences , we carefully compare the stellar sequence in the aegis to that of daddi ( 2004 ) , with the similar procedure outlined in kong ( 2006 ) , and apply small color terms to @xmath107 and @xmath108 , @xmath109 in aegis and in daddi ( 2004 ) , in order to obtain a fully consistent match . from this analysis
, we obtain our new color criteria @xmath108 and @xmath109 to select @xmath3 galaxies : @xmath110 can be used to select @xmath24 star - forming galaxies ( hereafter named ) , @xmath111 and @xmath112 can be used to isolate @xmath24 passively evolving galaxies ( hereafter named ) . to further elucidate the validity of these @xmath113 criteria
, we generate a set of templates , with different star formation history , by bruzual & charlot ( 2003 ) stellar population synthesis models , and find that galaxies at @xmath114 with ongoing star formation indeed lie in the @xmath115 region as expected , passively evolving galaxies lie in the @xmath111 and @xmath112 region of the @xmath108 and @xmath109 diagram . figure [ fig : samp](b ) shows the @xmath113 color diagram of @xmath2-selected objects in the aegis . using the color criterion @xmath116 ( the black solid line ) ,
1609 galaxies with @xmath48 are selected as ( blue squares ) , which occupy a narrow range on the left of the solid line in figure [ fig : samp](b ) . using @xmath111 and @xmath112 ( dot - dashed line ) ,
422 objects are selected as candidate ( red circles ) , which lie in the top - right part of figure [ fig : samp](b ) .
the sample of gzks ( and ) with open circles from the overlapping area between nmbs and aegis .
these gzks mostly lie along the boundary with non - gzk objects , comparing to the photometric redshift distribution for gzks in our sample , their photometric redshifts have lower values .
the surface density of @xmath48 @xmath117 is @xmath118 arcmin@xmath119 , which is in good agreement with those presented in daddi ( 2004 ) , about @xmath120 arcmin@xmath119 , and kong ( 2006 ) , about @xmath121 arcmin@xmath119 . in figure
[ fig : reds](a ) , we show the histogram of the photometric redshift for all @xmath2-selected galaxies in our sample . to check
how efficient is our @xmath113 color criteria in singling out @xmath24 galaxies , the cross - hatched region shows all non - gzk galaxies in our sample . from this figure , we find that only 4% of galaxies with photometric redshift @xmath24 can not be selected by the @xmath113 color criteria , which means the @xmath113 criteria are quite efficient in singling out high - redshift galaxies . figure [ fig : reds](b ) shows the distribution of the photometric redshifts of , with a median value @xmath122 .
most of them have redshifts at the range @xmath123 , only @xmath124 of them with @xmath125 and @xmath126 with @xmath127 .
we show the redshift distribution of in figure [ fig : reds](c ) ; it clearly shows that there is a considerable fraction ( 12.8% ) of objects at low redshift ( @xmath128 ) .
the median redshift value of is @xmath129 , which is similar to that ( @xmath130 ) of mccracken ( 2010 ) .
figure [ fig : cou ] shows the differential number counts of ( left panel ) and ( right panel ) . in both panels of figure .
[ fig : cou ] , results from the deep3a - f ( asterisks ) and daddi - f ( open diamonds ) from kong ( 2006 ) , ukidss ( open squares , lane 2007 ; crosses , hartley 2008 ) , musyc ( open triangles , blanc 2008 ) , and cosmos ( open circles , mccracken 2010 ) are also plotted for comparison . figure [ fig : cou](a ) shows the number counts of , no correction for incompleteness is applied to our data . as shown in the figure ,
our number counts of in the egs field are in good agreement with the number counts of @xmath3 star - forming galaxies in kong ( 2006 ) , blanc ( 2008 ) , and mccracken ( 2010 ) .
however , the counts presented by lane ( 2007 ) are offset compared to our counts , which is also found by mccracken ( 2010 ) , and the cause for this discrepancy may arises from an incorrect transformation to daddi s ( 2004 ) filter set in lane ( 2007 ) .
figure [ fig : cou](b ) shows the number counts of .
our number counts of , in general , are in good agreement with those presented in the literature for @xmath24 passively evolving galaxies . at bright magnitudes ( e.g. , @xmath131 ) our counts are in very good agreement with the pbzks counts from mccracken ( 2010 ) .
however , the number counts of ours and those of mccracken ( 2010 ) are slightly below those from blanc ( 2008 ) at faint magnitudes . combining literature data with the present work
, we can examine the global shape of the @xmath24 passively evolving galaxy counts over a very wide dynamic range .
this strongly suggests a break feature in the slope at @xmath131 , which was found in kong ( 2006 ) already .
the dot - dashed lines in figure [ fig : cou](a ) and ( b ) show the number counts from the semi - analytic model of kitzbichler & white ( 2007 ) .
the number counts of quiescent galaxies and star - forming galaxies were derived by mccracken ( 2010 ) , following the approach of daddi ( 2007 ) . from figure
[ fig : cou ] , we find the models overpredict the number of faint galaxies , but underpredict the number of bright galaxies in all two plots . given the narrow redshift range of our @xmath24 galaxy population , apparent @xmath2 magnitude is a good proxy for absolute @xmath2 magnitude , which can itself be directly related to underlying stellar mass ( daddi 2004 ) .
this implies that these models predict too many small low - mass star - forming and passively evolving galaxies , and too few large high - mass star - forming and passively evolving galaxies at @xmath24 .
similar conclusions were drawn by fontana ( 2006 ) and mccracken ( 2010 ) .
therefore , models incorporating agn feedback similar to kitzbichler & white underpredicted the number of high - mass galaxies , however predicted too many low - mass galaxies .
in this section , physical properties of , such as stellar mass ( @xmath132 ) , sfr , and ssfr ( sfr@xmath133 ) , are derived on the basis of the present photometric data . in order to estimate the stellar mass ( @xmath132 ) of , we perform sed fitting of the multi - band photometry as described above with population synthesis models .
the sed templates are generated with the sps package developed by bruzual & charlot ( 2003 ) .
we assume an universal initial mass function ( imf ) from chabrier ( 2003 ) .
masses derived assuming this imf can be converted to salpeter by adding 0.3 dex ( bundy 2005 ) .
a grid of model templates spanning a range of star formation histories ( parameterized as an exponential ) , ages , metallicities and dust content are used .
the stellar mass - to - luminosity ( @xmath134 ) ratios from the best - fit templates and the photometric redshifts ( spectroscopic redshifts are adopted if available ) from section 2.3 are used to calculate the stellar mass .
the histogram for the stellar mass of the is shown in figure [ fig : sfr](a ) .
about 50% of the in our sample have @xmath135 , and the median stellar mass of is @xmath136 . the ultraviolet continuum can be used to estimate the luminosity from young stars and hence the sfr .
we derive the sfr of each based on its rest - frame 2800 luminosity , which is calculated from the same best - fitting sed template as used in the estimation of the photometric redshift .
since our multi - band photometric data completely cover the rest - frame 2800 for the redshift range ( @xmath137 ) of , the estimated rest - frame 2800 luminosity is a robust quantity , especially in the case where a galaxy is detected in both bands which straddle the rest - frame 2800 .
the color excess , @xmath138 , from the sed fitting is used to derive the dust - corrected uv luminosity .
then , we convert the dust - corrected rest - frame 2800 luminosities into sfrs using the calibration by kennicutt ( 1998 ) : @xmath139 .
the sfr histogram of the is shown in figure [ fig : sfr](b ) .
about 85% of in our sample ( @xmath140 ) have sfr @xmath141yr@xmath50 , and the median sfr is about @xmath142 yr@xmath50 , which is similar to the values in daddi ( 2004 ) ( with a typical sfr of @xmath143yr@xmath50 ) and in kong ( 2006 ) ( with a median sfr of @xmath144 @xmath145yr@xmath50 ) .
the existence of a correlation between star formation and stellar mass of galaxies at different redshifts ( @xmath147 ) has been reported recently ( daddi 2007 ; elbaz 2007 ; pannella 2009 ) . to gain further insights into the nature of star formation at @xmath14
, we show the sfr as a function of @xmath132 in the four redshift bins in figure [ fig : sfrm ] .
to examine in detail the stellar mass dependence of the sfr , we plot a good - fit line ( dashed line ) for galaxies with @xmath148 in each redshift panel . from this figure
, we find a positive correlation between the sfr and stellar mass of in the all four redshift bins . in figure
[ fig : sfrm](d ) , we overplot the correlation between the sfr and stellar mass for @xmath149 galaxies in the sloan digital sky survey ( sdss ; brinchmann 2004 ) , @xmath150 and @xmath3 galaxies in the goods ( elbaz 2007 ; daddi 2007 ) , and @xmath151 and @xmath150 from millennium model ( kitzbichler & white 2007 ) .
figure [ fig : sfrm](d ) shows the slopes between the sfr and stellar mass of galaxies at different redshifts are very similar . on average ,
the sfr of galaxies on the sfr@xmath146 diagram increases with redshift at a fixed @xmath132 , by a factor about 6 from @xmath152 to @xmath153 , and a factor about 20 from @xmath152 to @xmath154 . at fixed stellar mass ,
star - forming galaxies were much more active on average in the past .
this is most likely due to a larger abundance of gas , depleted with passing of time ( daddi 2007 ) .
the millennium model overlaps very well with the sdss at @xmath155 and predicts a slope very similar to the observed trend at different redshift bins . for @xmath153 , however , we find that at fixed stellar masses the model predicts an increase of the sfr much below the observed trend .
it seems that a major change required for models would be to increase the star formation efficiency ( and thus the typical sfr ) at all masses for star - forming galaxies at redshifts @xmath19 .
ssfr is defined as the ratio of the current sfr to the current stellar mass , ssfr = sfr@xmath133 .
thus , higher values of the ssfr indicate that a larger fraction of stars were formed recently , and it can be used to characterize the star formation history of galaxies . figure [ fig : ssfrm ] shows the ssfr as a function of stellar mass for the in our sample , with the same four redshift bins as in figure [ fig : sfrm ] .
the dashed lines in each panel represent a good fit between the ssfr and @xmath132 for galaxies with @xmath148 . in figure
[ fig : ssfrm](d ) , we overplot the correlation between the ssfr and stellar mass for @xmath156 , @xmath157 , and @xmath158 galaxies in brinchmann & ellis ( 2000 ) . from this figure , we find that the general shape of ssfr@xmath146 correlation is in good agreement with a similar study for @xmath128 galaxies ( brinchmann & ellis 2000 ) , and the line of maximum ssfr runs parallel to lines of constant sfr .
furthermore , this ssfr is generally increasing with increasing redshift for all stellar masses , suggesting that star formation contributes more to the growth of low - mass galaxies than to the growth of high - mass galaxies .
finally , the most massive galaxies have the lowest ssfr at redshift @xmath159 ; they formed their stars earlier and more rapidly than their low - mass counterparts .
this is in agreement with the downsizing scenario .
the morphology of a galaxy reflects its dynamical history and evolution .
for instance , different hubble types are associated with different star formation histories and different patterns of motion of stars and gas .
furthermore , galaxy morphology correlates with a range of physical properties in galaxies , such as stellar mass , luminosity , size , velocity dispersion , concentration , and , particularly , color ; this suggests that morphology is crucial in our understanding of the formation and evolution of galaxies ( fang 2009 ; kong 2009 ) . with the advent of the new wfc3 on board _ hst _ , with its vastly improved sensitivity and resolution compared to nicmos ,
it has become possible to analyze the rest - frame optical structure of @xmath3 galaxies with an unprecedented level of detail .
to describe clearly the morphologies of both and , we have performed nonparametric measures of galaxy morphology by using data from _ hst _
wfc3/h(f160w ) imaging ( grogin 2011 ; koekemoer 2011 ) , such as gini coefficient ( the relative distribution of the galaxy pixel flux values , or @xmath8 ) and @xmath9 ( the second - order moment of the brightest 20% of the galaxy s flux ) .
early - type galaxies have higher @xmath8 and lower @xmath9 , while late - type galaxies have lower @xmath8 and higher @xmath9 . combining @xmath8 and @xmath9 permits classifying different galaxy population effectively ( lotz 2006 ; kong 2009 ) . in figure
[ fig : mor ] , the blue squares and red circles represent the morphological distribution of 173 sgzks and 50 pgzks ( these sources have @xmath160-band counterparts ) in the rest - frame wavelength @xmath161 , respectively .
the distribution of gzks is very similar to that of local galaxies ( e - sd ) in lotz 2004 , with pgzks showing high @xmath8 and low @xmath9 , and sgzks with lower @xmath8 and higher @xmath9 values .
this figure clearly illustrates the morphological variety present at @xmath3 . the median ( @xmath8 , @xmath9 ) values for sgzks are ( 0.49 , 1.48 ) in the rest - frame optical band , while pgzks are ( 0.64 , 1.74 ) . a sample of 73 local ulirgs ( @xmath162 ) is also shown in this diagram ( lotz 2004 ) , indicated as empty stars , using data from _ hst _ wfpc 2/f814w imaging ( borne 2000 ) .
the solid diamonds and triangles in figure [ fig : mor ] represent the starburst - dominated sources at @xmath3 from bussmann ( 2011 ) and frster schreiber ( 2011 ) , respectively ; their @xmath8 and @xmath9 values were derived from nicmos / f160w images , corresponding to the rest - frame optical wavelength .
the dashed line ( @xmath163 ) in this figure is drawn qualitatively at rest - frame optical band ( @xmath161 ; bussmann 2011 ) .
galaxies with @xmath164 have diffuse structures or multiple bright nuclei in appearance .
objects with @xmath165 and @xmath166 can be irregular , often with a bright nucleus with tidal tail or fainter knots . as for sources with @xmath165 and @xmath167 , they are relatively smooth with a single nucleus . from figure [ fig : mor ] , we find that star - forming galaxies at high redshift not only have diffuse structures , similar to starburst - dominated ulirgs at @xmath3 with lower @xmath8 and higher @xmath9 values , but also have single - object morphologies ( higher @xmath8 and lower @xmath9 ) .
moreover , the distribution of @xmath9 of sgzks is mainly located in the range from 1.0 to 1.8 ; these features of morphologies are consistent with expectations from simulations of major mergers during the beginning and end stages , respectively , of the phase of the merger when the sfr peaks and begins to turn over ( lotz 2008 ) .
figures [ fig : mors ] and [ fig : morp ] show the _
hst_/wfc3 @xmath75- and @xmath160-band stamp images for a subset of six sgzks and six pgzks , respectively .
their rest - frame optical morphologies for sgzks are very diversified including string - like , extended / diffused , and even early - type spiral morphologies , while pgzks are relatively smooth and compact , implying that there are different formation process for these galaxies .
therefore , we conclude that the population of hubble sequence galaxies roughly matches that of the peculiars sometime between @xmath1682 . - and @xmath160-band images of six sgzks in the egs .
top : @xmath169 , middle : @xmath170 , bottom : @xmath171 .
values of @xmath8 and @xmath172 are shown in each @xmath160-band image .
all images are in negative gray scale and are @xmath173 arcsec@xmath38 in size .
[ fig : mors ] ] - and @xmath160-band images of six pgzks in the egs .
top : @xmath174 , middle : @xmath175 , bottom : @xmath176 .
values of @xmath8 and @xmath172 are shown in each @xmath160-band image .
all images are in negative gray scale and are @xmath173 arcsec@xmath38 in size .
[ fig : morp ] ]
mid - ir photometry has been proven to be a robust and efficient tool to select agns without prior information ( e.g. , lacy 2004 ; stern 2005 ; donley 2008 ) , as their properties at these wavelengths are typically very different from those of stars and galaxies .
there have been several studies of agn selection using mid - ir color or other ir properties . in this section ,
two different agn selection methods are employed , to identify agns in our gzk sample .
the first one is based on the irac band color criteria of stern ( 2005 ) .
the second method is based on mid - ir spectral index ( barmby 2006 ; park 2010 ) .
figure [ fig : mid ] shows the mid - ir color space of the egs survey along with the stern ( 2005 ) agn selection criterion ( green dot - dashed lines ) .
small dots represent the distribution of mid - ir color for sources detected in all four irac channels in the aegis field ; they have very similar [ 3.6 ] [ 4.5 ] colors , as this is dominated by their stellar emission .
six hundred and twelve in our sample are detected in all the four irac channels and are plotted as blue open circles .
one hundred and twenty - two of them can be selected as agn candidates by the mid - ir selection criterion of stern ( 2005 ) .
two hundred and sixteen in our sample are detected in all the four irac channels and plotted as red open circles .
thirty - five of them are agn candidates .
therefore , using the mid - ir agn selection criterion , @xmath177 of and @xmath178 of are classified as agns , respectively . the agn selection criterion from stern ( 2005 ) were based on the spectroscopic sample of the agn and galaxy evolution survey , most of which have redshift @xmath179 .
their mid - ir color criterion is reliable for classifying agns and galaxies at low redshift .
in contrast , observations and templates suggest that a high degree of stellar contamination is unavoidable when the color selection technique is applied to deeper samples ( barmby 2006 ; donley 2008 ; georgantopoulos 2008 ; assef 2010 ) .
the redshift of and in our sample is @xmath154 ; with much deeper irac data , we will classify agns from normal galaxies with the mid - ir spectral index method in the next subsection .
the slope of a galaxy sed in the infrared can be characterized by a power - law behavior of flux density with frequency @xmath180 .
agn seds often follow a negative - sloping ( red ) power - law , which may arise from either thermal or non - thermal emission originating near the central region .
in contrast , stellar - dominated sources at redshift @xmath181 generally exhibit positive ( blue ) irac power law emission .
for this reason , mid - ir power - law @xmath17 of galaxies have been proposed and used as classification criteria for agns and normal galaxies , with a small level of galaxy contamination . in this paper ,
the irac fluxes for each gzks , covering the 3.68.0 @xmath182 m range , are fitted with a power law , @xmath17 .
the @xmath17 value is only accepted if the @xmath183 probability fit is @xmath184 .
galaxies fitted well with a power - law sed are further classified as red or blue .
a limit of @xmath185 , following donley ( 2008 ) and park ( 2010 ) , is chosen to classify gzks as agns .
the criterion reveals that the fraction of agns in our sample is @xmath10 10% . in the meantime
, we also check the x - ray luminosities of these agn candidates ( if they have x - ray counterparts ) , and the result shows that most of these objects have @xmath186 .
in this paper , we have described the construction of a sample of ( correspond to the bzks of daddi 2004 ) within the aegis . with this field s unique combination of area and depth at almost every waveband
observable , we analyze the photometric redshifts , number counts , sfrs , ssfrs , stellar masses , morphologies , and the fraction of agns of in the aegis .
our main conclusions are as follows .
we identify a sample of 2031 including 1609 star - forming galaxies , with @xmath187 and 422 passively evolving galaxies with @xmath111 and @xmath112 , to a limit @xmath188 .
the surface density of + is @xmath118 arcmin@xmath119 . of
the sgzk galaxies , 83.2% lie in the range @xmath123 , while 8.5% are at @xmath125 and 8.3% at @xmath127 . for pgzk galaxies
, there is a considerable fraction ( 12.8% ) of objects at low redshift ( @xmath128 ) .
the @xmath189 median values of both and are 1.8 and 1.5 , respectively . \2 . for a sample selected at @xmath4 , 2989 spectroscopic redshift of field galaxies ( including 35 lbgs at @xmath88 and 14 ulirgs at @xmath3 ) ,
we reach an average redshift accuracy of @xmath190 with the normalized median absolute deviation ( @xmath102 ) of 0.032 .
only 4.3% galaxies have @xmath101 .
we also compare our photometric redshifts with nmbs photometric redshifts in the aegis field ; they are in good agreement .
similar to the findings in other literatures , our counts of passively evolving galaxies turn over at @xmath7 and both the number of faint and bright sources ( including and ) in our catalogs exceed the predictions of a recent semi - analytic model of galaxy formation , implying that a more successful model must have to explain this diversity in the future . \4 . based on the reddening - corrected uv luminosities and sed fitting of the multi - band photometry , we find that _ gzk_-selected star - forming galaxies have a median sfr and stellar mass of @xmath191yr@xmath50 and @xmath192 , respectively .
moreover , we also find that the sfr and ssfr of increases with redshift at all masses , suggesting that star - forming galaxies were much more active on average in the past .
comparing to the predictions of the millennium model , there is a major change required for models in order to increase the star formation efficiency at all masses for star - forming galaxies at redshifts @xmath19 .
we also study the morphological properties of gzks in our sample , employing data from _ hst _ wfc3/f160w imaging within the egs field .
we derived the median ( @xmath8 , @xmath9 ) values for sgzks and pgzks are ( 0.49 , 1.48 ) and ( 0.64 , 1.74 ) in the rest - frame optical band , respectively .
moreover , we find that morphologies of @xmath3 galaxies in our sample are complex and varied : from compact , apparently early - type galaxies to large star - forming systems superficially similar to pure disk or irregular galaxies , implying that there are different formation process for these galaxies .
we conclude that the population of hubble sequence galaxies roughly matches that of the peculiars sometime between @xmath1931.52 .
6 . for the fraction of agns in the
, we compared the power - law selection technique with mid - infrared color selection , and we find that 82 sources with @xmath194 , corresponding to the fraction of agns , is @xmath10 10% in our sample of 828 gzk galaxies with four irac bands . in the meantime , we also check the x - ray luminosities of these sources ( if they have x - ray counterparts ) ; the result shows that most of these objects have @xmath186 .
we thank the anonymous referee for valuable comments and suggestions that have improved the paper .
we thank j .- s . huang and h. j. mccracken for helpful suggestions and discussions .
this study makes use of data from aegis , a multi - wavelength sky survey conducted with the _ chandra _ , _ galex _ , _ hubble _ , keck , cfht , mmt , subaru , palomar , _ spitzer _ , vla , and other telescopes and supported in part by the nsf , nasa , and the stfc .
this work is also based on observations taken by the candels multi - cycle treasury program with the nasa / esa _ hst _ , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas5 - 26555 . the work is supported by the national natural science foundation of china ( nsfc , no .
10873012 ) , the open research program of key laboratory for the structure and evolution of celestial objects , cas , and chinese universities scientific fund ( cusf ) .
abraham , r. g. , glazebrook , k. , mccarthy , p. j. , et al .
2004 , , 127 , 2455 arnouts , s. , walcher , c. j. , le fvre , o. , et al .
2007 , , 476 , 137 assef , r. j. , kochanek , c. s. , brodwin , m. , et al .
2010 , , 713 , 970 balogh , m. l. , mcgee , s. l. , wilman , d. , et al .
2009 , , 398 , 754 barmby , p. , alonso - herrero , a. , donley , j. l. , et al . 2006 , , 642 , 126 barmby , p. , huang , j .- s . , ashby , m. l. n. , et al .
2008 , , 177 , 431 bertin , e. , & arnouts , s. 1996 , , 117 , 393 blanc , g. a. , lira , p. , barrientos , l. f. , et al .
2008 , , 681 , 1099 bolzonella , m. , miralles , j .- m .
, & pell , r. 2000 , , 363 , 476 borne , k. d. , bushouse , h. , lucas , r. a. , & colina , l. 2000 , , 529 , 77 brammer , g. b. , & van dokkum , p. g. 2007 , , 654 , l107 brammer , g. b. , whitaker , k. e. , van dokkum , p. g. , et al .
2009 , , 706 , l173 brammer , g. b. , whitaker , k. e. , van dokkum , p. g. , et al . 2011 , , 739 , 24 brinchmann , j. , charlot , s. , white , s. d. m. , et al .
2004 , , 351 , 1151 brinchmann , j. , & ellis , r. s. 2000 , , 536 , l77 bruzual , g. , & charlot , s. 2003 , , 344 , 1000 bundy , k. , ellis , r. s. , & conselice , c. j. 2005 , , 625 , 621 bundy , k. , ellis , r. s. , conselice , c. j. , et al .
2006 , , 651 , 120 bussmann , r. s. , dey , a. , lotz , j. , et al .
2011 , , 733 , 21 chabrier , g. 2003 , , 115 , 763 chapman , s. c. , blain , a. w. , smail , i. , & ivison , r. j. 2005 , , 622,772 cimatti , a. , daddi , e. , mignoli , m. , et al .
2002 , , 381 , l68 conselice , c. j. , bundy , k. , trujillo , i. , et al .
2007 , , 381 , 962 cowie , l. l. , songaila , a. , hu , e. m. , et al .
1994 , , 432 , l83 daddi , e. , cimatti , a. , renzini , a. , et al .
2004 , , 617 , 746 daddi , e. , dickinson , m. , morrison , g. , et al .
2007 , , 670 , 156 davis , m. , guhathakurta , p. , konidaris , n. p. , et al .
2007 , , 660 , l1 desai , v. , soifer , b. t. , dey , a. , et al .
2009 , , 700 , 1190 dey , a. , soifer , b. t. , desai , v. , et al .
2008 , , 677 , 943 dickinson , m. , papovich , c. , ferguson , h. c. , & budavri , t. 2003 , , 587 , 25 donley , j. l. , rieke , g. h. , prez - gonzlez , p. g. , & barro , g. 2008 , , 687 , 111 drory , n. , bundy , k. , leauthaud , a. , et al .
2009 , , 707 , 1595 drory , n. , salvato , m. , gabasch , a. , et al .
2005 , , 619 , l131 dunne , l. , ivison , r. j. , maddox , s. , et al .
2009 , , 394 , 3 elbaz , d. , daddi , e. , le borgne , d. , et al .
2007 , , 468 , 33 fadda , d. , yan , l. , lagache , g. , et al .
2010 , , 719 , 425 fang , g .- w . , huang , j .- s . ,
kong , x. , et al .
2011 , , submitted fang , g .- w . ,
kong , x. , & wang , m. 2009 , research in astronomy and astrophysics , 9 , 59 farrah , d. , lonsdale , c. j. , weedman , d. w. , et al .
2008 , , 677 , 957 fiolet , n. , omont , a. , lagache , g. , et al .
2010 , , 524 , a33 fontana , a. , donnarumma , i. , vanzella , e. , et al . 2003 , , 594 , l9 fontana , a. , salimbeni , s. , grazian , a. , et al .
2006 , , 459 , 745 frster schreiber , n. m. , shapley , a. e. , erb , d. k. , et al .
2011 , , 731 , 65 franx , m. , labb , i. , rudnick , g. , et al .
2003 , , 587 , l79 furusawa , j. , sekiguchi , k. , takata , t. , et al .
2011 , , 727 , 111 georgantopoulos , i. , georgakakis , a. , rowan - robinson , m. , & rovilos , e. 2008 , , 484 , 671 grogin , n. a. , kocevski , d. d. , faber , s. m. , et al .
2011 , , 197 , 35 hartley , w. g. , lane , k. p. , almaini , o. , et al .
2008 , , 391 , 1301 huang , j .- s . ,
faber , s. m. , daddi , e. , et al .
2009 , , 700 , 183 kennicutt , r. c. , jr .
1998 , , 36 , 189 kitzbichler , m. g. , & white , s. d. m. 2007 , , 376 , 2 koekemoer , a. m. , faber , s. m. , ferguson , h. c. , et al . 2011 , , 197 , 36 kong , x. , daddi , e. , arimoto , n. , et al .
2006 , , 638 , 72 kong , x. , fang , g .- w .
, arimoto , n. , & wang , m. 2009 , , 702 , 1458 labb , i. , huang , j. , franx , m. , et al .
2005 , , 624 , l81 lacy , m. , storrie - lombardi , l. j. , sajina , a. , et al .
2004 , , 154 , 166 laird , e. s. , nandra , k. , georgakakis , a. , et al .
2009 , , 180 , 102 lane , k. p. , almaini , o. , foucaud , s. , et al .
2007 , , 379 , l25 lotz , j. m. , jonsson , p. , cox , t. j. , & primack , j. r. 2008 , , 391 , 1137 lotz , j. m. , madau , p. , giavalisco , m. , primack , j. , & ferguson , h. c. 2006 , , 636 , 592 lotz , j. m. , primack , j. , & madau , p. 2004
, , 128 , 163 mccracken , h. j. , capak , p. , salvato , m. , et al .
2010 , , 708 , 202 nandra , k. , laird , e. s. , adelberger , k. , et al .
2005 , , 356 , 568 noeske , k. g. , weiner , b. j. , faber , s. m. , et al .
2007 , , 660 , l43 onodera , m. , arimoto , n. , daddi , e. , et al .
2010 , , 715 , 385 pannella , m. , carilli , c. l. , daddi , e. , et al .
2009 , , 698 , l116 papovich , c. , moustakas , l. a. , dickinson , m. , et al .
2006 , , 640 , 92 park , s. q. , barmby , p. , willner , s. p. , et al .
2010 , , 717 , 1181 patel , s. g. , kelson , d. d. , holden , b. p. , franx , m. , & illingworth , g. d. 2011 , , 735 , 53 prez - gonzlez , p. g. , rieke , g. h. , villar , v. , et al .
2008 , , 675 , 234 pozzetti , l. , bolzonella , m. , lamareille , f. , et al .
2007 , , 474 , 443 quadri , r. f. , williams , r. j. , lee , k .- s . , et al .
2008 , , 685 , l1 richards , g. t. , strauss , m. a. , fan , x. , et al . 2006 , , 131 , 2766 scoville , n. , aussel , h. , brusa , m. , et al .
2007 , , 172 , 1 steidel , c. c. , adelberger , k. l. , shapley , a. e. , et al .
2003 , , 592 , 728 steidel , c. c. , shapley , a. e. , pettini , m. , et al .
2004 , , 604 , 534 stern , d. , eisenhardt , p. , gorjian , v. , et al .
2005 , , 631 , 163 stetson , p. b. 1987 , , 99 , 191 szomoru , d. , franx , m. , bouwens , r. j. , et al .
2011 , , 735 , l22 van dokkum , p. g. , franx , m. , frster schreiber , n. m. , et al .
2004 , , 611 , 703 van dokkum , p. g. , labb , i. , marchesini , d. , et al .
2009 , , 121 , 2 van dokkum , p. g. , quadri , r. , marchesini , d. , et al .
2006 , , 638 , l59 van dokkum , p. g. , whitaker , k. e. , brammer , g. , et al .
2010 , , 709 , 1018 whitaker , k. e. , labb , i. , van dokkum , p. g. , et al .
2011 , , 735 , 86 whitaker , k. e. , van dokkum , p. g. , brammer , g. , et al .
2010 , , 719 , 1715 williams , r. j. , quadri , r. f. , franx , m. , van dokkum , p. g. , & labb , i. 2009 , , 691 , 1879 williams , r. j. , quadri , r. f. , franx , m. , et al .
2010 , , 713 , 738 wolf , c. , aragn - salamanca , a. , balogh , m. , et al .
2009 , , 393 , 1302 wuyts , s. , labb , i. , franx , m. , et al .
2007 , , 655 , 51 yoshikawa , t. , akiyama , m. , kajisawa , m. , et al .
2010 , , 718 , 112 | using a simple two - color selection based on @xmath0- , @xmath1- , and @xmath2-band photometry , we pick out 1609 star - forming galaxies ( ) and 422 passively evolving galaxies ( ) at @xmath3 from a @xmath2-band- selected sample ( @xmath4 ) in an area of @xmath5 deg@xmath6 of the all - wavelength extended groth strip international survey .
the number counts of in our sample turn over at @xmath7 , and both the number of faint and bright objects ( including and ) exceed the predictions of a recent semi - analytic model of galaxy formation , a more successful model is need to explain this diversity .
we also find that the star formation rate ( sfr ) and specific sfr ( ssfr ) of increases with redshift at all masses , implying that star - forming galaxies were much more active on average in the past .
moreover , the ssfr of massive galaxies is lower at all redshifts , suggesting that star formation contributes more to the mass growth of low - mass galaxies than to high - mass galaxies . from _ hubble space telescope _
wide field camera 3 near - infrared imaging data , we find that morphologies of @xmath3 galaxies not only have diffuse structures with lower @xmath8 and higher @xmath9 values , but also have single - object morphologies ( higher @xmath8 and lower @xmath9 ) , implying that there are morphological variety and different formation process for these galaxies at @xmath3 .
finally , we also study the fraction of active galactic nuclei ( agns ) in the , 82 of 828 with four irac bands can be classified as agns ( @xmath10 10% ) .
most of these agn candidates have @xmath11 . |
* ( 1 ) the failure of the bondi accretion feedback * in several papers and talks we showed that the bondi accretion fails to explain feedback heating in cooling flows ( cfs ) in clusters of galaxies ( soker 2006 ) . for more arguments
the reader should consult that paper .
here we only point to the two main reasons for the failure of a feedback heating based on the bondi accretion . _
( i ) there is no time for the feedback to work .
_ in the bondi accretion process the inflow velocity at the bondi radius @xmath0 is about the sound speed @xmath1 , with typical values for cf clusters . from mass conservation
the inflow velocity varies as @xmath2 ( because the icm density profile is flat in the inner regions of clusters ) , and its typical value at @xmath3 is @xmath4 .
the time it takes for the gas to flow inward from a distance of @xmath5 and feed the smbh is @xmath6 this time is longer than the cooling time of most cf clusters at @xmath3 .
the situation at @xmath7 is hopeless , as the inflow time is more than the age of the cluster . _
( ii ) insufficient heating in clusters . _
most supporters of feedback by bondi accretion cite the paper by allen et al .
. however , allen et al .
( 2006 ) , noted that their results can not be extrapolated to cfs in cluster of galaxies .
most likely , the bondi accretion process can not explain the feedback heating in the process of galaxy formation as well .
* ( 2 ) the cold feedback mechanism * in the cold - feedback model ( pizzolato & soker 2005 ; soker 2006 , 2008b ) mass accreted by the central smbh originates in non - linear over - dense blobs of gas residing in an extended region of @xmath8 ; these blobs are originally hot , but then cool faster than their environment and sink toward the center ( see also revaz et al .
the mass accretion rate by the central black hole is determined by the cooling time of the icm , the entropy profile , and the presence of inhomogeneities ( soker 2006 ) .
most important , the icm entropy profile must be shallow for the blobs to reach the center as cold blobs .
wilman et al .
( 2008 ) suggest that the behavior and properties of the cold clumps they observe in the cluster a1664 support the cold feedback mechanism .
the cold feedback mechanism has the following consequences .
cooling flows do exist , but at moderate mass cooling rates .
the _ moderate cooling flow model _ ( soker et al . 2001 ) .
the cold feedback mechanism explains why real clusters depart from an ` ideal ' feedback loop that is 100% efficient in suppressing cooling and star formation .
simply , the feedback requires that non - negligible quantities of mass cool to low temperatures .
part of the mass falls to small radii .
part ( likely most ) of the inflowing cold gas is ejected back from the very inner region .
this is done by the original jets blown by the smbh ( soker 2008a ) .
the ejection of this gas is done in a slow massive wide ( smw ) bipolar outflow , which are actually two jets .
the most striking example of an smw outflow is presented in the seminal work of moe et al .
( 2009 apj submitted ) . by conducting a thorough analysis , moe et al .
( 2009 ) find the outflow from the quasar sdss j0838 + 2955 to have a velocity of @xmath9 , and a mass outflow rate of @xmath10 , assuming a cover fraction of @xmath11 .
the cooling and ejection back of large quantities of mass , make the feedback process not only of energy ( heating ) , but also of mass ( pizzolato & soker 2005 ) . as we showed in a series of five papers
, such smw jets can inflate the ` fat ' bubbles that are observed in many cfs , in clusters , groups of galaxies , and elliptical galaxies .
the implication is that narrow radio jets that are observed in many cases are not the main source of the energy inflating the bubbles .
we elaborate on this in the last section .
the same mechanism that form smw jets in cf clusters and galaxies , can expel large quantities of mass during galaxy formation , and might explain the smbh - bulge mass correlation ( soker 2009 ) .
in a series of five papers we conducted 2.5d hydrodynamical numerical simulations , and for the first time managed to systematically inflate ` fat ' bubbles ( i.e. , large and almost spherical bubbles ) as observed in cf clusters .
main results of these papers are as follows . *
( 1 ) * fat bubbles can be inflated by smw jets ( sternberg et al .
2007 ; sternberg & soker 2008b , 2009a , b ) .
typical values are : initial jet speed@xmath12 ; mass loss rate into the two jets@xmath13 ; a large half opening angle of @xmath14 . *
( 2 ) * rapidly precession narrow jets have the same effect as wide jets ( sternberg & soker 2008a , 2009a ) , and can also explain the formation of ` fat ' bubbles . *
( 3 ) * bubbles are known to be stable during the inflation phase ( pizzolato & soker 2006 ) .
the reason is that the bubble s boundary is decelerated by the denser icm gas , and hence it is actually rayleigh - taylor _ stable _ during the inflation phase .
when we inflate bubbles self - consistently with jets ( as it should be done , and as we did in all our simulations ) , we further find that the bubbles stay intact for a much longer time after the inflation phase ends ( sternberg & soker 2008b )
. basically , the dense shell moving outward ahead of the bubble ( formed during the inflation phase ) and the vortexes formed inside the bubbles , stabilize the bubble .
these effects can not be studied with bubbles that are artificially injected to the icm . * ( 4 ) * every bubble inflation episode excites multiple sound waves and shocks ( sternberg & soker 2009a ) .
such sound waves are observed in some cf clusters , most clearly in perseus ( fabian et al .
( 5 ) * mixing of very - hot bubble material with hot icm gas can be a major channel to heat the icm ( sternberg & soker 2008b , 2009a ; also brggen et al .
( 6 ) * the last two processes are effective not only in the directions of propagation of the jets and bubbles , but also perpendicular to this symmetry axis .
it seems that these processes provide the energy channel from the jets to the icm in all relevant regions .
again , self - consistent inflation of bubbles by jets is required to show that . * ( 7 ) * narrow very fast ( even relativistic ) jets can have negligible energy content and still fill the bubbles with radio radiation .
we elaborate on this preliminary result next .
in many clusters there is an imperfect morphological correlation between radio emission and deficient x - ray bubbles .
examples were given in the talks by fabio gastaldello and simona giacintucci .
more examples are : ( 1 ) in ngc 5044 the radio emission fills only one of three bubbles ( david et al . 2009 ) ; ( 2 ) hercules a does not show x - ray holes coincident with the radio lobes ( gizani & leahy 2004 ; nulsen et al .
2005 ) ; ( 3 ) in abell 2626 no x - ray deficient bubbles are observed at the locations of strong radio emission ( rizza et al .
2000 ; wong et al . 2008 ) .
we are also motivated by that the energy required to inflate bubbles is larger than the energy in radio emission ( e.g. , de young 2006 ) .
we note that weak radio emission at long wavelength can fill the bubbles .
this is expected as old radio jets might have filled the region . here
we are interested in the radio morphology of recently launched jets .
the simulations were performed using the _ virginia hydrodynamics - i _ code ( vh-1 ; blondin et al .
1990 ) , as described in our previous papers .
for the first 100 myr time period of our simulation we used the parameters of model i ( as described in sternberg & soker 2009b ) , at the end of which a fat bubble has been inflated .
all detail are in that paper .
we study a three - dimensional axisymmetric flow with a 2d grid ( referred to as 2.5d ) .
the fast ` radio ' jet was turned on at the same time the smw jet was turned off . while the smw jet was ejected along the symmetry axis ( horizontal axis in the figures ) ,
the fast jet was injected at an angle of @xmath15 to the symmetry axis . in our 2.5d numerical code
, this physically corresponds to a jet that is rapidly rotating around the symmetry axis at an angle of @xmath15 ( see sternberg & soker 2008a ) .
the fast jet has a half opening angle of @xmath16 , an injection velocity of @xmath17 , a mass loss rate per two jet of @xmath18 , and a total power of the two fast jets of @xmath19 , which is equal to the power of the smw jets that inflated the bubbles up to @xmath20myr . in reality
, we would like the fast ( radio ) jets power to decline from about the power of the smw jets to @xmath21 times the power of the smw jets , and be narrow jets with slow precession .
such a jet covers a solid angle of @xmath22 .
however , what matters for the penetration of the jets through the bubble is the momentum flux ( equals to the ram pressure ) . in our 2.5d code one jet
we inject covers a solid angle of @xmath23 . to maintain the desired momentum flux we had to use a power larger by a factor of @xmath24 .
the evolution of the fast ( radio ) jet , as shown in the figures , teaches us about the propagation of the fast jets and the morphology of their shocked gas , but the total volume filled by the shocked fast jet s material is larger than what a more realistic full 3d code will give .
our code does not explicitly separate the material originating in the smw and the fast jets .
we identify the material originating in the fast jet by its high postshock temperature , @xmath25k . in reality , this is where we expect to detect strong radio emission .
we note that the shocked fast jet material is spreading inside a large volume of the bubble .
however , it is not smooth , but has shapes of loops and arcs , as observed in some cases .
namely , there is no one - to - one ( perfect ) correlation between the shocked fast ( ` radio ' ) jet material and the bubble .
the main reason is that it is injected in one direction , and that the internal velocities of the gas inside the already inflated bubbles spreads the shocked fast wind gas around . in other words ,
the internal motion inside the already existing bubble spreads the shocked fast jet material to large distances and to different directions
. this might give the ( wrong ) impression that the bubble was inflated by the fast jet .
the justification for the transition from smw jets to much faster narrow jets can be explained in the following way in the frame of the cold feedback model .
after most of the cold gas was expelled from the vicinity of the smbh in one of the feedback cycles , two changes occur .
first , the accretion rate declines and so is the power of the jets .
second , the fast narrow jets launched by the smbh do not encounter large quantities of dense gas , that otherwise might prevent them from propagating to large distances ( soker 2008a ) .
therefore , instead of smw jets we observe the narrow radio jets . in reality
the transition will be gradual , and both types of jets might coexist .
we propose this as an explanation to many cases where no perfect correlation exist between the morphologies of strong radio emission and x - ray deficient bubbles .
if true , this implies that the observed radio jets are not the main energy source of the bubbles .
a. sternberg & n. soker _ mnras _ * 384 * , 1327 ( 2008a ) a. sternberg & n. soker _ mnras _ * 389 * , l13 ( 2008b ) a. sternberg & n. soker
_ mnras _ * 395 * , 228 ( 2009a ) a. sternberg & n. soker
_ mnras _ * 398 * , 422 - 428 ( 2009b ) r. j. wilman , a. c. edge , & a. m. swinbank _ mnras _ , * 395 * , 1355 - 1370 ( 2009 ) | for the recent four years we have been studying feedback heating in cooling flow ( cf ) clusters by agn activity that inflate bubbles by jets ; this short contribution to a meeting summarizes our main results . to achieve our results we had to self - consistently inflate the bubbles with jets , rather than inject them artificially .
our main results are as follows ( 1 ) feedback mechanisms that are based on bondi accretion fail . instead , the accretion to the central super - massive black hole ( smbh ) is in the form of cold dense blobs that fall - in from an extended region .
( 2 ) slow massive wide ( smw ) jets , or rapidly precessing jets , can inflate bubbles similar to those observed in cf clusters .
( 3 ) contrary to some claims in the literature , the inflated bubbles are stable for a relatively long time , becoming unstable only at later times .
( 4 ) a single bubble inflation episode excites multiple sound waves and shocks .
these can then heat the intracluster medium ( icm ) .
( 5 ) mixing of the bubble material to the icm is efficient , and can serve as a main heating channel .
( 6 ) the heating processes work in all directions , and can explain the heating of the icm in cf in clusters and in galaxies . address = dept . of physics , technion , haifa 32000 israel
; [email protected] address = dept . of physics , technion , haifa 32000
israel address = osservatorio astronomico di brera , via brera 28 , 20121 milano |
neutron stars have been serving as laboratories to probe the densest and most neutron - rich matter in the universe .
it is generally believed that the outer , low - density part of a neutron star ( crust ) consists of a body - center - cubic lattice of neutron - rich nuclei , embedded in a gas of electrons and , if any , dripped neutrons , and near normal nuclear density ( @xmath1 ) , the nuclei melt into uniform nucleonic matter , which mainly composes the star s core @xcite .
the equation of state ( eos ) of matter in the star , i.e. , neutron star matter , has one - to - one correspondence to the star s mass ( @xmath2 ) and radius ( @xmath3 ) relation via hydrostatic equilibrium .
observational data for @xmath2 have been accumulated @xcite , whereas those for @xmath3 have been recently estimated from observations of thermonuclear x - ray bursts with photospheric radius expansion and thermal spectra from quiescent low - mass x - ray binaries @xcite . in theoretically describing laboratory nuclei and neutron star matter
, it is useful to consider the energy of `` nuclear matter , '' i.e. , hypothetical infinite matter , composed of neutrons and of protons that have electric charge switched off . for simplicity , as neutron star matter
, we will consider zero - temperature , @xmath4 equilibrated , charge neutral matter made of real nucleons and electrons .
the eos of nuclear matter is still uncertain even near @xmath1 , while it can be constrained from terrestrial nuclear experiments @xcite and neutron star observations @xcite via theoretical calculations .
it is noteworthy that the candidates for low - mass neutron stars have been discovered in binary systems @xcite , which could give additional information on the eos once @xmath3 is measured .
the energy of uniform nuclear matter can be expanded around the saturation point of symmetric nuclear matter ( snm ) , i.e. , nuclear matter made of the same number of neutrons and protons , with respect to the nucleon number density , @xmath5 , and neutron excess , @xmath6 , defined as @xmath7 , where @xmath8 and @xmath9 denote the neutron and proton number densities . in practice , in the vicinity of the saturation point of snm at zero temperature , the energy per nucleon , @xmath10 , of uniform nuclear matter can be written as a function of @xmath5 and @xmath6 @xcite , i.e. , @xmath11\alpha^2 , \label{eq : w}\ ] ] where @xmath12 , @xmath13 , and @xmath14 are the saturation energy , the saturation density , and the incompressibility of snm , while @xmath15 and @xmath16 are associated with the symmetry energy coefficient @xmath17 .
that is , @xmath18 is the symmetry energy coefficient at @xmath19 , while @xmath16 characterizes the density dependence of the nuclear symmetry energy around @xmath19 , defined as @xmath20 . among these five parameters in eq .
( [ eq : w ] ) , @xmath12 , @xmath13 , and @xmath15 can be relatively easier to determine from empirical data for masses and radii of stable nuclei , while the remaining two parameters , @xmath14 and @xmath16 , are more difficult to fix @xcite .
this is why we focus on the various sets of @xmath14 and @xmath16 ( table [ tab : eos ] ) in analyzing neutron star matter .
.nuclear matter eos parameters [ cols="^,^,^,^,^",options="header " , ] these two parameters , @xmath14 and @xmath16 , mainly determine the stiffness of neutron - rich nuclear matter , but have yet to be fixed . it is also suggested that @xmath14 is related to the giant resonances of stable nuclei @xcite , while @xmath16 is associated with the structure and reactions of neutron - rich nuclei @xcite and the pressure of pure neutron matter at the saturation density of snm . additionally , one could constrain @xmath16 via quasi - periodic oscillations in giant flares observed from soft - gamma repeaters @xcite .
in contrast to the well - known empirical nuclear mass and radius formulas @xcite , the neutron star counterparts have to be theoretically given as function of not only the central density ( @xmath21 ) , but such eos parameters as @xmath14 and @xmath16 .
so far , however , the dependence of low - mass neutron star models on @xmath14 and @xmath16 remains to be examined systematically .
we thus start with construction of the neutron star models from various eoss of neutron star matter that meet the following conditions : 1 . unified description of matter in the crust and core based on the same eos of nuclear matter with specific values of @xmath14 and @xmath16 .
consistency of the masses and radii of stable nuclei calculated within the same theoretical framework with the empirical values .
mass and radius formulas for low - mass neutron stars are finally obtained in such a way as to approximately reproduce the neutron star models thus constructed .
among many available eoss of neutron star matter , we adopt the eoss that meet the above conditions , i.e. , unified eoss , which are categorized into three groups as in table [ tab : eos ] .
the first is based on the phenomenological eos of uniform nuclear matter that was constructed by two of us @xcite , using a simplified version of the extended thomas - fermi theory @xcite , in such a way as to reproduce empirical masses and radii of stable nuclei .
they adopted the pad - type potential energies with respect to the nucleon density @xmath5 for snm and for neutron matter , respectively , and connected them in a quadratic approximation with respect to neutron excess @xmath6 .
this form of the potential energy can well reproduce the variational calculations of @xcite , to which , in fact , the high - density behavior of neutron matter was adjusted .
the @xmath6 dependence of the potential energy is partially justified by the variational calculations of @xcite , and the expression for the total energy reproduces eq .
( [ eq : w ] ) in the limit of @xmath22 and @xmath23 . with such eoss of uniform nuclear matter obtained for various sets of ( @xmath14 , @xmath16 )
, they constructed the eoss of neutron star matter @xcite by generalizing the above thomas - fermi theory as done by @xcite .
hereafter , such eoss of neutron star matter are referred to as the oi - eoss .
we remark that generally accepted values of @xmath14 lie in the range of @xmath24 mev @xcite or so , while the oi - eoss include rather extreme cases of @xmath25 and 360 mev , as shown in table [ tab : eos ] , to cover the large parameter space .
the final mass and radius formulas would remain almost unchanged even if the oi - eoss with @xmath25 and 360 mev are not included in the fitting . in the second group ,
there are two eoss of neutron star matter calculated within the relativistic framework .
one is the shen eos based on the relativistic mean field theory with the tm1 nuclear interaction @xcite , and the other is the miyatsu eos based on the relativistic hartree - fock theory with the chiral quark - meson coupling model @xcite . in both eoss
, the same type of the thomas - fermi model as used for the oi - eoss is used in describing neutron star matter in such a way as to reproduce empirical masses and radii of stable nuclei . the third group is composed of the five eoss of neutron star matter based on the skyrme - type effective interactions : fps @xcite , sly4 @xcite , bsk19 , bsk20 , and bsk21 @xcite . the fps interaction , which was constructed by fitting the properties of uniform nucleon matter calculated by @xcite , well reproduces the empirical ground - state properties of doubly magic stable nuclei via the hartree - fock calculations .
the sly4 interaction was constructed by @xcite in such a way as to reproduce the microscopic eos of neutron matter calculated with the uv14+uvii nuclear force by @xcite as well as the empirical ground - state properties of doubly magic stable nuclei within the hartree - fock approximation .
the bsk19 , bsk20 , and bsk21 interactions are written in the form of the nuclear energy - density functionals , which are derived from generalized skyrme interactions in such a way as to fit all the available nuclear mass data @xcite . as a result , empirical charge radii were also well reproduced .
these interactions are different in the sense that bsk19 , bsk20 , and bsk21 are fitted to the eoss of neutron matter derived by @xcite , @xcite , and @xcite , respectively .
this difference is expected to play a role in estimating the effect of uncertainties in three - neutron interactions on the stellar properties , as we shall see . in describing neutron star matter ,
a compressible liquid - drop approach was used for fps and sly4 , while an extended thomas - fermi model was used for bsk19 , bsk20 and bsk21 . to calculate the neutron star models in the present study
, we adopt the analytical expressions for fps and sly4 given by @xcite and for bsk19 , bsk20 , and bsk21 given by @xcite . .
we plot the relations between the mass and radius ( a ) and between the gravitational redshift and radiation radius ( b ) .
the mark and end on each line denote the stellar models with @xmath26 and @xmath27 , respectively . in ( a )
, the labels on the lines denote the values of the nuclear matter parameter @xmath28 .
to distinguish between the oi - eoss , we add the values of @xmath14 to the oi - eos labels ; for example , we use oi 180 " for the two oi - eoss with @xmath29 mev ( left , smaller @xmath16 ; right , larger @xmath16 )
. the shaded region corresponds to the allowed region from the observed radiation radius of the neutron star in @xmath30 cen ( see text for details ) .
, width=377 ]
now , we construct nonrotating neutron stars by integrating the tolman - oppenheimer - volkoff equations from the stellar center of density @xmath21 outward up to the position where the pressure vanishes .
it is not clear up to what density the adopted unified eoss are applicable .
nonetheless , one can expect that non - nucleonic components such as hyperons and quarks do not occur below @xmath31 @xcite and that the uncertainty from three - neutron interactions in the eos of pure neutron matter becomes relevant above @xmath31 , as suggested by quantum monte carlo ( qmc ) calculations @xcite .
we thus examine the stellar models for @xmath32 , where @xmath1 is set to @xmath33 g @xmath34 , and the resultant @xmath2-@xmath3 relations are plotted in fig .
[ fig : mr](a ) . to systematically describe various stellar models ,
we introduce a new auxiliary parameter @xmath28 defined as @xmath35 .
the values of @xmath28 are shown in table [ tab : eos ] .
remarkably , the @xmath2-@xmath3 relation changes almost smoothly with @xmath28 .
note that the oi - eoss @xcite with @xmath36 mev are too soft to keep the pressure positive and thus not used here .
this implies the lower limit of @xmath28 of order 30 mev .
meanwhile , the eos models used here cover the values of @xmath28 up to @xmath37 mev , which is significantly larger than expected from existing nuclear experiments .
we remark that the powers of @xmath16 and @xmath14 in @xmath28 are chosen to be simple rational numbers in such a way that @xmath28 has the same unit as @xmath16 and @xmath14 , i.e. , mev . if one considers arbitrary real numbers as the exponents , therefore , one could choose different kinds of @xmath28 with which the @xmath2-@xmath3 relation changes as smoothly as the present choice . from the observational viewpoint , the radiation radius @xmath38 and the gravitational redshift @xmath39 with the gravitational constant @xmath40 and the speed of light @xmath41 could be more relevant in describing the stellar properties than @xmath2 and @xmath3 .
the calculated @xmath42-@xmath43 relation again shows a smooth change with @xmath28 ( fig .
[ fig : mr](b ) ) . the photon flux ,
if detected , would be proportional to @xmath44 , where @xmath45 is the distance from the earth , while the gravitational redshift could be determined from the possible shift of atomic absorption lines in spectra of the stars .
the smooth change of the stellar properties with @xmath28 suggests that not only future nuclear experiments but also simultaneous measurements of @xmath2 and @xmath3 or , equivalently , @xmath42 and @xmath43 could constrain @xmath28 , which could in turn lead to restriction of the stellar models .
in particular , observations of low - mass neutron stars would be essential .
for example , the radiation radius of the x - ray source , cxou 132619.7472910.8 , in the globular cluster ngc 5139 ( @xmath30 cen ) has been determined as @xmath46 km from the @xmath47 data @xcite .
the allowed region from this @xmath43 is shown in fig .
[ fig : mr](a ) and ( b ) with the shaded region .
this is still consistent with various values of @xmath28 , but future precise determination of @xmath43 could constrain @xmath28 , if @xmath2 is low enough .
additionally , thermal spectra detected from quiescent low - mass x - ray binaries are expected to give @xmath2 and @xmath3 simultaneously @xcite . .
the stellar models constructed from various unified eoss are given for @xmath48 , @xmath49 , and @xmath50 .
the solid , broken , and dotted lines are the linear fitting to the cases of @xmath48 , @xmath49 , and @xmath50 , respectively ( see text for details).,width=377 ]
to examine the dependence of the stellar properties on @xmath28 more clearly , we plot the stellar masses calculated for @xmath48 , @xmath49 , and @xmath50 ( fig .
[ fig : etam](a ) ) . from this figure
, we find that the stellar masses for fixed @xmath21 can be approximately expressed as a linear function of @xmath28 , @xmath51 , where @xmath52 and @xmath53 are adjustable parameters that depend on @xmath21 .
the validity of @xmath28 is now evident .
the deviation of the calculations from the linear fit at @xmath48 is larger than that at @xmath54 , particularly for bsk20 and bsk21 .
such deviation is of the order of uncertainties in @xmath2 due to three - neutron interactions obtained from the qmc evaluations @xcite .
the parameters @xmath52 and @xmath53 can then be expressed as a quadratic function of @xmath55 within the accuracy of errors less than a few percent ( fig .
[ fig : c0c1 ] ) .
finally , we obtain the mass formula : @xmath56 where we confine ourselves to @xmath57 ; otherwise , the stellar models can become unstable with respect to decompression , depending on the eos of neutron star matter . and @xmath53 in the mass formula .
the corresponding quadratic fitting curves ( solid and broken lines ) are also shown as a function of @xmath58 . here
we consider the stellar models only for @xmath59 to avoid unstable neutron star models.,width=377 ] we also find that the gravitational redshift calculated for fixed @xmath21 can be approximately expressed as a linear function of @xmath28 ( fig . [
fig : etam](b ) ) .
then , just like the mass formula ( [ eq : mass ] ) , we can obtain the theoretical formula for @xmath42 as @xmath60 using eqs .
( [ eq : mass ] ) and ( [ eq : zz ] ) , one could estimate the values of @xmath28 and @xmath61 from possible simultaneous measurements of @xmath2 and @xmath42 . in general ,
( [ eq : mass ] ) and ( [ eq : zz ] ) can have as many as four sets of solutions @xmath62 for given observational values of @xmath63 and @xmath42 .
as mentioned above , however , eqs .
( [ eq : mass ] ) and ( [ eq : zz ] ) are valid in the range of @xmath64 . in this range , as shown in fig .
[ fig : etam ] , the solution @xmath62 has to be unique . .
the stellar models constructed from various unified eoss are given for @xmath54 ( black ) , @xmath49 ( red ) , and @xmath27 ( blue ) .
the solid , broken , and dotted lines are the formula values for the cases of @xmath65 , 1.5@xmath1 , and 1.0@xmath1 , respectively , obtained from eqs .
( [ eq : mass ] ) and ( [ eq : zz ] ) . the thick straight line denotes the converging behavior expressed by eq .
( [ eq : fit - r]).,width=377 ] it is straightforward to obtain the formula for @xmath3 from eqs .
( [ eq : mass ] ) and ( [ eq : zz ] ) .
the obtained formula can be compared with the calculations of @xmath3 for @xmath54 , @xmath49 , and @xmath27 ( fig .
[ fig : etar ] ) .
we confirm a good agreement between those two except for @xmath66 mev .
the mass and radius formulas could help to constrain not only the nuclear matter parameter @xmath28 but also a star s @xmath21 via possible simultaneous measurements of the star s @xmath2 and @xmath3 .
if such measurements are precise , @xmath28 could be deduced to within the accuracy of @xmath67 mev , which would provide a basis for analyzing more massive neutron stars . from fig .
[ fig : etar ] , one can also observe that the calculated @xmath3 depends nonlinearly on @xmath21 at small values of @xmath28 , while converging on an approximately linear function of @xmath28 at sufficiently large values of @xmath28 : @xmath68 note that such nonlinear dependence at small values of @xmath28 arises from the flattened behavior of the corresponding @xmath2-@xmath3 relations that can be seen from fig .
[ fig : mr](a ) , while such convergence at large values of @xmath28 is related to the vertically straightened behavior of the corresponding @xmath2-@xmath3 relations .
in this paper , we have succeeded in constructing the theoretical formulas for the masses , gravitational redshifts , and radii of low - mass neutron stars as functions of the star s central density and the new eos parameter @xmath28 in a manner that is consistent with empirical masses and radii of stable nuclei .
the value of @xmath28 , which characterizes the stiffness of neutron star matter , remains unknown , but could be deduced from possible simultaneous @xmath2 and @xmath3 measurements via comparison with our formulas if the star observed is light enough .
thus , a firm evidence for the presence of low - mass neutron stars is first of all desired .
one promising candidate is the neutron star in the high - mass x - ray binary 4u 1538 - 52 , of which the mass could be significantly low or even the lowest among stars with known mass if the binary orbit is eccentric @xcite .
the x - ray burster 4u 1724 - 307 in the globular cluster terzan 2 is even more interesting because the x - ray data from the cooling phase of photospheric radius expansion bursts apparently allow the object to have a relatively low mass and still a significantly large radius @xcite .
such conclusions are tentative partly because of the dependence on the atmosphere models adopted and partly because of uncertainties in the distance to the object , but , if valid , might eventually suggest the @xmath28 value of order or even larger than 130 mev .
h.s . is grateful to c. ishizuka , k. sumiyoshi , t. tatsumi , and n. yasutake for comments . h.s .
, k.i . , and a.o .
acknowledge the hospitality of the facility for rare isotope beams , where this work was initiated , and thank t. enoto for helpful discussions .
this work was supported in part by grants - in - aid for scientific research on innovative areas through no .
24105001 and no .
24105008 provided by mext , in part by grant - in - aid for young scientists ( b ) through no .
24740177 provided by jsps , in part by the yukawa international program for quark - hadron sciences , and in part by grant - in - aid for the global coe program the next generation of physics , spun from universality and emergence " from mext . | neutron stars , produced at the death of massive stars , are often regarded as giant neutron - rich nuclei .
this picture is especially relevant for low - mass ( below about solar mass , @xmath0 ) neutron stars , where non - nucleonic components are not expected to occur . due to the saturation property of nucleonic matter ,
leading to the celebrated liquid - drop picture of atomic nuclei , empirical nuclear masses and radii can be approximately expressed as function of atomic mass number .
it is , however , not straightforward to express masses and radii of neutron stars even in the low - mass range where the structure is determined by a balance between the pressure of neutron - rich nucleonic matter and the gravity .
such expressions would be of great use given possible simultaneous mass and radius measurements .
here we successfully construct theoretical formulas for the masses and radii of low - mass neutron stars from various models that are consistent with empirical masses and radii of stable nuclei . in this process
, we discover a new equation - of - state parameter that characterizes the structure of low - mass neutron stars .
this parameter , which plays a key role in connecting the mass - radius relation of the laboratory nuclei to that of the celestial objects , could be constrained from future observations of low - mass neutron stars . |
the motion of a massive spinning test particle in general relativity is described by the mathisson - papapetrou model @xcite consisting of 10 equations @xmath0^{\alpha\beta},\end{aligned}\ ] ] for 13 unknowns : the generalized momentum of the particle @xmath1 ( 4 ) , the antisymmetric spin tensor @xmath2 ( 6 ) and the unit tangent vector @xmath3 to the world line used for the multipole reduction ( 3 ) . for the model to be complete ,
it must supplemented by three additional equations which in the literature are usually called supplementary conditions(scs ) .
there are three standard choices for the supplementary conditions : 1 .
papapetrou - corinaldesi @xcite : @xmath4 , where @xmath5 is some arbitrary timelike vector chosen for convenience of calculation ; 2 .
pirani @xcite : @xmath6 ; 3 .
tulczyjew @xcite : @xmath7 .
each of these is an algebraic condition restricting the spin tensor to have 3 fewer independent components .
if the vector in terms of which the supplementary condition is expressed is timelike , this just means that the spin tensor is equivalent to a spin vector lying in the local rest space associated with this timelike direction . in a sense ,
only the pirani and tulczyjew scs are intrinsic `` relations between the various unknowns of the model and have more credibility as physical conditions rather than just a mathematical convenience which is coordinate dependent '' , as is the papapetrou - corinaldes sc . the extension of the mathisson - papapetrou model to the case of a null multipole reduction world line with tangent
@xmath8 has been considered by mashhoon @xcite .
the equations of the extended model have exactly the same form as ( [ papcoreqs1 ] ) and ( [ papcoreqs2 ] ) but with @xmath9 replaced by @xmath8 and @xmath10 ( the proper time parameter along the timelike world line ) replaced by @xmath11 ( a parameter along the null world line such that @xmath12 ) @xmath13^{\alpha\beta}\ .\end{aligned}\ ] ] eqs .
( [ papcoreqs1n ] ) and ( [ papcoreqs2n ] ) must then be solved assuming some sc .
consider only the more physically motivated scs of pirani and tulczyjew , with pirani s conditions now naturally generalized to @xmath14 .
furthermore , since we are interested in the massless limit of the mathisson - papapetrou equations , and since the mass of the particle is defined by @xmath15 , the massless limit would imply @xmath16 , so we assume this to be true here .
mashhoon @xcite found tulczyjew s condition @xmath7 to be inconsistent if @xmath17 is also null : @xmath18 . while we could examine this condition for the nonnull case
, it seems most reasonable to consider pirani s sc as the only physically meaningful supplementary condition . under these conditions
mashhoon @xcite has shown that the particle follows a null geodesic ( so that one can assume @xmath11 to be an affine parameter ) with its spin vector ( denoted by @xmath19 ) [ spinvector ] s^=12 ^l_s _ parallelly transported along the null world line with tangent @xmath8 and either parallel or antiparallel to @xmath8 . moreover , again adopting mashhoon s point of view , the generalized momentum of the particle should be spacelike or even null and is determined by solving equations ( [ papcoreqs1n ] ) and ( [ papcoreqs2n ] ) supplemented by @xmath14 .
the other components of the spin tensor not equivalent to the spin vector that one can introduce must also be determined . and
@xmath17 and there is no agreement in the literature on this freedom .
for example , tod , de felice and calvani @xcite consider @xmath17 to be timelike assuming that it represents the total energy - momentum of the particle while they do not impose any causality condition on the world line @xmath9 , which plays the role of a mere mathematical tool " to perform the multipole reduction ; on the other hand according to mashhoon @xcite , @xmath17 can be considered to be analogous to a canonical momentum for the particle , in which case there is no reason to prefer a particular causality character for it , while the world line @xmath9 must be timelike in the massive case or null in the massless case because it is a world line intimately associated with the test particle location .
this uncertainty in the model itself , i.e. , the uncertainty in the sc , has to be considered as a limitation of our description of spinning test particles in general relativity .
we are aware of this problem and only consider the pirani sc here . ]
note that the spin vector depends on the scale of @xmath8 and hence on the parametrization chosen for the null world lines to which it is tangent . in the present paper , using the newman - penrose ( np ) formalism , we derive the equations of motion for a massless spinning test particle in any algebraically special vacuum spacetime for the special case in which the multipole reduction world line is aligned with a principal null direction of the spacetime .
this leads to very simple equations which are straightforward to solve .
explicit solutions corresponding to some familiar petrov type d and type n spacetimes ( including the schwarzschild , kerr , c metric , kasner , and single exact gravitational wave spacetimes ) are derived and discussed . finally , motion along null circular orbits is also studied and explicit solutions in black hole spacetimes are presented .
we adopt the standard np notation of chandrasekhar @xcite which includes the metric signature @xmath20 @xmath21 @xmath21 @xmath21 .
in this section we express the above equations for massless spinning test particles in the np formalism .
let @xmath22 , @xmath23 , @xmath24 , @xmath25 be an np frame consisting of four null vector fields satisfying the relations @xmath26 and @xmath27 , so that the frame components of the metric tensor are ( _ ab ) = ( ^ab ) = . exactly like the np representation of the electromagnetic faraday 2-form , the 6 real components of the spin tensor @xmath28 are represented in terms of 3 complex scalar quantities @xmath29 ( s_ab ) = , which have the well - defined spin - weights @xcite @xmath30 respectively .
is only fixed up to boosts and rotations which leave the directions of @xmath8 and @xmath31 unchanged , i.e. , up to class iii tetrad lorentz transformations @xmath32 a quantity @xmath33 which tranforms under this transformation as @xmath34 is said to be a quantity of type @xmath35 , with spin weight @xmath36 and boost weight @xmath37 .
most of the np quantities have a well - defined spin and boost weight .
the exceptions are the spin coefficients @xmath38 and the directional derivative operators @xmath39 along the frame vectors . ]
we first adapt the np frame to the null world line of the multipole reduction by choosing its tangent as the first frame vector : @xmath40 .
the zero mass condition @xmath41 then forces the second component @xmath42 of @xmath17 to be zero ( and hence @xmath43 , so @xmath17 is spacelike ) .
( [ papcoreqs1n ] ) and ( [ papcoreqs2n ] ) then take the following form in a vacuum spacetime where the riemann curvature tensor reduces to the weyl curvature tensor ( np scalars @xmath44 ) @xmath45 where @xmath46 is the directional derivative along @xmath8 . as usual
the equation for @xmath47 is just the complex conjugate of the one for @xmath48 .
pirani s supplementary conditions are simply : and @xmath49 and his choice of frame transport law along @xmath8 is equivalent to the conditions @xmath50 and @xmath51 on the np scalars which guarantee that @xmath8 is the tangent to an affinely parametrized geodesic and that @xmath31 is parallel transported along @xmath8 .
this latter assumption is unnecessary here . ]
@xmath52 i.e. , @xmath53 is purely imaginary with @xmath54 real and whose meaning is discussed below ( do nt confuse @xmath54 with the symbol for the spin - weight of tensorial quantities ) .
the only nonvanishing frame components of the spin tensor are then @xmath55 and @xmath56 . with these conditions
the equations of motion then become @xmath57 the solution @xmath58 of the first equation would lead to @xmath59 and hence vanishing spin tensor , a trivial case that will not be considered here .
one must therefore have instead @xmath60 , i.e. , @xmath8 is a null geodesic @xcite . hence , these equations further reduce to @xmath61 a further simplification can be made by using a type iii rotation of the np frame to set the spin coefficient @xmath62 .
this corresponds to require that the congruence of @xmath8 is affinely parametrized .
the final set of equation is the following : @xmath63 these equations are still general for any vacuum spacetime . in the special case that the 4-momentum is a null vector , equivalent to @xmath64 as seen above , then @xmath65 and the previous set of equations specialize to @xmath66
the spin vector defined in eq .
( [ spinvector ] ) as the contracted dual of the spin tensor ^= ^ * s^ l _ = i(|s_1 - s_1 )
l + i s_0 m - i |s_0 |m , when the pirani supplementary conditions @xmath53 and @xmath67 are imposed , reduces to the simple form [ spinvector_new ] s^= s l^ , so that @xmath54 represents the helicity of the particle ( see p. 80 of reference for the relation between the helicity and spin - weight of a massless field ) . since @xmath54 is constant along the world line , and @xmath8
is parallelly transported along itself , the spin vector is parallel transported along @xmath8 , recovering mashhoon s result .
however , the spin tensor is not completely determined by the spin vector alone as in the timelike case but also requires the independent component @xmath68 [ snp ] s = l ( s_2 m + |s_2 |m ) + is m |m .
the real vector @xmath69 can be interpreted as a sort of orbital angular momentum " contribution to the total spin ( see reference , p. 68 ) , as measured with respect to the null world line of the multipole reduction associated with @xmath8 . in a sense , since @xmath8 is a null vector , the multipole moments taken with respect to any spacetime point on the null world line are always those of a particle in motion with respect to any physical observer .
alternatively , there are no observers such that the associated electric part of @xmath28 vanishes identically as it does in the timelike case . to see this
let us consider the lorentz frame naturally associated with the np frame , defined by @xmath70 using the notation @xmath71 .
( [ snp ] ) then becomes s = ul(u)+^*(u)s(u ) , where @xmath72 e_2 -{\rm im}\ , [ s_2 ] e_3\ , \nonumber \\ s(u)&= & { \rm re}\ , [ s_2 ] e_3 + { \rm im}\ , [ s_2 ] e_2 + s\ , e_1\ , \end{aligned}\ ] ] which are respectively the electric and magnetic parts of @xmath28 with respect to @xmath9 .
the notation for the spatial dual of a spatial vector @xmath73 ( spatial with respect to @xmath9 , i.e. , @xmath74 ) , ^=12 ^u_x _ , as well as for the exterior product of two spatial vectors @xmath73 and @xmath75 ( spatial with respect to @xmath9 ) ^= ^u_x_y _ , where @xmath76 is the unit volume four form , will be used here and below @xcite . note that l(u)+is(u)=s_2 m + is e_1 .
the two algebraical invariants of @xmath28 can then be written as i_1 = 12 s^s _
= -l(u)^2+s(u)^2s^2 , i_2 = 12 s^^*s _ = l ( u ) s(u)0 .
the spin tensor can be classified in the same way as the electromagnetic 2-form .
if the ( complex ) invariant i = 14 ( s^+i^*s^)(s_+i^*s _ ) = i_1+ii_2 = 4(s_0 s_2 - s_1 ^ 2 ) is nonzero , the spin tensor will be said to be non - null or nonsingular , while if it is zero the spin tensor will be said to be null or singular . in the present case since @xmath77
, it reduces to @xmath78 and the spin tensor is singular only when @xmath58 .
note that the spinning test particle model holds when a dimensionless quantity formed from @xmath54 is very small : otherwise modification of the background due to backreaction should be considered and the particle can not be considered to be a test particle . due to the choice of p supplementary conditions ,
the quantity @xmath68 does not contribute to the spin invariant ; therefore , requiring the magnitude of @xmath54 to be small does not give any restriction on it .
the 4-velocity @xmath5 of any physical observer can be expressed in terms of its relative velocity with respect to @xmath9 and vice versa @xcite u=(u , u)[u+(u , u ) ] , u=(u , u)[u+(u , u ) ] , where @xmath79 with @xmath80 and @xmath81 related to @xmath82 and @xmath83 by a boost transformation exactly like the electric and magnetic fields @xcite @xmath84\ , \nonumber \\
p(u)s(u)&= \gamma ( u , u ) [ s(u)+\nu(u , u)\times_u l(u)]\ , \end{aligned}\ ] ] where @xmath85 projects orthogonally to @xmath9 . since @xmath86 , it follows that @xmath80 and @xmath81 are orthogonal , i.e. , any observer will see the electric and magnetic parts of the spin tensor to be orthogonal .
moreover , from ( [ elemag ] ) , it easy to see that l(u)+e_1 s(u)=0 , which compared with ( [ em_transf ] ) , gives @xmath87 only for an observer moving with respect to @xmath9 with velocity @xmath88 , i.e. , with unit speed , so that it does not correspond to any physical observer .
the goldberg - sachs theorem @xcite states that a vacuum metric is algebraically special if and only if it contains a shearfree null geodesic congruence with tangent @xmath8 ( taken as the first vector of an np frame ) , i.e. , = 0= , _ 0=0=_1 . hence , assuming that the ( single ) line of multipole reduction associated with @xmath8 is a world line in such a congruence , i.e. , @xmath8 is a principal null direction of the spacetime , then the set of equations ( [ eqsspinp_fin2 ] ) simplifies to @xmath89 the formal integration of these final equations is straightforward , leading to @xmath90 and @xmath91 being constant along @xmath8 and @xmath92 \rmd \lambda ' + p^1(\lambda_0)\ .\end{aligned}\ ] ] in the special case @xmath93 ( i.e. , @xmath17 null and aligned with @xmath8 ) , the solution ( [ sold ] ) reduces to @xmath94 \rmd \lambda ' + p^1(\lambda_0)\ .\end{aligned}\ ] ] eqs .
( [ sold ] ) and ( [ sold0 ] ) suggest the presence of a helicity - connection coupling `` term in @xmath68 , as well as a helicity - curvature coupling '' term in @xmath95 .
we will come back to these couplings when discussing explicit examples .
in each case we adopt a well - known np frame with the above properties , assuming that the world line of the multipole reduction is an affinely parametrized integral curve of the principal null direction @xmath8 .
we investigate the principal null direction corresponding to outgoing radial null geodesics in the schwarzschild case and those which are asymptotically radial outgoing null geodesics in the kerr case , and those in related spacetimes which are analogous to these .
the ingoing null geodesics of this type can be treated similarly .
consider the schwarzschild solution @xcite written in standard spherical coordinates s^2= ( 1-)t^2 - ( 1-)^-1r^2 - r^2 ( ^2 + ^2 ^ 2 ) .
introduce the np frame adapted to the outgoing radial null geodesics @xmath96\ , \nonumber \\
m&= & \frac{1}{\sqrt{2}r}\left(\partial_\theta + \frac{i}{\sin \theta}\partial_\phi \right)\ , \end{aligned}\ ] ] which has the following nonvanishing spin coefficients @xmath97 and the single nonvanishing weyl scalar @xmath98 .
the integral curves of @xmath8 can be expressed in the form t = r - r_0 + 2m+t_0 , r=-_0+r_0 , = _ 0 , = _ 0 , where @xmath99 are arbitrary integration constants taken at @xmath100 . then eqs . ( [ eqsspinp ] ) reduce to @xmath101 with solution @xmath90 , @xmath102 , @xmath103 and @xmath91 .
the case @xmath104 is clearly trivial . the taub - nut solution @xcite is often written in the form @xmath105\ , \end{aligned}\ ] ] where @xmath106 , and it represents the spacetime of a black hole with an additional gravitomagnetic monopole charge associated with the parameter @xmath107 .
introduce the np frame generalizing the previous schwarzschild situation to which it reduces when @xmath108 @xmath109\ , \nonumber \\ m&= & \frac{1}{\sqrt{2}(r+i\ell ) } \left[-2i\ell\cot \theta \partial_t + \partial_\theta + \frac{i}{\sin \theta}\partial_\phi \right]\ , \end{aligned}\ ] ] with nonvanishing spin coefficients @xmath110 and the single nonvanishing weyl scalar @xmath111 .
the integral curves of @xmath8 have the form @xmath112 + t_0\ , \nonumber \\ \fl\quad r&=&\lambda-\lambda_0+r_0\ , \quad \fl\quad \theta = \theta_0\ ,
\quad \fl\quad \phi = \phi_0\ , \end{aligned}\ ] ] where @xmath113 , @xmath114 , @xmath115 and @xmath116 are arbitrary integration constants taken at @xmath100 .
the formal integration of ( [ sold ] ) is then straightforward , leading to @xmath90 and @xmath91 being constant along @xmath8 and @xmath117 ^ 2 } -\frac{(r_0-m)^2-(m^2+\ell^2)}{(r_0 ^ 2+\ell^2)^2 } \right\ } \nonumber\\ \fl\quad & & + p^1(\lambda_0)\ .\end{aligned}\ ] ] when @xmath118 and @xmath119 , i.e. , the particle starts moving from the horizon @xmath120 , we have @xmath121 and the solution is given by @xmath90 and @xmath91 constant along @xmath8 and @xmath122 if in addition @xmath123 , then @xmath124 and @xmath17 is null and aligned with @xmath8 p= s(r_+ ) l . the behavior of the varying component @xmath95 of the total 4-momentum as a function of the radial parameter is shown in fig . [ fig:1 ] . moving from @xmath120 out to infinity the proportionality factor with @xmath8 passes from zero to a maximum positive value at the real solution of @xmath125 and then returns asymptotically to zero . of the total 4-momentum
is plotted versus the radial parameter @xmath126 , where @xmath120 is the ( outer ) horizon in each of these spacetimes .
the initial conditions are chosen so that @xmath127 with @xmath128 for all cases , @xmath129 and @xmath130 for taub - nut , and @xmath131 , im@xmath132=1/(10\sqrt{2}a)$ ] and @xmath133 for kerr .
@xmath95 represents the component of the generalized momentum of the particle along the world line of the multipole reduction which is assumed to coincide with a principal null direction of the spacetime . in the schwarzschild case
this quantity is constant all the world line and identically zero for our choice of initial conditions . in the taub - nut case
it varies but asymptotically returns to its initial value . in the kerr case ,
it varies and does not return to its initial value , although it does approach the asymptotic value @xmath134 , with our choice of initial conditions . ]
the c metric solution @xcite , which represents the spacetime associated with a uniformly accelerating black hole , is often written in the form s^2=h u^2 + 2u r-2ar^2 ^ 2r -^2-r^2 g ^2 , where @xmath135\ , \nonumber \\ g&=&\sin^2\theta-2am\cos^3\theta\ .\end{aligned}\ ] ] we limit our considerations to the range of parameters @xmath136 where the c metric is static @xcite . introduce the np frame @xmath137 with nonvanishing spin coefficients @xmath138\ , \nonumber\\ \fl\quad & & \gamma=\frac{a^2r^3g+ma^2r^2\cos^2\theta(a\cos \theta r-3 ) + m - a\cos \theta r^2}{2r^2}\end{aligned}\ ] ] and weyl scalar @xmath98 . in order to write down the equations for the integral curves of @xmath8 , it is convenient to introduce the new variable w= , in place of @xmath139 ( @xmath140 is constant along @xmath8 ) , which has proven to be helpful in studying the horizons of the c metric , i.e. , the solutions of the equation @xmath141 , equivalent to [ eqw ] w^3-w+2ma=0 . in the case @xmath142 its solutions are w_1=2u , w_2=-u+ , w_3=-u- , where + iv= ( - + i ) ^1/3 , i.e. , = -(13arccos ) , = ( 13arccos ) . in terms of these quantities
one finds @xmath143 where w_0= . finally , the formal integration of the equations of motion is straightforward , giving @xmath90 and @xmath91 constant along @xmath8 and @xmath144+p^1(\lambda_0)\ , \end{aligned}\ ] ] where in this case @xmath145 , depending only on @xmath140 , is constant along @xmath8 . in the equatorial plane ( @xmath146 )
we have @xmath147 .
thus @xmath148+p^1(\lambda_0)\ .\end{aligned}\ ] ] consider the kerr metric @xcite expressed in standard boyer - lindquist coordinates @xmath149
^ 2-\frac{\sigma}{\delta}\rmd r^2-\sigma \rmd \phi^2\ , \end{aligned}\ ] ] where the standard notation @xmath150 , @xmath151 has been adopted .
introduce the kinnersley frame @xmath152\ , \quad n = \frac{1}{\delta}\left [ ( r^2+a^2)\partial_t -\delta \partial_r + a \partial_\phi\right]\ , \nonumber \\
m&= & \frac{1}{\sqrt{2}\sin \theta ( r+ia\cos \theta ) } \left(ia\sin^2\theta \partial_t + \sin \theta \partial_\theta+i\partial_\phi \right)\ .\end{aligned}\ ] ] the nonvanishing spin coefficients are @xmath153}{2\sigma ( r - ia\cos \theta)}\ , \nonumber\\ \fl \quad & & \alpha = -\frac{1}{2\sqrt{2}}\frac{\cos \theta(r+ia\cos \theta)-2ia}{\sin \theta(r - ia\cos \theta)^2}\ , \nonumber \\
\quad & & \pi=\frac{ia\sin \theta}{\sqrt{2}(r - ia\cos \theta)^2}\ , \quad \beta=\frac{\cos \theta}{2\sqrt{2}\sin \theta(r+ia\cos \theta)}\ , \end{aligned}\ ] ] and the weyl scalar is @xmath154 .
the integral curves of @xmath8 are @xmath155\ , \nonumber \\ \fl & & r = r_0+\lambda-\lambda_0 , \quad \theta=\theta_0\ , \nonumber \\ \fl & & \phi=\phi_0-\frac{a}{\sqrt{m^2-a^2}}\left[{\rm arctanh } \left ( \frac{r - m}{\sqrt{m^2-a^2}}\right ) -{\rm arctanh } \left ( \frac{r_0-m}{\sqrt{m^2-a^2}}\right)\right]\end{aligned}\ ] ] and represent outgoing null geodesics which are asymptotically radial far from the horizon .
( [ eqsspinp ] ) then have the following solution @xmath156\ , \nonumber \\ \fl\quad p^1(\lambda)&=&p^1(\lambda_0)+\frac{1}{\sqrt{2}}ia\sin \theta\left\ { \bar p^3(\lambda_0 ) \left [ \frac{1}{\lambda-\lambda_0+r_0-ia\cos \theta}-\frac{1}{r_0-ia\cos \theta}\right ] \right.\nonumber \\ \fl\quad & & \left . - p^3(\lambda_0 ) \left [ \frac{1}{\lambda-\lambda_0+r_0+ia\cos \theta}-\frac{1}{r_0+ia\cos \theta}\right ] \right\ } \nonumber \\ \fl\quad & & -2am\cos \theta s(\lambda_0)\left [ \frac{\lambda-\lambda_0+r_0}{(\lambda-\lambda_0+r_0)^2+a^2\cos^2\theta}-\frac{r_0}{r_0 ^ 2+a^2\cos^2\theta}\right]\ , \end{aligned}\ ] ] where @xmath157 and @xmath158 refer to the fixed value @xmath159 . for example , on the equatorial plane @xmath131 , eq .
( [ solkerr ] ) reduces to @xmath160 \left ( \frac{1}{\lambda-\lambda_0+r_0 } -\frac{1}{r_0}\right)\ .\end{aligned}\ ] ] in the special case @xmath161 and @xmath119 , i.e.,the particle starts moving from the horizon @xmath120 , we have @xmath121 and @xmath162\left ( \frac{1}{r}-\frac{1}{r_+}\right)\ .\end{aligned}\ ] ] if in addition @xmath123 , then we have @xmath163 we see that @xmath68 ( carrying spin weight @xmath164 ) is coupled ( via the kerr rotational parameter @xmath165 ) to @xmath166 ( i.e.,@xmath54 , with spin weight @xmath167 ) : in a sense , a helicity - rotation coupling factor @xmath168 is necessary to ensure spin - weight conservation on both sides of the equation for @xmath68 .
the behaviour of the varying component @xmath95 of the total 4-momentum as a function of the radial parameter is shown in fig .
[ fig:1 ] .
@xmath95 is always negative in this case , and goes to the asymptotic value @xmath169/r_+$ ] for @xmath170 .
note that in the schwarzschild limit ( @xmath171 ) @xmath95 vanishes identically .
the kasner metric @xcite representing an anisotropic universe model is s^2= t^2 + t^2p_1x^2 + t^2p_2y^2 + t^2p_3z^2 , with @xmath172 .
choosing kasner index values @xmath173 , @xmath174 ( and permutations ) one has the petrov type @xmath175 case , and introducing the np tetrad @xmath176\ , , \
n=\frac{t^{p_1}}{\sqrt{2}}[\partial_t - t^{-p_1}\partial_x]\,,\ m=\frac1{\sqrt{2}}[t^{-p_2}\partial_y+it^{-p_3}\partial_z]\,,\end{aligned}\ ] ] one finds the nonzero spin coefficients = t^-2p_2=2 , = -3 t^-p_2 , while the only surviving weyl scalar is @xmath177 .
the integral curves of @xmath8 are @xmath178^{3/2}\ , \quad x = x_0+\frac16(\lambda-\lambda_0)^2+\frac{t_0^{2/3}}{\sqrt{2}}(\lambda-\lambda_0)\ , \nonumber\\ y&=&y_0\ , \quad z = z_0\ , \end{aligned}\ ] ] and @xmath90 , @xmath103 , @xmath179 , @xmath180 .
consider the metric associated with a single gravitational wave @xcite in coordinates @xmath181 s^2=-hu^2+u v -x^2-y^2 , where = f_+(u)(x^2-y^2)+2f_(u ) xy , and @xmath182 represent the two possible polarization states of the wave .
the natural np tetrad is l=2_v , n=_u + h_v , m=(_x+i_y ) with the single nonvanishing spin coefficient = -and weyl scalar @xmath183 $ ] .
the integral curves of @xmath8 are u = u_0 , v=2(-_0)+v_0,x = x_0 , y = y_0 , with @xmath184 arbitrary integration constants taken at @xmath100 . the mathisson - papapetrou equations for the world line of the multipole reduction chosen within the congruence @xmath8 are @xmath185 with solution : @xmath90 , @xmath102 , @xmath186 and @xmath179 .
the case @xmath64 is trivial .
instead of choosing @xmath8 to be a ( geodesic ) principal null direction of the background spacetime , one can allow it to be the tangent vector to any other null geodesic .
null circular geodesics in some of the above type d spacetimes are of special interest ; in the black hole case they represent the last possible circular orbits as one approaches the horizon . to describe this case , eqs .
( [ eqsspinp_fin ] ) reduce to @xmath187 where we have set @xmath62 , assuming @xmath8 to be affinely parametrized , and in addition @xmath188 , because this is the case in all the explicit examples considered below and we have used the property that all of the nonvanishing spin coefficients ( @xmath189 denotes the constant value of @xmath145 when evaluated along the orbit ) and the weyl scalars are constant along the orbit . then the system can be easily integrated and the solution is @xmath190\ } -2s(\lambda_0)\,{\rm i m } [ \psi_2(\lambda_0)](\lambda-\lambda_0 ) + p^1(\lambda_0)\ , \nonumber \\
\fl p^3(\lambda ) & = & \pi_0 is(\lambda_0)-\frac{\rmd s_2}{\rmd \lambda}\ , \end{aligned}\ ] ] where @xmath191 satisfies the following equation = -|_0 along the orbit ) is real and positive , the solution for @xmath68 is given by s_2 ( ) = + i .
when instead @xmath192 is not purely real , we have to solve a constant coefficient second order system of coupled ordinary differential equations for the real and imaginary parts of @xmath68 , which is also trivial to do .
we give the essential details to obtain the solution for the kerr , schwarzschild , taub - nut and c metric spacetimes in the following subsections . in the case of the kerr solution
, it is well known that in the equatorial plane ( @xmath146 ) there exist two circular null geodesics at radii r_(geo ) , = 2m\ { 1 + } which are real solutions of the cubic equation r^3 - 6mr^2 + 9m^2r-4ma^2=0 . to discuss this case , we consider the following np frame field adapted to generic co - rotating and counter - rotating circular null orbits @xmath193 } \partial_\phi \ , \nonumber \\ \fl\quad n_\pm & = & \frac{(r^2+a^2)\sigma + 2a^2mr\sin^2 \theta}{2\sigma \delta}\left [ \partial_t + \frac{2amr\sin \theta \mp\sqrt{\delta}\sigma}{\sin \theta [ ( r^2+a^2)\sigma + 2a^2mr\sin^2 \theta ] } \partial_\phi\right]\ , \nonumber \\ \fl\quad m & = & \frac{1}{\sqrt{2\sigma}}\left ( \sqrt{\delta}\partial_r
+ i \partial_\theta\right)\ .\end{aligned}\ ] ] the congruence @xmath8 contains the null geodesic orbits in the equatorial plane .
along such orbits , @xmath146 and @xmath194 and the only nonvanishing spin coefficients reduce to @xmath195 and the nonvanishing weyl scalars are @xmath196 ^ 2\ , \quad \psi_2 = \frac{m}{2r^3}\ , \nonumber \\ \psi_4&= & -\frac{3m}{4r^5\delta } \left[r^2+a^2\pm a\sqrt{\delta } \right]^2\ , \end{aligned}\ ] ] where @xmath194 is assumed .
thus the spin coefficients and weyl scalars are all constant along the orbit . from eqs .
( [ eqsspinp_fin ] ) we have @xmath197 which can be easily integrated because all the coefficients are constant . in the limit of the schwarzschild spacetime
all these results simplify .
the np tetrad adapted to the null circular orbits is @xmath198 \ ,
\nonumber \\ m & = & \frac{1}{\sqrt{2}}\left ( \sqrt{1-\frac{2m}{r}}\partial_r + \frac{i}{r } \partial_\theta\right)\end{aligned}\ ] ] along the two equatorial null geodesics orbits @xmath146 and @xmath199 the only nonvanishing spin coefficients are = = -=- , while the nonzero weyl scalars _ 0=- , _
2= , _ 4=- , are all constant . from eqs .
( [ eqsspinp_fin_new ] ) we easily get the solutions .
introduce an np frame adapted to a generic circular null orbit @xmath200\ , \nonumber \\
m&= & \frac{\sqrt{2}}{2(r^2+\ell^2)^{1/2}}\left[\sqrt{\delta_{\rm tn}}\partial_r+i\partial_\theta \right]\ .\end{aligned}\ ] ] null geodesics exist off the equatorial plane satisfying the conditions @xmath201 the first equation gives r_(geo)=m+2 .
the only nonvanishing spin coefficients ( evaluated along the orbits , i.e. , using eqs .
( [ tnnullgeos ] ) to simplify general formulas ) therefore reduce to @xmath202\ , \quad \nonumber\\ \fl\quad \alpha&=&\bar\beta+\frac{\sqrt{2}}{2}\frac{r\sqrt{\delta_{\rm tn}}}{(r^2+\ell^2)^{3/2 } } \nonumber\\ \fl\quad & = & i\frac{\sqrt{2}}{4}\frac{\ell\sqrt{\delta_{\rm tn } } } { ( r^2+\ell^2)^{3/2}\sin\theta}\frac{4\ell\sqrt{\delta_{\rm tn } } ( r^2+\ell^2)\cos\theta-[(r^2+\ell^2)^2 + 4\ell^2\delta_{\rm tn}]\sin\theta}{(r^2+\ell^2)^2 - 4\ell^2\delta_{\rm tn}}\ , \nonumber\\ \fl\quad \nu&=&i\frac{\sqrt{2}}{8}\frac{(r^2+\ell^2)^2 + 4\ell^2\delta_{\rm tn } } { \delta_{\rm tn}(r^2+\ell^2)^{5/2}\sin^3\theta}[2\ell\sqrt{\delta_{\rm tn}}\sin\theta+(r^2+\ell^2)\cos\theta]\ , \end{aligned}\ ] ] and the nonvanishing weyl scalars are @xmath203 ^ 2 } \big\{(r^2+\ell^2)^4 + 16\ell^4\delta_{\rm tn}^2\nonumber\\ \fl\quad & & + 4\ell\sqrt{\delta_{\rm tn}}(r^2+\ell^2)[(r^2+\ell^2)^2 + 4\ell^2\delta_{\rm tn}]\sin\theta\cos\theta\big\}\ , \nonumber\\ \fl\quad \psi_2&=&\frac{i}{2}\frac{\ell - im}{(r+i\ell)^3}\ , \quad \psi_4 = -\frac{i}{4}\frac{[(r^2+\ell^2)^4 - 16\ell^4\delta_{\rm tn}^2]^2 } { \delta_{\rm tn}^2(r^2+\ell^2)^3(\ell+ir)^6}\psi_0\ .\end{aligned}\ ] ] the integration of the equations of motion then follows easily . introduce the following np frame @xmath204\ , \nonumber\\ n&= & \frac{1}{\sqrt{2}}[\sqrt{h}\rmd u + \frac{\rmd r -ar^2\sin \theta \rmd \theta}{\sqrt{h}}-r\sqrt{g}\rmd \phi]\ , \nonumber \\ m&= & \frac{1}{\sqrt{2}}\left [ \frac{\rmd r}{\sqrt{h}}+r\sin \theta \left(\frac{i}{\sqrt{g}}-\frac{ar}{\sqrt{h}}\right)\right]\ .\end{aligned}\ ] ] null geodesics exist in the equatorial plane ( @xmath146 ) at @xmath199 .
evaluated along this orbit , the only nonvanishing spin coefficients are : @xmath205 while the weyl scalars reduce to _ 0=- , _ 2= , _ 4=- .
the equations of motion are easily integrated .
we have studied the motion of massless spinning test particles in algebraically special vacuum spacetimes according to the mathisson - papapetrou model supplemented by pirani s conditions in the framework of the newman - penrose formalism .
we have considered the case in which the particle follows a null geodesic world line belonging to a congruence of a principal null direction of the background spacetime , a condition which allows complete integration of the equations of motion and for this case we have discussed explicit examples for both type d spacetimes ( schwarzschild , kerr , taub - nut , c metric , kasner ) and type n spacetimes ( gravitational wave ) , revealing helicity couplings with connection coefficients as well as the curvature of the background .
finally , we have also considered for some familiar type d spacetimes the motion of a massless spinning test particle along a circular null geodesic , a case which can also be completely solved analytically . | the motion of massless spinning test particles is investigated using the newman - penrose formalism within the mathisson - papapetrou model extended to massless particles by mashhoon and supplemented by the pirani condition . when the multipole reduction world line " lies along a principal null direction of an algebraically special vacuum spacetime ,
the equations of motion can be explicitly integrated .
examples are given for some familiar spacetimes of this type in the interest of shedding some light on the consequences of this model . |
fast outflows are a common feature of active galactic nuclei ( agns ) over more than four orders of magnitude in luminosity ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
agn outflows appear to be a substantial component of the nuclear environment , and their ubiquity suggests that mass ejection is probably linked to or even required for mass accretion onto a supermassive black hole ( smbh ) .
for example , outflows could provide a mechanism for expelling angular momentum from the accreting material ( e.g. , * ? ? ? * ; * ? ? ? * ) .
moreover , outflows in luminous agns may play an important role in the feedback of smbhs into typical massive galaxies ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the outflowing material could drive away sufficient gas from the host galaxy to quench both star formation and smbh growth , leading to the observed relations between the mass of the smbh and the properties of the galaxy bulge ( e.g. , * ? ? ?
the strongest observational signature of outflows from luminous agns ( i.e. , quasars ) is broad absorption lines ( bals ; @xcite ) in the ultraviolet ( uv ) ; these are seen in of optically selected quasars ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
aside from dust reddening , bal quasars generally have indistinguishable continuum spectral energy distributions ( seds ) from non - bal quasars from the infrared ( ir ) to the uv ( e.g. , * ? ? ?
* ; * ? ? ?
it has been suggested that all / most quasars have bal winds , with bals being observed only when inclination angles are large and the line of sight passes through the outflowing wind ( e.g. , @xcite ) .
the intrinsic fraction of bal quasars , after correcting for selection effects , is @xmath7 ( see , e.g. , @xcite and references therein ) , suggesting that the wind has an average covering factor of @xmath8 .
is the average value .
quasars with winds having larger covering factors would have larger chances of being observed as bal quasars . ]
an alternative hypothesis is that bal quasars represent an early evolutionary stage of quasars and/or the appearance as a bal quasar might be related to the duty cycle of smbh growth ( e.g. , * ? ? ?
* ; * ? ? ?
a promising scenario for bal quasar outflows is the accretion - disk wind model , where the wind is launched from the disk at @xmath910@xmath10 cm and is radiatively driven by uv line pressure ( e.g. , * ? ? ?
* ; * ? ? ?
figure [ fig - art ] is a schematic illustration of the model .
uv absorption - line profiles predicted by this model are consistent with observations of bal quasars ( e.g. , * ? ? ?
since line - driving becomes less efficient when the ionization state of the gas is too high , the disk - wind model has invoked `` shielding '' material to prevent the wind from being overionized by the extreme uv ( euv ) and soft radiation from the innermost accretion disk and its corona .
one proposed origin for the shielding gas is a `` failed wind '' , which is located at the base of the uv - absorbing wind and consists of material that does not reach escape velocity due to overionization ( e.g. , * ? ? ?
* ; * ? ? ?
the detailed geometry of the shielding gas is still uncertain ( e.g. , it could also perhaps `` hug '' the uv - absorbing wind ) .
10@xmath10 cm and is driven by uv radiation from the disk .
bals are observed when the line of sight passes through the outflowing wind .
the shielding gas located at the base of the wind blocks the euv and x - ray radiation from the nucleus and prevents the wind from being overionized .
one origin for the shielding gas could be a `` failed wind '' ; the small arrows in this gas represent the velocity field ( e.g. , * ? ? ?
for a standard accretion disk , the disk region that emits most strongly in the uv has a radius @xmath11 cm ; see section [ sec - physicalimp ] below .
[ fig - art ] ] under the above scenario , agns with observable disk winds would appear x - ray weak due to absorption by the shielding gas . indeed , bal quasars are in general weak , and absorption is often seen ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
moreover , the observed weakness of bal quasars is correlated with the absorption strength and maximum velocity of the uv bals ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , suggesting that the shielding gas does play a role in shaping the properties of the uv wind .
the absorption in bal quasars often appears complex ( i.e. , not just a simple photoelectric absorption cutoff but an ionized or partially covering absorber ) ; the measured absorption column densities ( @xmath12 ) are typically in the range of @xmath13 to @xmath14 @xmath3 , although absorption constraints usually have significant uncertainties due to limited photon statistics and poorly constrained absorption complexity ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
x - ray absorption variability has been observed in a few bal quasars on timescales of years ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , indicating that the shielding gas can be a dynamical structure subject to rotational and outflow / inflow motions , but the details of the dynamics remain unclear .
significant flux variability has also been seen in a few objects ( e.g. , * ? ? ?
however , when compared with the quasar population in general , bal quasars do not show exceptional variability @xcite .
overall , the nature of the shielding gas in bal quasars remains poorly constrained . with the current spectra of bal quasars observed by _
asca _ , _ chandra _ , or _
xmm - newton _ , the absorption column density can be constrained up to @xmath15 @xmath3 . at higher column densities ,
even photons at energies of kev are severely absorbed , and the observed spectra may be dominated by a scattered / reflected component .
it is thus difficult to determine the absorption properties for heavily obscured ( @xmath16 @xmath3 ) or compton - thick ( @xmath17 @xmath3 ; see , e.g. , @xcite for a review ) objects . supplemented with multiwavelength data ,
the level of weakness can be estimated by comparing the seds of bal quasars to those of typical quasars ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
specifically , in such comparisons the power - law slope parameter ( @xmath18 ) is often used ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , which is a measure of the soft ( 2 kev ) luminosity of a quasar relative to its optical / uv luminosity .
this parameter is known to be correlated with the 2500 monochromatic luminosity ( @xmath19 ) of the quasar ( e.g. , * ? ? ?
* ) , and thus @xmath20 , defined as the difference between the observed @xmath18 and the one expected from @xmath19 ( @xmath21 ) , indicates the level weakness of the source . for many bal quasars , the observed weakness can be attributed entirely to absorption .
it is clear that their photons have penetrated the obscuring material , and , after correcting for the moderate absorption , their fluxes recover to nominal levels ( ) . in these cases , the absorption - corrected 2 kev flux , and thus @xmath22 and @xmath23 , are generally estimated assuming a power - law spectrum ( e.g. , with a nominal photon index @xmath24 ) and normalizing it to the hard ( e.g. , rest - frame @xmath25 kev ) continuum ( e.g. , * ? ? ?
* ; * ? ? ?
for some of the sources with better spectra , absorption corrections can be derived directly via spectral fitting ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) .
it has also been found that the effective photon index determined from hardness ratios is correlated with the level of x - ray weakness ( @xmath20 ) , i.e. , x - ray weaker sources are harder , suggesting that absorption plays an important role in causing the x - ray weakness of the general bal quasar population ( e.g. , * ? ? ?
on the other hand , some bal quasars show significant weakness which can not be accounted for by the _
apparent _ x - ray absorption determined using @xmath26 kev data . in these cases
the continuum levels are still factors of lower than the expected sed levels after absorption corrections ( @xmath27 to @xmath28 ; e.g. , @xcite ) .
the nature of the x - ray weakness for such objects is uncertain .
it is possible that they are heavily obscured , and the observed spectra are dominated by the scattered / reflected component .
alternatively , these objects could be intrinsically weak compared to typical quasars , not emitting at a nominal level ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
an example of a mechanism that could cause intrinsic weakness is wind quenching of the accretion - disk corona , as proposed by @xcite , where the coronal x - ray emission is suppressed when the failed disk wind falls into the corona and makes it `` too dense , too opaque , and consequently too cold '' .
a few bal quasars have been suggested to have compton - thick absorption ( e.g. , * ? ? ?
* ; * ? ? ?
these studies are based on the weak / non detection of the source in the x - ray band , making detailed spectral analysis infeasible .
therefore the scenario of intrinsic weakness can not be excluded .
an open question is whether even stronger absorption could be present in weak bal quasars , or if instead some of these quasars are actually intrinsically weak .
one way to address this is to observe at higher energies where the are considerably more penetrating .
for example , the _ bepposax _ observation of the bal quasar mrk 231 revealed an absorption column density of @xmath29 @xmath3 , with only the @xmath3050 kev x - rays observed by _ bepposax _
able to penetrate the obscuring gas @xcite ; this discovery was later apparently confirmed by the _ suzaku _
data in the kev band @xcite . with the successful launch of the _ nuclear spectroscopic telescope array _ ( _ nustar _ ; @xcite ) on 2012 jun 13 , it is possible to investigate this question systematically .
_ nustar _ is the first focusing telescope in orbit observing in the hard x - ray band ; it provides about two orders - of - magnitude improvement in @xmath31 kev sensitivity over previous hard missions , as well as accurate source positions ( @xmath32 ) .
a _ nustar _
survey of significantly weak bal quasars can detect penetrating the absorber at @xmath31 kev unless the absorber is very compton - thick , and thereby better constrain the nature of the shielding gas and the disk - wind mechanism . as a pilot program , we selected two well studied bal quasars , pg 1004 + 130 and pg 1700 + 518 , that appear weak and observed them with _ nustar_. we selected these two targets from the 87 palomar - green ( pg ) quasars @xcite at @xmath33 .
these pg quasars represent one of the best - studied samples of luminous quasars in the nearby universe , and the more luminous pg quasars are also representative local counterparts of quasars ( including bal quasars ) at higher redshifts ( e.g. , @xmath34 ) from the sloan digital sky survey ( sdss ; @xcite ) in terms of luminosity .
five bal quasars have been classified within this pg quasar sample ( see footnote 4 of * ? ? ?
* ) , two of which ( pg 1001 + 054 and pg 2112 + 059 ) show soft weakness that was considered to be accounted for by moderate absorption ( e.g. , * ? ? ?
* ; * ? ? ?
weaker after absorption correction ) . ] the other three are also weak but the nature of their x - ray weakness is not as well understood ; among these three , we chose the two lower redshift objects as our targets here ( pg 1004 + 130 at @xmath35 and pg 1700 + 518 at @xmath36 ; the other object is pg 0043 + 039 at @xmath37 ) .
pg 1004 + 130 and pg 1700 + 518 are among the optically brightest bal quasars known ; see figure [ fig - lz ] .
also shown in figure [ fig - lz ] are the 87 pg quasars and the @xmath33 sdss data release 7 ( dr7 ; @xcite ) quasars from the catalog in @xcite .
pg 1004 + 130 and pg 1700 + 518 show clearly detected bals ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
aside from dust reddening in pg 1700 + 518 , their optical and uv spectra ( continua and emission lines , excluding the bal regions ) appear normal compared to typical quasars .
pg 1700 + 518 is also sub - classified as a low - ionization bal ( lobal ) quasar . ) of bal quasars that have bals from ions at lower ionization states such as or ( e.g. , * ? ? ?
* ; * ? ? ?
they often show signs of dust reddening and are weaker than hibal quasars ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
[ footnote - bal ] ] both objects are well studied and have superb multiwavelength coverage ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
pg 1004 + 130 is radio loud with a radio - loudness parameter ( @xmath38 ) ; such radio - loud bal quasars are relatively rare ( e.g. , * ? ?
* ; * ? ? ?
pg 1700 + 518 is radio quiet ( @xmath39 ) . in the band ,
pg 1004 + 130 and pg 1700 + 518 do not show the expected level of x - ray emission ( determined using the relation ) for a luminous quasar at energies below @xmath4010 kev .
chandra _ and
_ xmm - newton _ spectra of pg 1004 + 130 show modest absorption ( @xmath41 @xmath3 ) with a partial - covering absorbed power - law model , and the derived photon index is consistent with the typical value for radio - loud quasars ( @xmath42 ; e.g. , @xcite ) ; however , the x - ray continuum flux after correction for this absorption is still @xmath43 times lower than that expected from its optical / uv flux @xcite .
pg 1700 + 518 has been observed by _ xmm - newton _ with a @xmath44 ks exposure .
the spectrum is flat ( @xmath45 ) compared to the typical photon index of @xmath46 for radio - quiet quasars ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and the absorption column density was constrained to be @xmath47 @xmath3 with an absorbed power - law model @xcite .
after correction for this strong absorption , pg 1700 + 518 is still @xmath48 times x - ray weaker than expected .
it was also weakly detected in a @xmath40 ks exposure by _
chandra_(@xmath49 counts in the 0.58 kev band ; @xcite ) .
note that in these @xmath18 calculations the optical / uv fluxes of these two quasars have not been corrected for any intrinsic reddening , which would render them even x - ray weaker .
as mentioned in section 1.1 above , absorption column densities constrained by data below @xmath4010 kev could be biased for objects that are heavily obscured or even compton - thick , as the observed spectra are probably dominated by the scattered / reflected component . given the expected underlying x - ray continua assuming normal quasar seds for pg 1004 + 130 and pg 1700 + 518 ( i.e. , the underlying 2 kev luminosities satisfying the relation ) , if we were able to detect direct nuclear hard with _ nustar _ , we would expect to obtain @xmath0 - 600 hard x - ray ( @xmath1 kev ) counts . therefore , pg 1004 + 130 and pg 1700 + 518 are ideal targets for an initial sampling of the hard ( @xmath31 kev ) properties of weak bal quasars .
this paper is organized as follows . in section 2
we describe the _ nustar _ observations and our data - analysis approach .
we present photometric and spectroscopic properties of the two targets when available . in section 3
we present multiwavelength properties of the two targets , which show weak hard emission compared to typical quasars . in section 4
we infer the absorption column densities from the data assuming that the weak hard emission is caused by obscuration by the shield gas , and we discuss physical implications and issues related to the fe k@xmath5 line .
we also discuss the possibility of the two targets being intrinsically weak based on a coronal - quenching model .
we summarize in section 5 . throughout this paper , we use j2000.0 coordinates and a cosmology with @xmath50 km s@xmath51 mpc@xmath51 , @xmath52 , and @xmath53 ( e.g. , * ? ? ? * ) .
for the spectral modeling , we use the cosmic abundances of @xcite and the photoelectric absorption cross - sections of @xcite .
we adopt the terminology that has been used in previous studies to describe weakness ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) : the term `` x - ray weak '' indicates that the observed emission is significantly weaker than that expected from the optical uv continuum sed , while the term `` intrinsically weak '' refers to one possible cause for the observed weakness where the object simply does not produce x - ray emission at a nominal level ( one other apparent cause would be absorption ) .
_ nustar _ carries two co - aligned x - ray telescopes with a focal length of 10.15 m focusing hard photons ( 379 kev ) onto two shielded focal plane modules ( fpms a and b ; @xcite ) .
each fpm consists of four cdznte pixel sensors placed in a two - by - two array , providing a @xmath54 field of view at 10 kev .
_ has excellent angular resolution compared to previous hard x - ray missions , with a half - power diameter ( hpd ) of @xmath55 and a full width at half maximum ( fwhm ) of @xmath56 independent of energy .
pg 1004 + 130 and pg 1700 + 518 were observed by _ nustar _ with exposure times of @xmath57 ks and @xmath58
ks , respectively .
the details of the observations are listed in table [ tbl - obs ] .
we processed the data using the _ nustar _ data analysis software ( nustardas ) v0.9.0 with nustar caldb 20121126 .
cleaned calibrated event files were created using the nupipeline script .
for each source in each of the two fpms , we created images in five bands : , , , , and using the _ chandra _ interactive analysis of observations ( ciao ) v4.4 tool dmcopy .
the images are oversampled , and the pixel size is @xmath59 .
we searched for sources in these images using the ciao tool wavdetect @xcite with a false - positive probability threshold of 10@xmath60 and wavelet scales of 2 , 4 , 8 , and 16 pixels .
pg 1004 + 130 is relatively bright , and it is detected in multiple bands in both fpms .
pg 1700 + 518 appears to be faint , and it is detected only in the 420 kev image of fpm a. the background in fpm b around the source position of pg 1700 + 518 is @xmath3040% higher than that in fpm a at lower energies ( @xmath61 kev ) , rendering the source undetectable in this fpm . this higher level of background in fpm b is caused by a larger level of stray light at the position of the source due to unfocused aperture leakage @xcite .
we adopted positions based on the wavdetect detection in the 420 kev band , which appears to be the most sensitive band among the five bands we studied , and the positions appear good upon visual inspections .
for faint sources , _ nustar _ provides positional accuracy to better than 5 .
the position of pg 1004 + 130 in fpm a is 3.1 away from its optical position , and in fpm b the offset is 0.1 . for pg 1700 + 518 in fpm a , the position is 1.5 away from the optical position .
overall , _ nustar _ provided accurate positions for these two bal quasars , and the positional offsets are within expectations for faint sources .
this assures us that the emission detected comes from our two targets .
neither of the two objects is detected in the or 3079 kev bands .
more than 100 net counts would be expected for either source in the 2030 kev band if it had a typical quasar sed ( i.e. , a 2 kev luminosity given by the relation and a power - law x - ray continuum with @xmath46 ) .
this suggests that their hard photons did not penetrate the obscuring material or they are intrinsically x - ray weak .
no serendipitous sources were detected in the fields of view of the _ nustar _
the 420 kev images of pg 1004 + 130 and pg 1700 + 518 in fpm a are shown in figure [ fig - img ] , centered on the source positions .
we performed aperture photometry to extract source counts in the _ nustar _ bands .
background counts were extracted from an annular region centered on the x - ray position with inner radius 120 and outer radius 180 .
different choices of the background - extraction region ( e.g. , a circular region in the source vicinity but outside the 120-radius aperture ) do not appear to affect the background estimate systematically , and the number of background counts generally fluctuates at the @xmath62 level . for the relatively bright source , pg 1004 + 130 , we used a 60-radius circular aperture to extract source counts ; this aperture approximates the @xmath63 encircled - energy fraction ( eef ) contour of the point spread function ( psf ) . extended jet emission from pg 1004 + 130 has been detected by _
chandra _
@xcite , and it is included in the _ nustar _ aperture extraction here .
however , the observed x - ray flux from the extended jet is only @xmath64 of the nuclear flux @xcite , and therefore it does not affect our analysis of the nuclear emission . pg 1700 + 518 is only weakly detected in fpm a , and thus we chose a smaller source - extraction region to optimize the signal - to - noise ratio .
we used a circular aperture with a radius of 25 , corresponding to an eef of @xmath65 . for both sources in each band , we derived their aperture - corrected net ( background - subtracted ) counts .
we calculated a binomial no - source probability , @xmath66 , to assess the significance of the source signal , defined as @xmath67 in the above equation , @xmath68 is the total number of counts in the source - extraction region ; @xmath69 , where @xmath70 is the total number of counts in the background region ; @xmath71 , where @xmath72 is the area scaling factor between the background and source regions . @xmath66
represents the probability of observing the source counts by chance ( due to a background fluctuation ) under the assumption that there is no source at the relevant location .
it has been used to filter out low - significance sources and create reliable catalogs of _ chandra _ sources ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
for our two targets here , if the @xmath66 value in a band is smaller than 0.01 ( @xmath73 ) , we considered the source to be detected and calculated the 1@xmath74 errors on the net counts , which were derived from the 1@xmath74 errors @xcite on the extracted source and background counts following the numerical method in section 1.7.3 of @xcite .
if the @xmath66 value is larger than 0.01 , we considered the source undetected in this band and derived an upper limit on the source counts using the bayesian approach of @xcite for a 90% confidence level . under this criterion ,
pg 1004 + 130 and pg 1700 + 518 are detected in the , , and bands ( except for pg 1700 + 518 in fpm b ) .
the source counts and upper limits of the two targets are listed in table 2 . using the band ratio , defined here as the ratio between the observed 1020 kev and 410 kev counts , we derived an effective photon index ( @xmath75 ) for a power - law model with the galactic absorption column density ( table 1 ) .
we utilized the _
nustar _ spectral response files ( produced by the pipeline extraction of the spectrum at the source location ) and the fakeit command in xspec ( version 12.8.0 ; @xcite ) to calibrate the relation between the effective photon index and band ratio .
similarly , we calibrated a count - rate - to - flux conversion factor that depends on the effective photon index assuming a power - law model , and then converted the source count rates to fluxes . the 1@xmath74 error for @xmath75
was derived using the errors on the counts , and flux errors were derived using the errors on the counts and @xmath75 .
the effective photon indices and fluxes of the two targets are shown in table 2 .
pg 1004 + 130 appears to be a fairly soft source with @xmath76 , while pg 1700 + 518 has a hard spectrum with @xmath77 . the small @xmath75 value ( although with a large uncertainty ) for pg 1700 + 518 suggests that significant absorption ( @xmath78 @xmath3 ) and likely also compton reflection is present .
note that a column density of @xmath47 @xmath3 was derived from the _ xmm - newton _ data with an absorbed power - law model @xcite .
the luminosities in the 420 kev band ( listed in table 2 ) for the two objects are @xmath79 erg s@xmath51 and @xmath80 erg s@xmath51 , smaller than expectations for typical quasars ( @xmath81 erg s@xmath51 ) .
the photometric properties of pg 1004 + 130 in fpms a and b appear consistent . for pg 1700 + 518 ,
the upper limits on the counts and fluxes in fpm b are consistent with those measurements in fpm a. spectral analysis for the _ nustar _
spectra of pg 1700 + 518 is not feasible as the extracted spectra are dominated by background .
however , we were able to perform basic spectral analysis for pg 1004 + 130 .
we extracted spectra of pg 1004 + 130 in fpms a and b using the nustardas script nuproducts .
the same source- and background - extraction regions as used for the photometry above were adopted , and psf corrections have been applied to the auxiliary response files ( arfs ) . to extend the spectral coverage to lower energies ( @xmath82 kev ) , we fit the _ nustar _ data jointly with a _
chandra _ spectrum .
the 41.6 ks _ chandra _ observation of pg 1004 + 130 with the s3 ccd of the advanced ccd imaging spectrometer ( acis ; @xcite ) was described in detail in @xcite .
we extracted the _
chandra _
spectrum using the ciao tool specextract , with a circular source aperture of 4 in radius and a source - free background annulus of 1220 in radius .
all the spectra were grouped with a signal - to - noise ratio of 5 , and we fit the 0.38 kev _ chandra _ and 320 kev _ nustar _
spectra together with a partial - covering absorber model as suggested by @xcite .
the xspec model used was zpcfabs*zpowerlw*wabs , where zpcfabs is a partial - covering absorption model , zpowerlw is an underlying power - law spectrum , and wabs is to account for galactic absorption .
the spectral - shape parameters for the model ( absorption column density @xmath12 , covering factor @xmath83 , and photon index @xmath84 ) were free to vary but tied for the three spectra , and we let the normalization parameters for the three spectra vary to allow for possible flux variation and cross - calibration uncertainties .
the joint spectra overlaid with the best - fit model are displayed in figure [ fig - spec ] .
the best - fit model is statistically acceptable , with a null - hypothesis probability of 0.39 ( @xmath85 ) .
the model parameters are @xmath86 @xmath3 , @xmath87 , and @xmath88 ; the quoted errors are at the 90% confidence level for one parameter of interest ( @xmath89 ) .
these parameters are comparable to those derived in @xcite using the _ chandra _ data alone .
the intrinsic photon index , @xmath88 , is consistent with the @xmath42 typical of radio - loud quasars ( e.g. , * ? ? ?
we also tried to fit only the hard data ( 420 kev ) with a simple power - law model modified by galactic absorption .
the derived photon index is @xmath90 for either the _ nustar _ data
alone or the _ nustar _ plus
_ chandra _ data set , consistent with the @xmath75 value estimated from the band ratio for pg 1004 + 130 ( table 2 ) .
the apparent moderate absorption and soft spectral shape of pg 1004 + 130 do not suggest compton - thick absorption .
however , the apparent absorption is not sufficient to explain the weakness of this object , and a compton - thick absorber may still be present , if the observed spectra are dominated by a fraction of the jet x - ray emission that is not obscured by the absorber ( see the further discussion in section [ sec - physicalimp ] below ) . in the best - fit partial - covering absorber model , the normalization parameters for the _ nustar _
spectra in fpms a and b are only 42% and 46% of that for the
_ chandra _ spectrum , indicating that the 38 kev flux of pg 1004 + 130 observed by _ nustar _ has dropped by a factor of @xmath91 compared to the _ chandra _ observation in 2005 .
the _ nustar _
fluxes derived from the best - fit model are consistent with those derived from the photometric approach above ( table 2 ) within the 1@xmath74 errors , suggesting that the flux discrepancy is not likely caused by uncertainties introduced during the xspec fitting .
there was a simultaneous _ swift _ xrt observation of pg 1004 + 130 during the _ nustar _ observation . however , the xrt exposure is only 2.0 ks , and pg 1004 + 130 is not detected
. the upper limit on the xrt flux does not provide useful constraints on the variability .
pg 1004 + 130 is known to be variable in the band .
the 28 kev _
chandra_flux is @xmath92 times the _ xmm - newton _ flux observed in 2003 , and it is @xmath93 times the 28 kev flux limit inferred from the 1980 _ einstein _ 0.54.5 kev nondetection @xcite .
therefore , it is likely that the flux of pg 1004 + 130 has decreased by a factor of @xmath91 in the 2012 _ nustar _
observation compared to its _ chandra _ flux in 2005 .
we caution that variability between _ nustar _ and _ chandra _ observations might also affect the best - fitting model above , since we tied the spectral - shape parameters in the modeling and spectral variability has been observed in several bal quasars ( although it is not well constrained ; e.g. , @xcite ) .
we note that there is no apparent fe k@xmath5 line emission at rest - frame 6.4 kev ( 5.2 kev in the observed frame ) shown in the spectra in figure [ fig - spec ] , as has been generally observed in spectra dominated by a reflection component ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
adding a narrow line at 6.4 kev with a fixed width of 0.01 kev does not improve the fit .
the 90% confidence - level upper limit on the rest - frame fe k@xmath5 line equivalent width ( ew ) is @xmath94 ev . there is no evidence for a he - like or h - like fe k@xmath5 line at 6.7 kev or 7.0 kev either . for details ) .
there is no apparent fe k@xmath5 line emission at rest - frame 6.4 kev ( marked by the green arrow ) as is typically seen in a reflection - dominated spectrum .
[ fig - spec ] ]
we constructed radio - to - x - ray seds for pg 1004 + 130 and pg 1700 + 518 , which have superb multiwavelength coverage .
we collected photometric data from the literature @xcite , the nrao vla sky survey ( nvss ; @xcite ) , the _ wide - field infrared survey explorer _ ( _ wise _ ; @xcite ) , the two micron all sky survey ( 2mass ; @xcite ) , the sdss , and/or the _ galaxy evolution explorer _ ( _ galex _ ; @xcite ) catalogs . we adopted the @xmath95 and/or @xmath96 band magnitudes from @xcite , which have the host - galaxy contribution removed via two - dimensional fitting of the images .
we also obtained _ hubble space telescope _ ( _ hst _ ) ghrs and _ international ultraviolet explorer _ ( _ iue _ ) uv spectra for pg 1004 + 130 from the mikulski archive for space telescopes ( mast ) , and an _ hst _ fos spectrum for pg 1700 + 518 from @xcite .
the optical and uv data have been corrected for galactic extinction following the dereddening approach presented in @xcite . for pg 1004 + 130
, we adopted the spectral model ( with the _ nustar _ normalization ) determined in section [ sec - spec ] along with the upper limit on the 40 kev _ nustar _
luminosity derived from the photometric information in table 2 . for pg 1700 + 518 , we used the 2 kev _
chandra_luminosity @xcite , kev x - ray emission of pg 1700 + 518 likely has a small contribution from its star - forming activity ( @xmath97 ; @xcite ) , which does not affect our analyses here . ]
7 kev and 15 kev _ nustar _
luminosities , and 25 kev and 40 kev _ nustar _
luminosity upper limits ; the _ nustar _
data are again derived from the photometric information in table 2 .
the rest - frame seds are shown in figure [ fig - sed ] .
the @xmath98 bal features of these two quasars are also shown in figure [ fig - sed ] as inset panels : the broad troughs are clearly visible with the one in pg 1700 + 518 being deeper and broader . to compare the seds of these two bal quasars to those of typical quasars
, we overlaid in figure [ fig - sed ] the composite radio - loud ( for pg 1004 + 130 ) quasar seds from @xcite and @xcite or radio - quiet ( for pg 1700 + 518 ) quasar seds from @xcite and @xcite.@xmath99 relation and is thus consistent with our interpretation of the underlying quasar x - ray spectra here .
] we extrapolated the @xcite sed to higher energies assuming a @xmath100 power law to provide a fair comparison with the observed _ nustar _
data ; studies of high - redshift quasars ( e.g. , * ? ? ?
* ; * ? ? ?
* ) indicate that their rest - frame @xmath3040 kev continua generally follow such a power law .
the composite seds were normalized to the observed seds at the @xmath95 band , where the seds are largely free of intrinsic reddening and the data have been corrected for host - galaxy contamination @xcite .
the composite quasar seds from different studies agree well in general except that the @xcite radio - quiet sed is biased toward x - ray bright quasars . from the radio to the uv , the continuum seds of our two targets agree well with the composite seds .
the sed of pg 1700 + 518 shows intrinsic reddening in the optical and uv , which is consistent with previous findings that bal quasars , and especially lobal quasars , are in general redder than non - bal quasars ( e.g. , * ? ? ?
* ; * ? ? ?
? * ; * ? ? ?
* ) . by comparing the pg 1700 + 518
sed to the composite seds , we estimated the reddening to be @xmath101 , in agreement with the average value for lobal quasars ( e.g. , * ? ? ? * ) .
both bal quasars are significantly weak compared to typical quasars .
we computed the @xmath18 parameter , defined as @xmath102 , for quantitative comparison .
the rest - frame 2500 flux density ( @xmath103 ) was determined by interpolating / extrapolating the photometric data points . for pg 1700 + 518 ,
the dust reddening was not corrected ; reddening correction would increase the 2500 flux density by @xmath65 ( i.e. , more negative @xmath18 ) .
the rest - frame 2 kev flux density was derived from the best - fit spectral model ( see section [ sec - spec ] above ) for pg 1004 + 130 , and it was adopted from the weighted - average value of _ chandra _ and _ xmm - newton _ observations in @xcite and @xcite for pg 1700 + 518 .
the @xmath18 values are @xmath104 and @xmath105 for pg 1004 + 130 and pg 1700 + 518 , respectively .
the @xmath18@xmath99 plot is shown in figure [ fig - aox ] . for comparison ,
the @xcite relation for radio - quiet quasars predicts @xmath106 for pg 1700 + 518 .
pg 1004 + 130 is radio loud , and the @xcite relation is not applicable .
we thus adjusted the expected @xmath18 value by accounting for the excess luminosity expected for a radio - loud quasar with the radio loudness of pg 1004 + 130 ( see the relation in section 4 of @xcite ) .
the resulting expected @xmath18 value for pg 1004 + 130 is @xmath107 .
therefore , although the continuum radio - to - uv seds of these two bal quasars resemble those of typical quasars , their soft ( @xmath108 kev ) luminosities are @xmath109 ( for pg 1004 + 130 , with a 1@xmath74 range of 376 ) and @xmath110 ( for pg 1700 + 518 , with a 1@xmath74 range of 36400 ) times lower than the typical values .
after corrections for apparent absorption determined from @xmath26 kev data , their soft luminosities are still 11 and 12 times lower than expected ( see section 1.2 ) . it has been suggested that there is an additional correlation between @xmath18 and the eddington ratio ( e.g. , * ? ? ?
* ) , where @xmath18 is lower ( i.e. , more x - ray weak ) when the eddington ratio is higher .
however , the correlation suggests super - eddington accretion for @xmath111 ( also lacking sampling in this regime ) , which does not appear to be the case for our two targets ( see section 4.1.2 for their eddington ratios ) . in the hard bands probed by _
nustar _ , the luminosities of pg 1004 + 130 and pg 1700 + 518 are also below expectations . at rest - frame 20
kev , the _ nustar _ luminosities are more than an order of magnitude lower than those of typical quasars ; even at rest - frame 40 kev , where the _ nustar _
sensitivity to faint sources is lower , the 90% confidence - level upper limits on the luminosities are still a few times smaller than expectations . if these two bal quasars are able to produce emission as typical quasars do , the _ nustar _
data indicate that compton - thick obscuration is present that blocks not only the soft x - rays but also the hard x - rays .
@xmath99 relation .
the dashed red line shows the @xcite relation modified with the excess luminosity expected for the radio loudnesss of pg 1004 + 130 ( derived based on section 4 of @xcite ) .
the errors on @xmath18 for the two bal quasars were propagated from the errors on the fluxes ; it is smaller than the symbol size and is thus not visible for pg 1004 + 130 .
pg 1004 + 130 and pg 1700 + 518 are @xmath109 and @xmath110 times weaker than expected in the soft x - rays ( @xmath112 kev ) , respectively .
[ fig - aox ] ]
given the _ nustar _ and
multiwavelength data presented above for the two bal quasars targeted , it appears that either ( 1 ) there is highly compton - thick obscuration and even the hard x - rays probed by _ nustar _ can not penetrate the obscuring material , or ( 2 ) these objects are intrinsically weak . in the following ,
we discuss these two scenarios in more detail . for many bal quasars
, the observed x - ray weakness can be fully accounted for by absorption ( see section 1.1 ) .
the estimated absorption column densities for these objects have a continuous distribution between @xmath113 @xmath3 and @xmath15 @xmath3 , and thus we expect that there should probably be objects with higher column densities ( e.g. , @xmath114@xmath115 @xmath3 ) as well .
furthermore , one of our targets , pg 1700 + 518 , shows an effective photon index of @xmath116 , suggesting significant absorption .
therefore , it is natural to consider the possibility of compton - thick absorption for objects that show significant hard x - ray weakness . the underlying continua of these objects would then be comparable to those of typical quasars , but their observed x - ray spectra would probably be dominated by the scattered / reflected component . in this scenario , we can estimate the expected absorption column density for pg 1004 + 130 and pg 1700 + 518 based on the ratio of the observed broad - band flux and the expected intrinsic 20 kev flux density for a given value of @xmath18 ( cf .
@xcite ) ; to obtain this intrinsic 20 kev flux density , we determined the intrinsic 2 kev flux density using @xmath18 and extrapolated to 20 kev assuming a power - law spectrum with @xmath24
. parameter has also been defined at energies higher than 2 kev ( e.g. , 10 kev ; @xcite ) .
the @xmath18@xmath99 relation with @xmath18 defined at 2 kev has a relatively small dispersion due to the small errors in the 2 kev flux measurements , and it is therefore adopted here .
the intrinsic 20 kev flux density predicted with @xmath18 defined at 10 kev would lead to consistent results within the errors . ] the relation between this observed ratio and the column density was calibrated using the mytorus model @xcite implemented in xspec .
the mytorus model computes the transmitted and scattered spectra from a toroidal - shaped absorber / reprocessor in a physical and self - consistent way , and it was designed to model spectra in the compton - thick regime .
the model was calculated for neutral material . however ,
for our purpose of constraining the basic absorption column density in these bal quasars , the model is likely also applicable to ionized material , as high - energy attenuation is dominated by compton scattering ( not photoelectric absorption ) in the compton - thick regime .
two important geometrical parameters of the mytorus model are the angle of the obscuring medium and the inclination angle ( 0 corresponds to a face - on viewing angle ) .
the default angle was set to 60 ( corresponding to a covering factor of 0.5 ) , and we assumed an inclination angle of 80 ( large inclination angles are generally expected for bal quasars ; see section 1.1 ) .
we also explored the effects of different inclination angles and a different geometry with a angle of 37(corresponding to a covering factor of 0.8 ) .
the other parameters of the model , such as the relative cross - normalization factors of different components , were set as the default values ( see section 8.2 of the mytorus manual ) .
we derived column - density constraints using the _ nustar _ fluxes in three bands , , 1020 kev , and 2030 kev ( see table 2 ) , under a range of assumed @xmath18 values .
since neither object is detected in the 2030 kev band , the column densities constrained in this band are 90% confidence - level lower limits .
the results are displayed in figures [ fig-1004nh ] and [ fig-1700nh ] .
the harder bands ( e.g. , 1020 kev ) probed by _
nustar_generally provide much tighter constraints than the softer bands ( e.g. , ) .
the different geometric assumptions ( angle and inclination angle ) also affect the results somewhat .
note that the highest column density available in the mytorus model is @xmath115 @xmath3 , and thus any constraint above this value was derived from extrapolation and may have a large uncertainty .
nevertheless , compton - thick absorption appears required for any typical assumption about the intrinsic @xmath18 value . for the @xmath18 values expected from the @xmath18@xmath99 relation ( see section [ sec - sed ] ) , shown as the vertical dotted lines in figures [ fig-1004nh ] and [ fig-1700nh ] , the 1020 kev _ nustar _
data indicate @xmath117 @xmath3 for pg 1004 + 130 and @xmath118 @xmath3 for pg 1700 + 518 , for a angle of 60 and an inclination angle of 80 .
these column densities correspond to thomson optical depths of @xmath119 .
the uncertainty of the estimated @xmath12 was determined from the scatter in the expected @xmath18 value which is much more significant than the uncertainty of the observed flux . at @xmath120 @xmath3 , the observed spectrum computed from the mytorus model
is completely dominated by the scattered component ; it appears flat with a high - energy hump peaking around observed - frame 20 kev ( e.g. , see figure 6.1 of the mytorus manual ) , and the observed 1020 kev flux is absorbed by a factor of @xmath1 .
the parameterization of the mytorus model can not , of course , fully reproduce the complex absorption environments of these bal quasars , but we consider it to be the best available approximation for the purpose of deriving basic column - density constraints .
if the observed emission has contributions from additional continuum components that were not considered in the mytorus modeling , our estimation of the column density above would likely be an underestimate as the scattered / reflected component associated with the absorber would be weaker than the observed emission .
examples of possible additional x - ray components include jet emission and the nuclear continuum scattered by a large - scale medium ; see section [ sec - feline ] below for details . in the disk - wind scenario for bal quasars ,
the outflowing wind is mainly radiatively driven by uv line pressure . in some models ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , the uv radiation originates from the center of the agn and is filtered by the shielding gas .
this assumption requires the shielding gas not be compton thick ; otherwise , the uv radiation would also be blocked and the wind would lose its driving force .
in other models ( e.g. , * ? ? ?
* ) , this assumption is relaxed , and the uv radiation is allowed to be produced exterior to the shielding gas ( e.g. , see figure [ fig - art ] ) ; our finding of potential compton - thick absorption in these two bal quasars supports this latter geometry .
given the surface - temperature distribution of a standard accretion disk @xcite , the radius of the disk region that emits most strongly in the uv can be estimated as @xmath121 where @xmath122 is the accretion efficiency and @xmath123 is the accretion rate in units of the eddington accretion rate ( eddington ratio ) .
the black - hole masses ( @xmath124 ) for pg 1004 + 130 and pg 1700 + 518 are @xcite and @xcite , respectively ( the radius does not depend strongly on the black - hole mass , and thus the uncertainty on the mass estimate does not affect the derived compactness of the absorber significantly ) .
their bolometric luminosities can be calculated using the normalized @xcite or @xcite composite seds in figure [ fig - sed ] , and they are @xmath125 erg s@xmath51 for pg 1004 + 130 and @xmath126 erg s@xmath51 for pg 1700 + 518 .
the eddington ratios are then @xmath127 for pg 1004 + 130 and @xmath128 for pg 1700 + 518 .
the derived uv - emitting radii are @xmath129 ( ) for pg 1004 + 130 and @xmath130 ( @xmath9 cm ) for pg 1700 + 518 , where @xmath131 is the schwarzschild radius .
therefore , compton - thick absorption constrains the absorbing medium ( i.e. , shielding gas ) to be compact , located within @xmath9 cm of the smbh .
recent microlensing studies ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) suggest that the optical / uv emitting regions of quasar accretion disks are @xmath13230 times larger than those predicted from the standard model , and thus the above constraint on the compactness of the absorbing medium could be relaxed .
pg 1004 + 130 is a radio - loud quasar . under the assumption that its intrinsic emission is normal , most ( based on its radio loudness ) of its emission should come from the radio jets ( see section 4 of * ? ? ?
the observed weakness thus implies that the jet emission is also heavily absorbed .
based on an survey of 21 radio - loud bal quasars , @xcite concluded that jet emission is likely partially absorbed in these objects . therefore , it is possible that the underlying jet emission of pg 1004 + 130 is mostly absorbed by the same compton - thick material ( i.e. , the shielding gas ) that blocks the other nuclear emission .
the observed spectra may even be dominated by the unobscured portion of the jet emission if the nuclear emission is strongly absorbed .
such a jet - dominated spectrum can explain the soft spectral shape ( @xmath133 ; see section [ sec - spec ] ) observed that is consistent with those for radio - loud quasars , and it may also be responsible for the nondetection of the fe k@xmath5 line ( see details below ) . for a compton - thick agn with obscuration by neutral matter , a strong narrow fe k@xmath5 emission line at 6.4 kev with an ew of order 12 kev is expected if the continuum is reflection dominated ( e.g. , * ? ? ?
* ; * ? ? ?
a strong fe k@xmath5 line is observed in the majority of the known compton - thick agn population ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
there is no fe k@xmath5 line detected in the spectra of pg 1004 + 130 ; the upper limit on the ew is @xmath134 ev ( see section [ sec - spec ] ) . for pg 1700 + 518 ,
no fe k@xmath5 line is detected either , though the _ xmm - newton _ spectrum can not constrain an upper limit due to the dominating high background at high energies , which prevents even a detection of the continuum in the fe k@xmath5 band @xcite .
therefore , pg 1004 + 130 appears to be an unusual compton - thick agn without a strong fe k@xmath5 line .
however , pg 1004 + 130 is a luminous radio - loud type 1 bal quasar , and the physical nature ( e.g. , location , geometry , and ionization state ) of compton - thick absorption in such objects might differ from those in local type 2 seyfert galaxies .
one plausible cause for pg 1004 + 130 lacking a strong fe k@xmath5 line is that the line is diluted by a jet - linked continuum which could dominate over the scattered / reflected nuclear continuum ( cf .
jet dilution of the fe k@xmath5 line has been observed in the general population of radio - loud agns ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . in the compton - thick regime ,
the fe k@xmath5 line flux drops rapidly ( although the line ew increases ) when @xmath12 increases , especially at large inclination angles ( e.g. , * ? ? ?
therefore , the ew of the fe k@xmath5 line could be reduced substantially by any increase of the continuum level . utilizing the mytorus model above including the default fe k@xmath5 emission - line component and assuming @xmath135 @xmath3 , we estimated that @xmath136 ( the fraction that would make pg 1004 + 130 @xmath109 times weaker than expected ; see section [ sec - sed ] ) of the intrinsic jet continuum that is not absorbed could dilute the ew of the fe k@xmath5 line from @xmath137 kev to @xmath138 ev .
if the covering factor of the shielding gas is smaller than the mytorus default value ( 0.5 ) , which is likely the case given the small covering factor of the disk wind ( @xmath8 ; see section 1.1 ) , the ew of the fe k@xmath5 line could be even smaller ( e.g. , * ? ? ?
we caution that with the dilution from the jet emission , the column density estimated in section [ sec - columnd ] is likely a lower limit . for completeness
, we mention below a few additional possible explanations for the lack of a strong fe k@xmath5 line from pg 1004 + 130 , which might also be applicable for other bal quasars . 1 .
the strong fe k@xmath5 line could be diluted by the scattered continuum from a highly ionized compton - thin medium that surrounds the smbh on a larger scale than the compton - thick material ( e.g. , * ? ? ?
* ; * ? ? ?
we note that the scattering geometry in the may differ significantly from that studied spectropolarimetrically in the optical ( e.g. , the broad emission - line region or the shielding gas itself could be responsible for the scattered light seen in the optical ; @xcite ) .
the scattering medium must be highly ionized so that there is no strong fe k@xmath5 line produced in the scattering process ( see more discussion about ionization state in point 2 below ) . in the disk - wind model of bal quasars
, the compton - thin scattering medium could be the hot low - density outflow in the polar region ( e.g. , * ? ? ?
* ) , although the column density and covering factor of the outflow shown in the simulation results of @xcite are probably insufficient to produce the @xmath139 scattering fraction for diluting the line ew to the @xmath140 ev level , and the simulations were not designed for radio - loud objects with jets ( such as pg 1004 + 130 ) either . with such continuum dilution ,
the column density estimated in section [ sec - columnd ] is again likely a lower limit .
2 . the strength and centroid energy of the fe k@xmath5 line depend on the ionization state of the scattering / reflecting medium ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
as the ionization parameter ( @xmath141 ) increases , the ew of the line generally decreases and the centroid energy increases ; for a highly ionized medium ( @xmath142 ) , the ew of the fe k@xmath5 line can drop below 100 ev ( e.g. , * ? ? ?
* ) . in the disk - wind model of bal quasars ,
the scattering / reflecting medium is the shielding gas at the base of the disk wind , and it is likely ionized and could be highly ionized with @xmath1436 ( e.g. , * ? ? ?
* ) .a ) , the location of the absorber ( @xmath9 cm ) , the estimated column density ( @xmath144 @xmath3 ) , and an assumed size of the absorber ( @xmath9 cm ) , we estimated the ionization parameter to be @xmath145 ; this is likely an overestimate , as we neglected the effects of compton scattering in the estimate and also the absorber could be located further out . ]
therefore , a bal quasar may be compton - thick but without a strong fe k@xmath5 line due to highly ionized compton - thick absorption , although it would appear difficult to ionize highly a compton - thick medium due to the effects of compton scattering ( e.g. , see section 3 of @xcite ) .
the strong narrow fe k@xmath5 line could be affected by line broadening ( or smeared out in the extreme case ) if the absorber / reflector has outflow motion and consists of multiple velocity components . while the kinematics of the absorbing component of the disk wind are still uncertain , broadening and shifting of the fe k@xmath5 emission line has perhaps been seen in bal quasars ( e.g. , * ? ? ? * ; * ? ? ?
* ; * ? ? ?
we consider these explanations less likely than jet - linked dilution but still possible .
of course , it is also possible that pg 1004 + 130 is not compton - thick and thus does not have a prominent fe k@xmath5 line feature . in this case
, it would not produce emission as typical quasars do and would be intrinsically weak , as discussed in the next subsection .
pg 1004 + 130 and pg 1700 + 518 could also be intrinsically weak . a small fraction of quasars
have been suggested to be intrinsically weak ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
one of the best - studied cases is the narrow - line type 1 quasar phl 1811 , which is rapidly variable by a high amplitude but always appears weak and shows no evidence for intrinsic absorption @xcite .
a systematic survey for such intrinsically weak quasars demonstrated that these objects are rare in optically selected ( non - bal ) quasar samples ; the fraction of sdss quasars with @xmath146 is @xmath147 @xcite .
many bal quasars are weak due to absorption ( see section 1.1 ) , and thus they are often excluded in searches for intrinsically weak quasars ( e.g. , * ? ? ?
* ; * ? ? ?
however , intrinsic weakness is in fact an attractive possibility to account for the bal nature in the disk - wind scenario . as introduced in section 1.1 , to launch successfully a wind through the uv line - driving mechanism , the soft emission from the nucleus must be shielded to prevent the wind from being overionized .
if the nucleus were incapable of producing strong emission , the wind could be launched with little / no shielding , and then a bal quasar would be observed if the viewing angle were appropriate .
it is clear that some bal quasars do emit at a nominal level as their fluxes recover to expected levels after absorption corrections ( see section 1.1 ) .
however , there are other bal quasars , including pg 1004 + 130 and pg 1700 + 518 , that still appear weak after basic absorption corrections ; these are candidates for being intrinsically weak quasars .
the nature of intrinsic weakness remains unclear .
possible explanations generally invoke mechanisms that weaken or destroy the emitting accretion - disk corona ( see , e.g. , the discussions in @xcite and @xcite ) .
one specific mechanism relevant to bal quasars was proposed by @xcite , who suggested that part of the accretion - disk outflow ( i.e. , a dense , highly ionized `` failed wind '' produced by overionization ) could fall into ( see figure 1 for the velocity field ) the corona and thereby suppress its emission . with the presence of dense gas from the outflow ,
the coronal magnetic field becomes insufficient to liberate and transport energy from the disk to heat the corona via magnetized bubbles , and also relativistic electrons in the corona will cool efficiently via bremsstrahlung instead of inverse - compton radiation ( bremsstrahlung is less effective in making hard x - rays ) .
we caution that the natures of both the `` failed wind '' and the corona remain uncertain , and therefore there are inevitable uncertainties associated with the interaction of these two components considering the modeled dynamical nature of the outflow ( e.g. , * ? ? ?
* ) and the above coronal - quenching model , we suggest that there may even be a cyclical mechanism that switches on / off the coronal emission : after the quenching of the coronal emission , the wind can be successfully launched and there will be no `` failed wind '' falling into the corona ; the corona thus recovers to a standard emitting mode , and overionizes the inner portion of wind which will then again fall into the corona and suppress its emission .
the outflow is expected to settle down to a steady state over a timescale of years ( e.g. , * ? ? ?
* ) , and thus we might expect the above mechanism to operate over a timescale of years or longer .
such a cyclical mechanism could be used to explain why some bal quasars are normal ( after absorption corrections ) and some are perhaps intrinsically weak , and it may be responsible for the significant flux and spectral variability observed in some bal quasars ( e.g. , pg 2112 + 059 ; * ? ? ? * ) .
if bal winds exist in all / most quasars , such a mechanism could also be responsible for the strong variability seen from some non - bal quasars when they have entered weak states due to quenched coronal emission ( e.g. , phl 1092 ; * ? ? ?
pg 1004 + 130 and pg 1700 + 518 appear to have some intrinsic absorption regardless of whether they are intrinsically weak . for pg 1004 + 130 ,
the column density derived from spectral modeling is @xmath86 @xmath3 ( see section [ sec - spec ] ) . for pg 1700 + 518 ,
the _ xmm - newton _ data reveal a column density of a few @xmath148 @xmath3 @xcite ; significant absorption is also indicated by the small effective photon index ( @xmath149 ) derived using the _ nustar _ band ratio ( see section [ sec - pho ] ) . within the above coronal - quenching mechanism
, the apparent absorption could still be attributed to the shielding gas in bal quasars as there may be a period when the coronal emission is quenched while the shielding gas has not fully disappeared .
the values of @xmath20 are both @xmath150 for these two bal quasars after corrections for apparent absorption @xcite , indicating that they are @xmath151 times intrinsically weaker than expected .
intrinsic weakness would naturally explain the lack of a strong fe k@xmath5 line in pg 1004 + 130 , as the observed x - ray spectra are not scattering / reflection dominated . for the radio - loud object ,
pg 1004 + 130 , the intrinsic weakness scenario implies that the radio jet does not produce emission at a nominal level either ( @xmath1 times weaker ) .
although unusual , a small fraction of radio - loud quasars appear to be providing little / no excess emission compared to their radio - quiet counterparts ( see , e.g. , figure 7 of * ? ? ?
the likely large inclination angle of pg 1004 + 130 ( @xmath152 ; e.g. , @xcite ) could be one factor that reduces the observed x - ray emission compared to other radio - loud quasars due to a lack of relativistic beaming .
intrinsic weakness in quasars may affect the appearance of the optical uv emission lines by modifying the photoionization properties of the emission - line region .
phl 1811 , for example , has unusual line emission : its high - ionization lines ( e.g. , @xmath98 and @xmath153 ) are very weak , there is no evidence for forbidden or semiforbidden lines ( e.g. , [ ] @xmath154 ) , and the and pseudo - continuum in the near - uv is very strong @xcite .
the phl 1811 analogs investigated by @xcite have similar emission - line properties and are also weak quasar candidates .
such unusual emission lines can probably be attributed to a soft ionizing continuum ( uv nominal but weak ; e.g. , @xcite ) . for pg 1004 + 130 and pg 1700 + 518 , the @xmath98 and @xmath153 lines
are strongly absorbed due to their bal nature ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , rendering reliable line - strength measurements impossible .
pg 1004 + 130 shows normal [ ] @xmath154 emission ; pg 1700 + 518 is a narrow - line type 1 quasar ( as for phl 1811 ) and shows no visible [ ] @xmath154 line ( e.g. , * ? ? ?
their optical emission - line properties in general do not appear unusual compared to other pg quasars when considering `` eigenvector 1 '' correlations ( e.g. , * ? ? ?
* ; * ? ? ?
it is difficult to determine if pg 1004 + 130 and pg 1700 + 518 share the same usual emission - line properties as phl 1811 when the weakness of their high - ionization lines can not be constrained .
however , even if they have generally normal line emission , the intrinsic weakness scenario may still work , based on the following consideration . for a typical quasar ,
the ionizing continuum is dominated by euv emission , which is largely unobservable and is often estimated via interpolating the uv and photometric data points assuming a power - law spectrum ( e.g. , see the composite quasar seds in figure [ fig - sed ] ; also adopted in @xcite ) .
the coronal - quenching mechanism discussed above could perhaps mainly reduce the @xmath155 kev emission while leaving the euv radiation from the accretion disk largely the same .
although the significantly weakened emission would certainly affect the ionization state of the emission - line region somewhat , it is possible that the appearance of the emission lines is affected less dramatically due to ionization from euv photons ; further photoionization calculations of emission lines are required to assess this in detail . motivated by our _ nustar _ results , we derived basic statistical constraints upon the fraction of intrinsically weak bal quasars from the relatively well - studied large bright quasar survey ( lbqs ) bal quasar sample in @xcite . these are distant bal quasars at @xmath343 ; at their mean redshift of @xmath156 , the 0.58 kev _ chandra _ observations are probing the rest - frame @xmath13724 kev band , close to the hard band observed by _
nustar_. these quasars are all radio quiet except for one , and
thus we do not expect significant jet - linked contributions to the x - ray emission as for the case of pg 1004 + 130 .
we searched for intrinsically x - ray weak quasar candidates in these sample sources by comparing their @xmath18 and @xmath22 ( the absorption - corrected @xmath18 ; derived by assuming a @xmath100 power - law spectrum and normalizing it to the rest - frame continuum ) values from @xcite values instead of using upper limits . ] to the expected @xmath18 values derived from the @xcite relation .
these quasars have a limited range of @xmath99 ( with standard deviation @xmath157 dex ) , and thus we adopted the mean expected @xmath18 value ( @xmath158 ) and its 90% confidence - level uncertainty ( @xmath159 ; from figure 5 of @xcite ) in the comparison .
we also performed the comparison by deriving @xmath20 and @xmath23 for each object individually , and we got consistent results with those below . figure [ fig - aox06]a shows the @xmath18 vs. @xmath22 plot for the 35 bal quasars in the sample .
the shaded regions indicate the expected mean @xmath18 for this sample and its 90% uncertainty .
sources that lie near the slanted dashed line ( @xmath100 soft spectra , while heavily obscured or compton - thick spectra are usually hard ) and outside the shaded regions ( being x - ray weak ) are candidates for being intrinsically weak quasars . a significant fraction of the sources lie above the @xmath160 line , indicating that x - ray absorption is likely present .
only six objects have @xmath18 values in the expected range for typical quasars ( dark - shaded region ) .
however , after absorption corrections , 15 additional objects have @xmath22 values within the expected range ( light - shaded region ) , indicating that they probably have nominal underlying x - ray emission . there are four objects below the light - shaded region ( two measurements and two upper limits ) , and another ten in the light - shaded region have upper limits on @xmath22 .
we consider these 14 objects as possible candidates for being intrinsically weak quasars , and thus the fraction is @xmath161 ( i.e. , @xmath162 ) among bal quasars .
such a constraint is relatively loose as 12 of the 14 candidates have only upper limits on @xmath22 .
we performed a _
chandra _ stacking analysis to obtain the average x - ray properties of these 12 sources ( cf .
section 4.2 of * ? ? ?
we added the total counts and background counts of the sources extracted from @xmath163 eef apertures and larger annular background regions , respectively , in the observed 0.52 kev ( soft ) and 28 kev ( hard ) bands . for each band
, we computed the binomial no - source probability ( see equation 1 ) to determine if the stacked source is detected . using the portable , interactive , multi - mission simulator ( pimms )
, we derived the effective photon index from the ratio between the soft - band and hard - band counts .
we adopted the average values of redshift , galactic absorption column density , and @xmath103 ( there was only a small spread in @xmath103 , with standard deviation @xmath164 dex ) to compute the @xmath18 and @xmath22 parameters ( @xmath22 was derived using the hard - band flux and assuming @xmath100 ) for this subsample . the stacking results are listed in table 3 , and the stacked data point is shown as the black star in figure [ fig - aox06]b . the stacked source is detected in both the soft and hard bands , with net counts of @xmath165 and @xmath166 after aperture corrections , respectively , and it has @xmath167 . given the high luminosity ( @xmath168 erg s@xmath51 ) in the soft band , the observed soft x - ray spectrum ( covering rest - frame @xmath1376 kev ) should have negligible contribution from any host - galaxy emission .
it appears that the stacked source is a candidate for being intrinsically x - ray weak , as it is close to the slanted dashed line in the unshaded region ( having a fairly soft x - ray spectrum and being x - ray weak ) . as a first - order estimate
, we consider that about half ( six ) of the 12 quasars being stacked are probably softer ( @xmath169 ) than the stacked signal , and the other half are harder ( @xmath170 ) .
all these objects are outside the dark - shaded region ( x - ray weak based on @xmath18 ) , while the six having a soft spectrum ( @xmath169 ) would lie close to the @xmath18=@xmath22 line if they could be detected individually .
therefore , we expect that at least six ( @xmath171 ) of the 35 bal quasars are candidates for being intrinsically weak . combined with the @xmath172 upper limit , we constrain the fraction to be @xmath17340% .
deeper _ chandra _ observations of these 12 objects that detect them individually could significantly narrow the estimated range of this fraction . among the 12 stacked objects ,
there are seven hibal and five lobal quasars ( see footnote [ footnote - bal ] ) . as lobal quasars are x - ray weaker than hibal quasars in general , we stacked these two groups separately , and the results are also shown in table 3 and figure [ fig - aox06]b . the stacked signal of the 12 objects is dominated by the seven hibal quasars , but both of these groups of seven hibal quasars and five lobal quasars could contain intrinsically x - ray weak quasars , given their stacked x - ray properties and the positions of the stacked sources in the @xmath18 vs. @xmath22 plot .
the fraction of intrinsically x - ray weak objects among bal quasars estimated here ( @xmath17340% ) is considerably larger than the @xmath147 fraction of intrinsically weak objects among non - bal quasars @xcite .
if non - bal quasars also have bal winds lying out of the line of sight ( see section 1.1 ) , the significant difference in the fractions would suggest that intrinsically weak quasars are preferentially seen as bal quasars .
one possible scenario is that although the average covering factor of the bal wind is @xmath8 ( see section 1.1 ) , the wind in an intrinsically weak quasar has a considerably larger covering factor , as it is likely easier to launch the wind when the nuclear x - ray emission is weak .
therefore , we would tend to observe bals preferentially in the spectra of intrinsically weak quasars .
compared to the sdss bal quasar sample , the lbqs bal quasars have somewhat higher optical luminosities ( by @xmath174 dex on average ) and stronger bal features , and they are preferentially x - ray weaker ( e.g. , see section 4.6.2 of @xcite ) . the weaker x - ray emission and stronger bal features of lbqs bal quasars could be related to more absorption / coronal - quenching and thus stronger radiatively driven winds . as the nature of intrinsic x - ray weakness is highly uncertain
, we caution that the constraints on the fraction ( @xmath17340% ) derived from the lbqs sample might not be applicable to the sdss or other bal quasar samples , and further x - ray studies of large bal quasar samples are required to constrain the fraction better .
in this paper , we have investigated the hard emission observed by _ nustar _ from two optically bright bal quasars ,
pg 1004 + 130 and pg 1700 + 518 , and we have discussed the nature of their weak hard emission .
the key points from this work are summarized below : 1 .
pg 1004 + 130 and pg 1700 + 518 have been observed by _ nustar _ with exposure times of @xmath57 ks and @xmath58
ks , respectively .
pg 1004 + 130 was detected in both fpms a and b , while pg 1700 + 518 was only detected in fpm a. the positions of both objects are in good agreement with their optical positions ( @xmath175 ) , indicating the detections are reliable .
see section [ sec - pho ] .
2 . we provided aperture - photometry results for the two targets in five _ nustar _ bands : , , , , and .
the luminosities in the 420 kev band are @xmath176 erg s@xmath51 for pg 1004 + 130 and @xmath177 erg s@xmath51 for pg 1700 + 518 .
we also derived an effective power - law photon index based on the _ nustar _ band ratio .
pg 1004 + 130 is soft with @xmath178 , and pg 1700 + 518 is hard with @xmath179 .
see section [ sec - pho ] .
we jointly analyzed the _
nustar _ and _ chandra _ spectra for pg 1004 + 130 .
the spectra were modeled with a partial - covering absorber model .
the resulting absorption is moderate , @xmath86 @xmath3 , and it is insufficient to explain the weakness of this quasar .
furthermore , it is likely that the 38 kev flux has decreased by a factor of @xmath91 in the 2012 _ nustar _
observation compared to its _ chandra _ flux in 2005 .
there is no fe k@xmath5 emission line detected at rest - frame 6.4 kev , with an upper limit on the rest - frame ew of @xmath94 ev .
see section [ sec - spec ] .
4 . we constructed radio - to - x - ray seds for pg 1004 + 130 and pg 1700 + 518 . from the radio to the uv , the continuum seds of the two targets agree with composite quasar seds ( aside from dust reddening in pg 1700 + 518 ) . besides the significant weakness around rest - frame 2 kev
, these two objects also appear more than an order of magnitude less luminous than typical quasars at rest - frame 20 kev , and even the 40 kev luminosity upper limits are below expectations .
this hard weakness implies that the two bal quasars either have compton - thick absorption or are intrinsically weak .
see section [ sec - sed ] .
we derived column - density constraints using the mytorus model , under the assumption that the weak hard emission is caused by absorption . for both objects ,
compton - thick absorption appears required for any typical assumption about the intrinsic @xmath18 value .
the derived column densities are for both bal quasars , corresponding to thomson optical depths of @xmath119 .
compton - thick absorption in the disk - wind model requires the shielding gas to be located within @xmath9 cm of the smbh ( assuming a standard accretion disk ) to prevent blocking of the uv radiation that drives the outflowing wind .
we discussed jet - linked dilution and a few other possibilities that could cause the lack of a strong fe k@xmath5 line in pg 1004 + 130 .
see section [ sec - ct ] .
we discussed an intrinsic weakness scenario that may relax the shielding requirement for launching the accretion - disk wind in bal quasars .
pg 1004 + 130 and pg 1700 + 518 are @xmath151 times intrinsically weaker than typical quasars under this scenario . based on the @xcite coronal - quenching model
, we suggested a cyclical mechanism that could switch on / off the coronal emission and be responsible for a mix of intrinsically normal and intrinsically weak bal quasars .
motivated by our _ nustar _
results , we estimated the fraction of intrinsically weak bal quasars to be @xmath17340% via a _
chandra _
stacking analysis with the lbqs bal quasar sample .
see section [ sec - xweak ] . irrespective of its cause , the hard weakness of these two bal quasars discovered by _
nustar_has implications for the detection and characterization of agns in deep surveys .
pg 1004 + 130 and pg 1700 + 518 are among the most optically luminous bal quasars known at low redshift ( figure [ fig - lz ] ) , and in each object substantial smbh growth is clearly occuring .
however , we expect that pg 1700 + 518 could not be detected in a 600 ks _ nustar _ survey800 ks . ] if placed at @xmath180 . in the deepest
_ chandra_survey to date , the 4 ms _ chandra _ deep field - south ( e.g. , * ? ? ?
* ) , pg 1700 + 518 would be detectable if placed at high redshifts ( e.g. , @xmath181 ) , as _ chandra _ is actually probing the hard bands at these redshifts .
however , with @xmath182 counts detected , the _ chandra _ data could not constrain its intrinsic spectrum accurately , and the derived properties would have a large uncertainty ( e.g. , the derived luminosity and the amount of smbh growth would be an order of magnitude lower than the real values ) . therefore
, detection completeness and agn characterization in deep surveys should be interpreted carefully , considering the likely existence of x - ray weak bal and related quasars in the local and distant universe ( e.g. , * ? ? ?
* ; * ? ? ?
in this case , a uv - excess selection of bright quasars that are x - ray weak could be utilized to search for and study the distant counterparts of pg 1004 + 130 and pg 1700 + 518 .
given the limited data for the two targets studied here , we can not strongly prefer the compton - thick absorption scenario or the intrinsic weakness scenario .
a _ nustar _
survey of a large sample of bal quasars showing similar weakness ( i.e. , being significantly weak even after corrections for apparent absorption ) may help to discriminate between these two scenarios .
if heavy absorption is responsible for the weakness , a continuous distribution of column densities would be expected , probably ranging from @xmath78 @xmath3 to being significantly compton - thick ( @xmath183 @xmath3 ) .
if all the observed survey targets show weak hard emission like pg 1004 + 130 and pg 1700 + 518 , this would indicate an unlikely scenario where we had found an isolated population of highly compton - thick objects via soft selection , and thus we would consider that intrinsic weakness is probably the correct resolution .
alternatively , obtaining a hard spectrum for a similar object with sufficient counts that allows more detailed spectral analysis may also provide some useful constraints that would shed light on the nature of these bal quasars .
we hypothesized a cyclical mechanism in section 4.2.1 that could explain why some bal quasars are intrinsically normal and others are perhaps intrinsically weak .
further numerical simulations that carefully treat the temporal evolution of the shielding gas will be required to assess this dynamical model .
an variability study of a large sample of bal quasars over a long timescale ( years ) may also help to assess whether bal quasars vary between normal and x - ray weak states ( see , e.g. , * ? ? ?
this could be achieved by snapshot monitoring observations of a sample of weak bal quasars using _ chandra _ , _ xmm - newton _ , and/or _ nustar _ which would provide multi - epoch and large - bandpass coverage .
we estimated the fraction of intrinsically x - ray weak bal quasars ( @xmath17340% ) in section 4.2.3 .
this constraint could likely be tightened significantly if we could obtain _ chandra _ hard - band ( 28 kev ; @xmath18424 kev in the rest frame ) flux measurements of the 12 sources with only relatively weak upper - limit information presently .
these objects likely contain strong candidates for intrinsically x - ray weak quasars based on the stacking analysis .
they were previously observed by _
chandra _ with 57 ks exposures
@xcite . it could be a useful investment to obtain additional @xmath3030 ks _ chandra _ observations that will improve the hard - band detection limit by a factor of @xmath185 and detect a significant fraction of these 12 sources individually .
+ + we acknowledge support from the california institute of technology ( caltech ) _ nustar _
subcontract 44a-1092750 ( bl , wnb ) , nasa adp grant nnx10ac99 g
( bl , wnb ) , the leverhulme trust and the science technology and facilities council ( dma ) , basal - cata grant pfb-06/2007 and conicyt - chile grants fondecyt 1101024 and anillo act1101 ( feb ) , and conicyt - chile grant fondecyt 3120198 ( cs ) .
we thank m. young for help with the planning of this project and k. forster for help with the _ nustar _
data access , and we thank m. balokovic , k. boydstun , t. n. lu , b. p. miller , jianfeng wu , and t. yaqoob for helpful discussions .
we thank the referee , s. c. gallagher , for carefully reviewing the manuscript and providing helpful comments .
this work was supported under nasa contract no .
nng08fd60c , and made use of data from the _ nustar _
mission , a project led by caltech , managed by the jet propulsion laboratory , and funded by the national aeronautics and space administration .
we thank the _ nustar _ operations , software and calibration teams for support with the execution and analysis of these observations .
this research has made use of nustardas jointly developed by the asi science data center ( asdc , italy ) and caltech ( usa ) .
pg 1004 + 130 a & @xmath186&@xmath187&@xmath188&@xmath189&@xmath190&@xmath191&@xmath192&@xmath193&@xmath194&@xmath195 & @xmath196 + pg 1004 + 130 b & @xmath197&@xmath198&@xmath199&@xmath200&@xmath201&@xmath202&@xmath203&@xmath204&@xmath205&@xmath206 & @xmath196 + pg 1700 + 518 a & @xmath207&@xmath208&@xmath209&@xmath210&@xmath211&@xmath212&@xmath213&@xmath214&@xmath215&@xmath216&@xmath217 + pg 1700 + 518 b & @xmath218&@xmath219&@xmath220&@xmath221&@xmath222&@xmath223&@xmath224&@xmath225&@xmath226&@xmath227&@xmath228 [ tbl - pho ] cccccccccc all & 12&1.99 & 68.4 & @xmath165 & @xmath166 & @xmath229 & 43.3 & @xmath230 & @xmath231 + hibal quasars & 7&2.10 & 40.2 & @xmath232 & @xmath233 & @xmath229 & 43.5 & @xmath234 & @xmath235 + lobal quasars & 5&1.84 & 28.2 & @xmath236 & @xmath237 & @xmath238 & 42.9 & @xmath239 & @xmath240 [ tbl - stack ] | we present _ nustar _ hard observations of two weak broad absorption line ( bal ) quasars , pg 1004 + 130 ( radio loud ) and pg 1700 + 518 ( radio quiet ) .
many bal quasars appear x - ray weak , probably due to absorption by the shielding gas between the nucleus and the accretion - disk wind .
the two targets are among the optically brightest bal quasars , yet they are known to be significantly x - ray weak at rest - frame 210 kev ( times fainter than typical quasars ) . we would expect to obtain @xmath0600 hard x - ray ( @xmath1 kev ) photons with _ nustar _ , provided that these photons are not significantly absorbed ( @xmath2 @xmath3 ) . however ,
both bal quasars are only detected in the softer _ nustar _
bands ( e.g. , 420 kev ) but not in its harder bands ( e.g. , 2030 kev ) , suggesting that either the shielding gas is highly compton - thick or the two targets are intrinsically x - ray weak .
we constrain the column densities for both to be @xmath4 @xmath3 if the weak hard emission is caused by obscuration from the shielding gas .
we discuss a few possibilities for how pg 1004 + 130 could have compton - thick shielding gas without strong fe k@xmath5 line emission ; dilution from jet - linked x - ray emission is one likely explanation .
we also discuss the intrinsic weakness scenario based on a coronal - quenching model relevant to the shielding gas and disk wind of bal quasars .
motivated by our _ nustar _
results , we perform a _ chandra _
stacking analysis with the large bright quasar survey bal quasar sample and place statistical constraints upon the fraction of intrinsically weak bal quasars ; this fraction is likely @xmath6 . |
a branching random walk is a collection of points which , starting from a single point , diffuse and branch independently of the time , of their positions or of the other points , as in figure [ fig : brw ] . branching random walks appear in many contexts ranging from mathematics @xcite to biology @xcite .
they can for example be used to describe how a growing population invades a new environment . in the one dimensional case ,
see figure [ fig : brw ] , there is , at a given time @xmath0 , a rightmost individual at position @xmath1 , a second rightmost at @xmath2 and so on .
( note that the rightmost @xmath3 at a time @xmath4 is not necessarily a descendant of the rightmost @xmath1 at time @xmath0 . )
the expected position @xmath5 of the rightmost individual as well as the probability distribution of its position @xmath1 around @xmath6 are well understood @xcite ; the goal of the present paper is to describe the statistical properties of the positions of all the rightmost points in the system , in particular the distribution of the distances between the two rightmost points , the average density of points at some fixed distance from the rightmost @xmath1 , etc .
one motivation for studying these distances is that the problem belongs to the broader context of extreme value statistics @xcite : trying to understand the statistical properties of the rightmost points in a random set of points on the line is a problem common to the studies of the largest eigenvalues of random matrices @xcite , of the extrema of random signals @xcite , or of the low lying states of some disordered systems such as spin glasses @xcite . in fact , the points generated after some time @xmath0 by a branching random walk can be viewed as the energies of the configurations of a directed polymer in a random medium @xcite , and the distances between the rightmost points as the gaps between the low lying energy states .
the most studied example of branching random walk is the branching brownian motion : one starts with a single point at the origin which performs a brownian motion and branches at a given fixed rate ( right part of figure [ fig : brw ] ) .
whenever a branching event occurs , the point is replaced by two new points which evolve themselves as two independent branching brownian motions .
while the number of points generated after some time @xmath0 grows exponentially with time , the expected position @xmath6 of the rightmost point increases only linearly with time @xcite . in one dimension ,
mc kean @xcite and bramson @xcite have shown that the probability distribution of the rightmost point is given by the traveling wave solution of the fisher - kpp equation , with a step initial condition . here
we will see that all the statistical properties of the rightmost points can be understood in terms of solutions to the fisher - kpp equation with appropriate initial conditions @xcite .
we will also show that the distribution of the distances between these rightmost points has a long time limit which exhibits the striking property of superposability : the distances between the rightmost points of the union of two realizations of the branching brownian motion have the same statistics as those of a single realization .
this paper is organized as follows : in section [ statistics ] we introduce some generating functions useful to study random sets of points on the line and show how one can use them to obtain all the properties of these random sets . in section [ sec : bbmfkpp ] we show that , for the branching brownian motion , all these generating functions are solutions of the fisher - kpp equation .
we also show that the distribution of all the rightmost points as seen from @xmath6 or , alternatively , as seen from @xmath1 , has a long time limit which can be computed as the delay of fisher - kpp traveling waves .
this distribution has the property of superposability . in section [ quantitative ] ,
we present results , mostly numerical , on some specific aspects of the limiting distribution of points in the branching brownian motion , namely the distribution of the distance between the two rightmost points and the average density seen from the rightmost point . in section [ sec : disc ] we explain how the results on the branching brownian motion can be extended to more general branching random walks .
finally , we study in section [ statmeas ] the distribution of all the rightmost points in a specific frame which depends on the realization and which was introduced by lalley and sellke @xcite .
in this section , we introduce some useful quantities ( generating functions ) to characterize random sets of points on the line such that the number @xmath7 defined as @xmath8 is finite and vanishes for @xmath9 large enough . the first generating function one can define is @xmath10 from the knowledge of this function , one can extract the probability distribution function @xmath11 of the position @xmath9 of the @xmath12-th rightmost point .
indeed , by definition of @xmath13 , @xmath14 where @xmath15 is the probability that there are exactly @xmath12 points on the right of @xmath9 . one can notice that @xmath16 is the probability to have less than @xmath12 points on the right of @xmath9 .
the generating function of these sums is , from , @xmath17\lambda^2+\big[q_0(x)+q_1(x)+q_2(x)\big]\lambda^3+\cdots . \label{sumterms}\ ] ] but @xmath16 is also the probability that the @xmath12-th rightmost point , if it exists , is on the left of @xmath9 .
therefore , @xmath18 where @xmath19 is the probability that the @xmath12-th rightmost point exists and is in the interval @xmath20 $ ] .
( note that @xmath21 is the probability that there are at least @xmath12 points on the line . )
the knowledge of @xmath22 gives in particular the average distances between the points : from ( [ pix ] ) , one can see that @xmath23,\notag\\ & = \langle x_1 \rangle-\lambda[\langle x_1 \rangle-\langle x_2 \rangle ] -\lambda^2[\langle x_2 \rangle-\langle x_3 \rangle]-\cdots,\end{aligned}\ ] ] where @xmath24 is the average position of the @xmath12-th point ( with the convention that @xmath25 if there are less than @xmath12 points in the system ) . therefore @xmath26 = \sum_{i\ge1 } \langle d_{i , i+1}\rangle \lambda^i , \label{psid}\ ] ] where @xmath27 is the average distance between the @xmath12-th and the @xmath28-th point . to obtain the correlations between the positions of pairs of points ,
one can start , for @xmath29 , from the generating function @xmath30 the coefficient in front of @xmath31 in the expansion of @xmath32 in powers of @xmath33 and @xmath34 is the probability that there are exactly @xmath12 points on the right of @xmath9 and @xmath35 points on the right of @xmath36 . as in , the coefficient of @xmath31 in the expansion of @xmath37 is the probability that there are less than @xmath12 points on the right of @xmath9 and less than @xmath35 points on the right of @xmath36 , which is also the probability that the @xmath12-th rightmost point ( if it exists ) is on the left of @xmath9 and the @xmath35-th rightmost point ( if it exists ) is on the left of @xmath36 .
thus , for @xmath38 , @xmath39 where @xmath40 is the probability that both the @xmath12-th and @xmath35-th rightmost points exist and lie respectively in the intervals @xmath20 $ ] and @xmath41 $ ] . one can generalize ( [ defpsi],[defpsi2 ] ) by defining , for @xmath42 , the generating functions @xmath43 of the numbers @xmath44 of points on the right of positions @xmath45 , and get as in ( [ pix],[twopoint ] ) all the higher correlation functions . in that way ,
all the statistical properties of the measure can be derived from the knowledge of the generating functions ( [ defpsi3 ] ) . in the following
we will often try to characterize the random set of points as seen from the rightmost point ( i.e. in the frame where the rightmost point is at the origin ) . to do so , let us define the generating functions of the numbers @xmath46 of points at the right of @xmath47 in the frame of the rightmost point .
( note that if @xmath48 is the position of the rightmost , then @xmath49 and one has @xmath50 for @xmath51 and @xmath52 for @xmath53 . )
@xmath54 ( as in , we assume @xmath55 . ) these generating functions , as in section [ generating - functions ] , allow one to calculate all the statistical properties of the measure in the frame of the rightmost point ( in particular the distribution of the relative distances between the points ) .
they can be determined from the knowledge of the generating functions @xmath56 defined in ( [ defpsi3 ] ) by @xmath57 in section [ quantitative ] we will calculate the density of probability @xmath58 that the two rightmost points are separated by a distance @xmath59 ( and that there are at least two points on the line ) and the average density @xmath60 at a distance @xmath59 from the rightmost point . from ( [ defchi3 ] ) one can see that @xmath61 then using that @xmath62 , one gets @xmath63 by a similar calculation one can show that the average density @xmath60 of points at distance @xmath59 from the rightmost point is @xmath64 we now describe a few examples of such measures .
our first example is a poisson process on the line with a density @xmath65 .
we assume that @xmath65 decays fast enough to the right so that a rightmost point exists , and that @xmath66 so that there are infinitely many points on the line .
by definition of a poisson process , each infinitesimal interval @xmath20 $ ] is occupied by a point with probability @xmath67 and empty with probability @xmath68 , and the occupation numbers of disjoint intervals are uncorrelated .
the probability @xmath15 that there are exactly @xmath12 points on the right of @xmath9 is given by @xmath69 from this , we obtain @xmath22 from ( [ defpsi],[exppsi ] ) and @xmath70 from ( [ defpsi2 ] ) in the poisson process : @xmath71 using , the generating function of the average @xmath72 between the @xmath12-th and @xmath28-th points is @xmath73.\ ] ] the probability distribution function @xmath58 that the distance @xmath74 is equal to @xmath59 and the average density @xmath60 seen at a distance @xmath59 from the rightmost point are given by @xmath75 these expressions can be understood directly from the definition of the poisson process or , with a little more algebra , from ( [ p12],[rho ] ) . one can notice that @xmath58 and @xmath60 are given by the same expression with @xmath59 replaced by @xmath76 and are therefore analytic continuations of each other whenever @xmath65 is analytic . in the special case where the density of the poisson process is an exponential @xmath78 , one can simply replace @xmath79 in the previous expressions by @xmath80 .
this gives @xmath81 = \exp \left [ -e^{-\alpha \big(x - \frac{\ln(1- \lambda)}\alpha\big)}\right ] , \\
\psi_{\lambda\mu}(x , y ) & = \exp \left [ - \mu ( 1-\lambda ) \frac{e^{-\alpha x}}\alpha - ( 1-\mu ) \frac{e^{-\alpha y}}\alpha \right ] , \end{aligned } \label{a1}\ ] ] so that from ( [ psid ] ) @xmath82 and thus @xcite @xmath83 one also has from @xmath84 start with a collection of points @xmath85 , distributed according to some measure @xmath86 and , independently for each point @xmath85 , replace it by a realization of another measure @xmath87 shifted by @xmath85 .
we say that the points @xmath85 are _ decorated _ by the measure @xmath87 and call the resulting measure as @xmath86 decorated by @xmath87 .
we assume that @xmath86 and @xmath87 are such that the decorated measure has a rightmost point .
if the functions @xmath22 , @xmath70 , for the measure @xmath87 are known , the decorated measure is characterized by functions @xmath88 , @xmath89 , given by @xmath90 where the average is over all realizations @xmath91 of the measure @xmath86 .
for instance ,
if @xmath86 is a poisson process of density @xmath92 , then @xmath93 = \exp \left [ \int \big [ \psi_{\lambda}(x - u ) -1\big ] r(u ) \ ,
\diffd u \right ] , \\
\psi_{\lambda\mu}(x , y)= \exp \left [ \int \big [ \psi_{\lambda\mu}(x - u , y - u ) -1\big ] r(u ) \ , \diffd u \right ] .
\end{gathered } \label{psi}\ ] ] for a decorated measure where the decoration @xmath87 is a poisson process of density @xmath94 , the average over the @xmath85 s in leads in general to complicated expressions for @xmath88 or @xmath95 .
the expressions for @xmath58 and @xmath60 are however the same as in for the pure poisson process of density @xmath94 .
in fact , all the statistical properties of the distances between the rightmost points are the same as those in the exponential poisson process .
this can be understood from the following reason : decorating the points @xmath96 by independent realizations of a poisson process of density @xmath94 is equivalent to drawing a single realization of a poisson process of density @xmath97 , which is just the same as one realization of a poisson process of density @xmath94 shifted by the random variable @xmath98 .
the same argument applies to ruelle cascades , which can be defined as follows @xcite : take an increasing sequence of positive numbers @xmath99 and start with a poisson process of density @xmath100 . at each step
@xmath101 , each point in the system is decorated by a poisson process of density @xmath102 . at step @xmath103 , the measure of points in the system is simply , from the previous argument , a poisson process of density @xmath102 globally shifted by a random variable which depends on the positions of the points at step @xmath104 .
therefore , the statistics of the distances of the rightmost points is the same as for the poisson process of density @xmath102 .
we are now going to see how the generating functions ( [ defpsi],[defpsi2],[defpsi3 ] ) can be determined when the random set of points on the line are the points generated at time @xmath0 by a branching brownian motion . to define the branching brownian motion
we start at time @xmath105 with a single point at the origin .
this point diffuses and branches , and its offspring do the same . after some time @xmath0 ,
a realization of the process consists of a finite number of points located at positions @xmath106 for @xmath107 then , during the next time interval @xmath108 , each point , independently of what the others do , moves a random distance @xmath109 with @xmath110 and @xmath111 , and , with probability @xmath112 , is replaced by two new points located at the same position @xmath106 . for any function @xmath113 one can define the generating function @xmath114 by @xmath115\right\rangle , \label{dual}\ ] ] where the @xmath106 for @xmath116 are the positions of the @xmath117 points of the branching brownian motion at time
@xmath0 and @xmath118 denotes an average over all the possible realizations . by analyzing what happens during the very first time interval @xmath112
, one can see that the evolution of @xmath114 satisfies @xmath119 the first term in the right hand side represents the motion of the initial point during the first time interval @xmath112 and the second term represents the branching event which occurs with probability @xmath112 during this first time interval .
taking @xmath112 to zero , one gets @xmath120 which is the fisher - kpp equation @xcite .
( the fisher - kpp equation is often written as @xmath121 , but this is identical to by the change of variable @xmath122 . ) because there is a single point at the origin at time @xmath105 , the initial condition is simply , from , @xmath123 the generating function ( [ defpsi3 ] ) at time @xmath0 @xmath124 can be written , for @xmath125 , as @xmath126\right\rangle = h_\phi(x , t ) , \label{psi - h}\ ] ] where the function @xmath127 is given by @xmath128 and where @xmath129 is the heaviside step function defined by @xmath130 see figure [ figphi ] for the general shape of . . ] with the choice ( [ choicef ] ) of @xmath113 , the generating function ( [ defpsi3 ] ) and , therefore , all the properties of the point measure in the branching brownian motion at time @xmath0 can be obtained as solutions of the fisher - kpp equation with the initial condition ( [ initcond ] ) . in the special case @xmath131 and @xmath132 of , i.e. for the initial condition @xmath133 , one gets @xmath134 and one recovers the well - known fact @xcite that the solution @xmath135 of the fisher - kpp equation with a step initial condition is the cumulative distribution function of the position of the rightmost point . in section
[ quantitative ] we will choose @xmath136 and @xmath137 , other special cases of , given by @xmath138 to calculate the generating functions ( [ defpsi],[defpsi2 ] ) at time @xmath0 @xmath139 needed to determine the distribution @xmath58 and the density @xmath60 defined at the end of section [ statistics ] .
the fisher - kpp equation has two homogeneous solutions : @xmath140 , which is unstable , and @xmath141 , which is stable .
when the initial condition @xmath127 is given by the step function @xmath129 , see , the solution @xmath135 of ( [ fkpph ] ) becomes a traveling wave with the phase @xmath142 invading the phase @xmath143 @xcite . as the front is an extended object
, one can define its position @xmath6 in several ways ; for example one could define @xmath6 as the solution of @xmath144 for some @xmath145 . here
it will be convenient to use the following definition @xmath146 one can see using ( [ hprob ] ) that @xmath6 defined by ( [ defmt ] ) is the average position of the rightmost point .
if the initial condition ( [ initcond ] ) is not a step function but is such that @xmath147 for all large enough @xmath9 and @xmath127 is a constant smaller than 1 for all large negative @xmath9 , as in ( [ choicef],[choicefsimple ] ) , the solution @xmath114 of ( [ fkpph ] ) becomes also a traveling wave .
its position @xmath148 can be defined as in ( [ defmt ] ) by @xmath149 we are now going to show that the whole measure seen from the rightmost point can be written in terms of this position @xmath150 : one can rewrite ( [ the - frame ] ) as @xmath151 then from and ( [ defmtbis ] ) one gets @xmath152 where @xmath113 is the function with @xmath132 .
therefore , _ with the definition ( [ defmtbis ] ) of the position of the front _ , the whole information about the measure in the frame of the rightmost point , at any time @xmath0 , can be extracted from the @xmath113 dependence of @xmath153 in the long time limit , it is known @xcite that the traveling wave solution @xmath135 of ( [ fkpph ] ) , with the initial condition ( [ teta ] ) , takes an asymptotic shape , @xmath154 .
this means that @xmath155 { } f(x ) , \label{hf2}\ ] ] where @xmath154 satisfies @xmath156 it is also known , since the work of bramson @xcite , that , in the long time limit , the traveling wave moves at a velocity 2 and that its position ( [ defmt ] ) is given by @xmath157 if the function @xmath127 is not the step function but is of the form ( [ choicef],[choicefsimple ] ) , the solution @xmath114 of ( [ fkpph ] ) becomes also a traveling wave with the same shape @xmath154 .
this wave is centered around the position @xmath148 , defined in ( [ defmtbis ] ) , and one has @xmath158 { } f(x ) .
\label{hmphif}\ ] ] for large times @xmath148 is still given by ( [ bramson ] ) , but with a different constant @xcite .
this means that @xmath159 { } f[\phi ] , \label{defdelay}\ ] ] where @xmath160 $ ] is the long time delay in the position of the front due to the modified initial condition , as compared to a front starting with a step function .
taken together , and give @xmath161 { } f\big(x+f[\phi]\big ) .
\label{delay}\ ] ] using , this becomes @xmath162 { } f\big(x+f[\phi]\big ) , \label{stat1}\ ] ] which shows that the measure of @xmath163 ( the rightmost points in the branching brownian motion seen from the @xmath6 frame ) does converge when @xmath164 to a limiting point measure characterized by the functions @xmath154 and @xmath160 $ ] .
the measure of @xmath165 ( the rightmost points in the branching brownian motion seen from the @xmath1 frame ) also has a well - defined limit when @xmath164 . indeed , using ( [ the - frame-2 ] ) and
, one gets @xmath166 { } 1 - \left ( \partial_{z_1 } + \cdots + \partial_{z_{k } } \right ) f[\phi ] .
\label{the - frame-3}\ ] ] therefore , in the long time limit , all the information on the distribution of the rightmost points seen from @xmath1 is contained in the @xmath113 dependence of the delay @xmath160 $ ] .
note that , in contrast to ( [ the - frame-2 ] ) which requires the position to be defined by ( [ defmtbis ] ) , the delay @xmath160 $ ] in or ( [ the - frame-3 ] ) depends only on @xmath113 : it would not change if we had chosen another definition of the front position . let us now consider @xmath167 independent branching brownian motions starting at @xmath105 at positions @xmath168 . following the same argument as in section [ sec : fkpp ] , the generating function ( [ defpsi3 ] ) of the union of the points at time @xmath0 of these @xmath167 branching brownian motions is given by the following generalization of ( [ psi - h ] ) @xmath169 where @xmath114 is the same solution of ( [ fkpph ] ) with the initial condition ( [ choicef ] ) as in the case of a single branching brownian motion starting at the origin .
in the long time limit , using , @xmath170 { } \prod_{\alpha=1}^m f(x+ f[\phi]-u_\alpha ) .
\label{stat2}\ ] ] this means that here again , there is a limiting measure when @xmath164 for the rightmost points in the @xmath6 frame .
this measure is not the same as before ( when one starts with a single point at the origin ) , as can be seen by comparing and . in particular
, the distribution of the rightmost point is different . in the frame of the rightmost point
, however , one can see using and ( [ the - frame-2-bis ] ) that @xmath171 { } 1 - \left ( \partial_{z_1 } + \cdots + \partial_{z_{k } } \right ) f[\phi],\ ] ] as in .
it is remarkable that the generating function @xmath172 does depend neither on the number @xmath167 of starting points nor on their positions @xmath173 .
the picture which emerges is that if we superpose the rightmost points of several branching brownian motions , starting at arbitrary positions , the limiting measure in the frame of the rightmost point is , when @xmath174 , the same as for a single branching brownian motion .
we will say that , in the long time limit , the measure of the distances between the rightmost points in a branching brownian motion becomes _ superposable _ : the union of two ( or more ) realizations of the process ( even moved by arbitrary translations @xmath173 ) leads to the same measure in the frame of the rightmost point as for a single branching brownian motion . as a remark , it is easy to check , that the poisson process with an exponential density @xmath175 , see section [ sec : pppexp ] , is an example of a superposable measure : the superposition of @xmath167 such poisson processes translated by arbitrary amounts @xmath168 is identical to a single poisson process with an exponential distribution translated by @xmath176 .
one can also check that , for the same reason , all the decorated measures of section [ sec : deco ] are superposable when @xmath86 is a poisson process with an exponential density . in section [ statmeas ]
, we will state a stronger version of the superposability property of the branching brownian motion .
in this section we obtain , by integrating numerically the equation ( [ fkpph ] ) with the appropriate initial condition , some statistical properties of the limiting measure seen from the rightmost point .
the analytic calculation of the delay @xmath160 $ ] is in general not easy . for @xmath136 given by ( [ choicefsimple ] ) , however ,
it was possible to show @xcite that when @xmath177 , the delay is given by @xmath178 \simeq - \ln(1-\lambda ) - \ln [ - \ln(1-\lambda ) ] + { \cal o}(1 ) , \label{fphi1asympt}\ ] ] and , from this , one could deduce that , in the long time limit , the average of the distance @xmath179 between the @xmath12-th and the @xmath28-th rightmost points is given for large @xmath12 by @xmath180 in @xcite , the numerical values of the distances between the rightmost points were also obtained by integrating the fisher - kpp equation with the initial condition @xmath181 in ( [ choicefsimple ] ) and by using ( [ psid ] ) ( in practice we integrated numerically the equations satisfied by the coefficients of the expansion of @xmath22 in powers of @xmath182 . it was found that @xmath183 the results ( [ asympdi],[valuesdn ] ) gave evidence that the distances between the rightmost points of the branching brownian motion were different from those of a poisson process with an exponential density ( [ diexpo ] ) . according to ( [ p12 ] ) , to obtain the distribution @xmath58 of the distance between the two rightmost points , one needs to calculate @xmath184 to first order in @xmath34 .
we first remark that at time @xmath0 for @xmath185 , @xmath186= \operatorname{proba}[n(x - a)=0]=h_\theta(x - a , t)$ ] where @xmath135 is the standard fisher - kpp front with the step initial condition ( it is also easy to see from the definition of @xmath187 . ) then writing at time @xmath0 that @xmath188 is solution of the fisher - kpp equation , and using the initial condition @xmath187 in , one gets @xmath189 then , from one gets @xmath190 figure [ fig : distrib_d ] shows our numerical result for the distribution @xmath58 of the distance between the two rightmost points in the long time limit .
more details on our numerical procedure is given in appendix [ sec : numeric ] .
of observing a distance @xmath59 between the two rightmost points in the @xmath164 limit , as a function of @xmath59 .
for @xmath59 small ( left part ) , the distribution is very close to @xmath191 . for larger values of @xmath59 , one observes a faster exponential decay of order @xmath192.,title="fig : " ] of observing a distance @xmath59 between the two rightmost points in the @xmath164 limit , as a function of @xmath59 .
for @xmath59 small ( left part ) , the distribution is very close to @xmath191 . for larger values of @xmath59 , one observes a faster exponential decay of order @xmath192.,title="fig : " ] we see that @xmath58 is very close to @xmath193 for the values of @xmath59 which have a significant probability of occurring .
this is of course consistent with an average distance ( [ valuesdn ] ) close to @xmath194 . for large @xmath59 ( events with a small probability ) , however , the exponential decay is faster .
we now present a simple argument leading to the following prediction , which is consistent with our numerical data , @xmath195 in the long time limit , the right frontier of the branching brownian motion moves at velocity @xmath196 .
let us assume that a large distance @xmath59 between the two rightmost points is produced by the following scenario : by a rare fluctuation , the rightmost point escapes and , without branching , goes significantly ahead while the rest of the points go on as usual , forming a frontier moving at velocity @xmath196 .
such an event leads to the distance @xmath59 between the two rightmost points if , during a time @xmath197 , the rightmost point moves ( by diffusion alone ) by a distance @xmath198 without branching .
the probability of such a scenario is @xmath199\times e^{-\tau}. \label{prob1}\ ] ] the first term is the probability of diffusing over a distance @xmath200 during time @xmath197 , and the second term is the probability of not branching .
the probability to observe a large distance @xmath201 is then dominated by the events with @xmath197 chosen to maximize , that is @xmath202 and this leads to ( [ decayratedist2 ] ) in good agreement with the numerical data of figure [ fig : distrib_d ] .
there is a remarkable relation between the decay rate in ( [ decayratedist2 ] ) and the shape of the traveling wave solution of ( [ deff2 ] ) . around
the _ stable _ region @xmath203 , the equation can be linearized and one has @xmath204 we emphasize that this is a linear analysis of the _ stable _ region , which is usually uninteresting ( in contrast to the _ unstable _ region which determines the velocity ) .
the solutions for @xmath205 are @xmath206 @xmath207 is the correct root as it is the only positive solution and @xmath208 has to vanish .
the other solution @xmath209 ( the wrong root ) coincides ( up to the sign ) with the decay rate of the distribution @xmath58 for the distance @xmath59 between the two rightmost points ( [ decayratedist2 ] ) .
as explained in appendix [ sec : largedeviation ] , this coincidence exists in a broad class of branching processes : each variant of the branching brownian motion is linked to a variant of the fisher - kpp equation , and the wrong root in the linear analysis of the stable region always gives the asymptotic decay rate of @xmath58 . to obtain the average density of points at a distance @xmath59 on the left of the rightmost point
, one needs , according to , to calculate @xmath184 for @xmath34 close to 1 . as in section
[ sec : distances ] , we first remark , from the definition , that @xmath210=h_\theta(x , t)$ ] is the standard fisher - kpp front with the step initial condition .
then , writing at time @xmath0 that @xmath211 \label{expnpsirtilde}\ ] ] is solution of the fisher - kpp equation , and using the initial condition @xmath187 in , one gets @xmath212 it is the same equation as for @xmath213 in , but with a different initial condition . then , from one gets @xmath214 one can notice the great similarity between the expressions for the average density @xmath60 of points at a distance @xmath59 from the rightmost ( [ eqratilde],[rarho ] ) and the probability distribution @xmath58 for the distance between the two rightmost points ( [ eqra],[rap12 ] ) : one goes from one to the other by simple changes of signs , as in the example of a poisson process .
figure [ fig : density ] presents our numerical results for @xmath60 in the long time limit .
we see that @xmath60 increases as @xmath215 note that a poisson process with such a density would lead to asymptotic distances between points given by .
the branching brownian motion is however not a poisson process as the points are correlated , at least near the tip .
of points at a distance @xmath59 of the rightmost in the long time limit grows like @xmath216 .
when the data is multiplied by @xmath217 , as shown in the figure , the linear prefactor is clearly visible . ]
all the results of sections [ sec : bbmfkpp ] and [ quantitative ] can be generalized to other branching processes on the line where points move and branch independently of the positions and of the motions of the other points . in such systems , the function
@xmath114 defined in is also solution of an equation similar to the fisher - kpp equation . here are four examples : * the points perform brownian motions and branch as before , but at each branching event there is a probability @xmath218 to branch into three points and @xmath219 to branch into two
. then @xmath114 evolves according to @xmath220 * time is discrete with steps of duration @xmath221 ; at each time step , a point at position @xmath9 branches into two points at positions @xmath222 and @xmath223 , where the @xmath224 take independent random values distributed according to some given @xmath225 .
the evolution of @xmath114 is then given by @xmath226 ^ 2 .
\label{ex3}\ ] ] in this example , the positions of the points can be thought of as the possible energies of a directed polymer on a caley tree with independent random energies @xmath227 on the edges of the tree @xcite .
* time is continuous but space is discrete with steps 1 ; during @xmath112 , each point at position @xmath9 has a probability @xmath112 of being removed and replaced by two points at position @xmath228 .
the equation satisfied by @xmath114 is @xmath229 this example is relevant to the theory of binary search trees @xcite .
* time and space are discrete with steps @xmath230 for space and @xmath221 for time ; in a given time step , a point at position @xmath9 has a probability @xmath221 of branching into two points at position @xmath9 , a probability @xmath231 of jumping to the left , @xmath231 of jumping to the right , and @xmath232 of doing nothing .
then : @xmath233 , \label{ex5}\ ] ] which is of course a discretized version of the original fisher - kpp equation .
is actually the equation we used in our numerical simulations , see appendix [ sec : numeric ] . in all cases
, these equations have @xmath140 as an unstable fixed point , and @xmath141 as a stable fixed point . for initial conditions @xmath234 of the type ( [ choicef],[teta],[choicefsimple ] ) ,
the function @xmath114 develops into a traveling wave moving at a specific velocity @xmath235 .
we recall briefly the procedure to determine the asymptotic velocity @xmath235 of the front ( which is also , through , the velocity of the rightmost point in the branching process ) .
one looks for traveling wave solutions moving at velocity @xmath236 of the form @xmath237 and solve the linearized equation around the unstable fixed point by writing @xmath238 . this leads to a relation between @xmath239 and @xmath236 , and the minimal value @xmath235 of @xmath236 reached at some @xmath240 is the velocity selected by the front @xcite .
( we only consider here cases where the function @xmath241 has a minimum . ) for our four examples @xmath242 , \\\text{c)}\quad & v= \frac{2e^\gamma-1}\gamma;\ v^*\simeq 4.311 , & \text{d)}\quad & v=\frac1{\gamma\delta}\ln\left[1 + 2\delta\frac{\cosh(\gamma s)-1}{s^2}+\delta\right ] . \end{aligned } \label{vgammas}\ ] ] once the equation for @xmath243 of a particular branching process is written , one has access to all the generating functions @xmath244 , @xmath245 , etc .
, see ( [ defpsi],[defpsi2],[defpsi3 ] ) , by choosing the appropriate initial conditions ( [ choicef],[choicefsimple ] ) for the front equation .
the whole measure in the frame of the rightmost point is then obtained from at any finite time @xmath0 .
note that to prove the existence of a long time limit to the point measure in this frame for a specific branching process , one would need a version of bramson s result for this process which is , to our knowledge , not known in the general case .
it is natural to ask which properties of the branching brownian motion can be extended to other branching processes .
if the measure for the distances between the rightmost points has a long time limit , then the arguments of section [ superpos1 ] can be easily generalized and one can show that it is superposable .
we have checked that the analytical argument @xcite leading to the asymptotic expression for the average distances @xmath246 at large times can be extended in case b ) for a large class of densities @xmath225 and yields @xmath247 we have also checked numerically on examples c ) and d ) that the density at a distance @xmath59 of the rightmost point is , as in , @xmath248 for the tail of the distribution @xmath58 of the distance @xmath59 between the two rightmost points , we discussed a scenario , at the end of section [ sec : distances ] , which can be generalized ( see appendix [ sec : largedeviation ] ) to calculate the exponential decay @xmath58 for more general branching processes .
this scenario , however , can only hold if points can move without branching , as in our examples a ) and d ) ; for instance , in example a ) , it predicts an exponential with a decay rate equal to @xmath249 . in examples
b ) and c ) , the points branch whenever they move and the tail of @xmath58 is in general not an exponential . note that special care should be taken if the points are located on a discrete lattice , as in cases c ) , d ) and possibly b ) : quantities such as @xmath58 , see ( [ p12 ] ) , become probabilities rather than densities of probability and quantities such as @xmath60 , see ( [ rho ] ) , become average numbers rather than average densities , and all the formulas in the previous sections need to be adapted : integrals become discrete sums , derivatives become finite differences , etc . if one interprets @xmath7 as the number of points _ strictly _ on the right of @xmath9 , then the generating functions @xmath13 , @xmath32 ,
are still related to @xmath243 as in ( [ psi - h],[phi1phi2 ] ) with the choices ( [ choicef],[choicefsimple ] ) for the initial condition @xmath113 .
then , for instance , one can show easily that becomes @xmath250 = \sum_{i\ge1 } \langle d_{i , i+1}\rangle \lambda^i,\end{gathered}\ ] ] where @xmath230 is the lattice spacing .
for these systems on the lattice , there are new properties that can be investigated . as an example , if @xmath251 is the number of points on the rightmost occupied site , then it is easy to check that @xmath252.\ ] ] the whole distribution of @xmath251 can then be determined by numerical integration . in the case of our example
c ) , the number @xmath251 corresponds to the number of leaves at the deepest level in a binary search tree @xcite and we found numerically that , at large times , @xmath253 , @xmath254 , @xmath255 , @xmath256 , etc .
we recall from the results of section [ sec : fkpp ] , see and , that the distribution of the rightmost point is given in the long time limit by @xmath257=f(x ) , \label{nerg}\ ] ] where @xmath6 is the average position of the rightmost point and is asymptotically given by and where @xmath258 is defined in .
the property is however non - ergodic , as shown by lalley and sellke @xcite in the sense that for a single realization of the branching brownian motion @xmath259\ne f(x ) .
\label{nonergo}\ ] ] in fact , the left hand side of is not self - averaging and depends on the realization .
this is illustrated in figure [ fig2bbm ] : for the realization on the left , there were few branching events at early times and the first points wandered to the left , leading at larger times to an asymmetric picture . for the realization on the right of figure [ fig2bbm ] , there were many branching events early .
for the right realization , the rightmost point is almost always on the right of @xmath6 while it is almost always on the left of @xmath6 in the left realization . .
the horizontal direction represents space , and time increases downwards .
the dotted gray lines are @xmath6 and @xmath260 , the average positions of the rightmost and leftmost points , as measured from equation .,title="fig : " ] . the horizontal direction represents space , and time increases downwards .
the dotted gray lines are @xmath6 and @xmath260 , the average positions of the rightmost and leftmost points , as measured from equation .,title="fig : " ] visually , these strong memory effects of the early stages of the branching brownian motion do not seem to decay with time , and it looks like the fluctuating right frontier of the system settles at some random fixed distance @xmath261 from @xmath6 : @xmath262 where @xmath261 would depend on the realization but not the time , and where @xmath263 would be a time - dependent random number centered around zero .
a natural question is whether it is possible to define @xmath261 for each realization in such a way that the distribution of @xmath263 becomes in the long time limit independent of @xmath261 and @xmath0 , the idea being that the branching brownian motion at long times seen from @xmath264 would `` look the same '' for any realization , whatever is the value of @xmath261 .
a related question was addressed by lalley and sellke @xcite in the following way : for each realization of the branching brownian motion , define @xmath265 as @xmath266e^{x_i(t)-2t}. \label{defz}\ ] ] ( the sum is over all the points @xmath106 present at time @xmath0 . ) as shown in @xcite , @xmath267 has a limit @xmath265 for almost every realization ; that limit is finite and positive .
lalley and sellke prove then a limit theorem for the frontier of the branching brownian motion which we interpret as follows : @xmath268}\big ) , \label{lsint}\ ] ] where @xmath269 is a constant related to the large @xmath9 behavior of @xmath154 , see . in words ,
if one considers only the realizations of the branching brownian motion with a given value of @xmath265 , then the large time distribution of the rightmost point is given by a gumble located around @xmath270 .
to make the link with , @xmath271 in appendix [ apxls ] , we present the precise theorem stated by lalley and sellke and we argue that should be equivalent to their result . a natural extension to lalley s and sellke s result is to write for all the points @xmath12 @xmath272 as in , and ask whether the joint distribution of @xmath263 , @xmath273 , @xmath274 , reaches a long time limit which is independent of the value of @xmath265 .
we show in appendix [ apxls ] that lalley s and sellke s result can be extended to all the generating functions @xmath243 .
our interpretation of this extension is @xmath275}}\big ) = \exp\big(- e^{-[x-\ln(az)+f[\phi]]}\big ) , \label{ls2int}\ ] ] where the delay function @xmath160 $ ] is the same as in . by choosing @xmath276 ( the step function ) , reduces to . by choosing @xmath113 as in
, one sees from that the distribution of points at the right of the branching brownian motion conditioned by @xmath265 reaches a long time limit where @xmath265 only appears through the global shift @xmath277 .
this means that at large times , the distribution of the rightmost points in a branching brownian motion has a well defined measure _ independent of @xmath265 _ located around @xmath278 . as an example , if one chooses the function @xmath181 defined by , one can easily show from and that , in the @xmath278 frame , the average density of points at any position is infinite in the long time limit . if one considers two branching brownian motions @xmath59 and @xmath279 starting at arbitrary positions , then the points in @xmath59 at large time will be characterized by a random value @xmath280 and a realization of the point measure described by ; idem for the points in @xmath279 . if one considers the union of these two branching brownian motions , one gets from @xmath281}}\big ) = \exp\big(- e^{-[x-\ln(az)+f[\phi]]}\big),\ ] ] with @xmath282
this means that the point measure reached in the long time limit in the @xmath278 frame is the same whether one started initially with one , two or , by extension , any finite number of initial points at arbitrary positions on the line .
what does depend on the initial number of points is only the law of the random number @xmath265 , not the positions around @xmath278 .
this is to be related to the discussion in section [ superpos1 ] , where we showed that , in the long time limit , the measure seen from @xmath6 depends on the initial number of points while the measure seen from @xmath1 does not .
furthermore , the large time measure of the points in the @xmath278 frame has the following property : @xmath283 ( think of @xmath284 as the offspring of @xmath59 in the @xmath285 $ ] frame and of @xmath286 as @xmath287 ; idem for @xmath279 .
the shifts @xmath286 and @xmath288 are arbitrary because @xmath280 and @xmath289 are unbounded independent random numbers . )
the property of the point measure in the @xmath278 frame is a stronger version of the superposability property discussed in section [ superpos1 ] : clearly , it implies that the distribution of distances between the rightmost points is invariant by superposition , but it gives more information on the measure as it encompasses the position of the rightmost point .
in particular , one can check that , in any such measure , the rightmost point is gumble distributed .
the simplest point measure with the property is the poisson process with an exponential density @xmath290 , for an arbitrary @xmath291 .
furthermore , all the decorated measures of section [ sec : deco ] when @xmath86 is a poisson process with an exponential density @xmath290 are also superposable measures .
a natural question is then : can any superposable point measure be constructed as a decorated exponential poisson process for a well chosen decoration measure ?
a more specific question would be : is the limiting point measure of the branching brownian motion in the @xmath278 frame can be constructed as a decorated exponential poisson process ?
we conjecture that the answer to the last question is yes . indeed , it is known @xcite that if one considers two points chosen at random among the rightmost points in a branching brownian motion at a large time @xmath0 , then the time one needs to go back to find the most recent common ancestor of those two points is either very short ( of order 1 ) or very long ( of the order of the age @xmath0 of the system ) .
this means that one can group the rightmost points into families where two points belong to the same family if the branching event that generated them occurred recently .
the branching brownian motion could then be a decorated exponential poisson process , where the rightmost points of all the families are poisson distributed with an exponential density @xcite and where the members of a given family form the decoration .
this interpretation helps to understand a question raised by our previous work @xcite : the distances between points in the branching brownian motion are given by ( [ asympdi],[valuesdn ] ) , but if one considers only the rightmost point of each family , then the distances are given by with @xmath292 , as in the grem or in the ruelle cascade , see section [ ruellecascade ] .
in the present work , we have shown that all the statistical properties of the rightmost points in a branching brownian motion can be obtained by solving a front equation with a well - chosen initial condition .
the distribution of the positions of the rightmost points seen in the frame @xmath6 ( the average position of the rightmost ) has a long time limit .
the properties of the limiting distribution can be expressed as the long time delays of the traveling wave solution of the fisher - kpp equation when one varies the initial condition .
this limiting distribution is however modified if one considers the union of several branching brownian motions .
if one considers , however , only the distances between the points , for example if one looks at the distribution of all the positions of the rightmost points seen in the frame of the rightmost one , one obtains at large times another limiting distribution which does not depend on the initial positions of the branching brownian motions ( as long as there are finitely many of them ) .
we called this property superposability . in section
[ quantitative ] we have measured a few properties of this limiting distribution , and in section [ sec : disc ] we explained how our results can be extended to more general branching random walks .
lastly , in section [ statmeas ] , we argued that in the lalley and sellke frame , the branching brownian motion satisfies a stronger version of the superposability property , see . in the future , it would be interesting to characterize more precisely the limiting measure of the branching brownian motion and of the branching random walks to see whether some universal properties emerge .
for example , we believe that the average density seen from the rightmost point should always grow as in
. it would also be interesting to check whether , in lalley s and sellke s frame , the limiting distribution is indeed a decorated exponential , and to determine the properties of the decorating measure .
the question of ergodicity , raised at the beginning of section [ statmeas ] , is also an interesting open question .
we would like to thank julien berestycki and simon harris for interesting discussions .
we performed the numerical simulations by discretizing the branching brownian motion in space ( with a grid length @xmath221 ) and time ( with steps @xmath230 ) as in the example d ) of section [ sec : disc ] .
the corresponding front equation is given by , which is of course the most straightforward discretization of the fisher - kpp equation .
the solutions to the discrete equation converge to the solutions of the fisher - kpp equation , if @xmath293 with @xmath231 held constant and small enough .
we used three sets of values for @xmath230 and @xmath221 and computed the exact asymptotic velocity @xmath235 of the front and the decay rate @xmath240 of the asymptotic shape @xmath154 by minimizing numerically the function @xmath241 given in ( [ vgammas]d ) .
the decay rate @xmath288 for the probability of observing a distance @xmath59 between the two rightmost points , see equation , was computed using the recipe given at the end of section [ sec : distances ] and explained in appendix [ sec : largedeviation ] .
all these values are presented in table [ tabval ] .values of @xmath235 , @xmath240 and @xmath288 for our discretized branching processes , compared to the values in the fisher - kpp case .
[ cols= " < , < , < , < , < " , ] the simulations were made on a finite but large domain centered around the position of the front ; typically it extended to about a distance 1000 ahead and behind the center of the front ( respectively 4000 , 10000 or 20000 lattice sites depending on @xmath230 ) .
the values at @xmath294 were exactly computed and used for the boundaries of the domain . whenever the front moved by more than one unit space lattice , the whole data set was recentered .
the simulations were performed up to large times of order @xmath295 ( 0.8 to 24 millions of time steps ) and the data was extrapolated to obtain a value at @xmath296 . to do this extrapolation
, we used a more precise asymptotic expansion of the position of the front than : according to @xcite , @xmath297 where the number @xmath298 does not depend on the initial condition .
as we measure the delay @xmath299 , many terms cancel and one gets @xmath300+\frac{\delta a_1}t+\frac{\delta a_{3/2}}{t^{3/2}}+\cdots . \label{deltaasympt}\ ] ] all the quantities we measure are derivatives of @xmath160 $ ] , see , and have therefore the same large time expansion as .
thus , we extrapolated our numerical data to the large time limit by fitting it with the function @xmath301 for times larger than ( typically ) 5000 , see figure [ fit ] , and by using @xmath269 as the end result . by the function @xmath301 ( line ) .
the inset shows the quality of the fit by displaying the ratio between the data points and the fitting function . ] on figure [ fig : distrib_d ] , the three data points were presented together ; on figure [ fig : density ] , we have drawn together for each data set the function @xmath302 using in each case the value of @xmath240 of table [ tabval ] . in both cases ,
the superposition was nearly perfect , and so we expect that on the scales of the figure , the curves would not change noticeably for smaller values of @xmath230 and @xmath221 .
in this appendix we generalize , to any branching random walk , the argument leading to the asymptotic decay of the distribution of the distance between the two rightmost points in the branching brownian motion .
we consider a generic branching random walk in discrete space ( with spacing @xmath230 ) and time ( with intervals @xmath221 ) defined by the following family of functions @xmath303 we assume that @xmath304 , so that there is no extinction .
then @xmath305 can be thought as the probability that the point does not branch but moves by a distance @xmath205 . the continuous time and/or space cases
can be obtained as suitable @xmath293 and/or @xmath306 limits .
let @xmath307 $ ] be the generating function of the displacement during time @xmath0 of one point _ conditioned on the fact that this point does not branch _ : @xmath308 as the time steps are independent , the function @xmath309 can be computed during the time interval @xmath221 which gives @xmath310 note that @xmath311 as soon as the branching probability is non - zero .
we want now to evaluate the probability that a point moves a distance @xmath205 , without branching , during time @xmath0 . for large @xmath0
, it takes the form @xmath312,\ ] ] where @xmath313 is a large deviation function . using
, one finds that @xmath313 and @xmath309 are related by a legendre transform @xmath314 now , assuming as in section [ sec : distances ] that the events which contribute most to a large distance @xmath59 between the two rightmost points are those where the rightmost point moves , without branching , a distance @xmath59 ahead of the frontier of the branching brownian motion , one gets @xmath315 \right\ } , \label{prob2}\ ] ] where @xmath235 is the velocity of the front . for large @xmath59 ,
the optimal @xmath197 is also large and it satisfies , by derivation , @xmath316 let @xmath317 . using , equation becomes @xmath318 remarkably , this equation does not depend on @xmath59 .
replacing into gives @xmath319 the asymptotic decay rate of the probability distribution function of the distance between the two rightmost points is therefore simply the positive solution @xmath288 of with @xmath309 given by . in the branching brownian motion , @xmath320 , @xmath321 ,
so that gives indeed @xmath322 .
as can be checked easily from ( [ giveg ] ) , the function @xmath309 is convex .
therefore , as @xmath311 , equation has at most one positive solution and at most one negative solution .
the positive solution is the relevant one here .
we are now going to show that the negative solution of ( [ whichbeta ] ) , if it exists , gives the asymptotic shape of the associated traveling wave .
we write the front equation associated to the branching point process . using the same method as in section [ sec : bbmfkpp ] ,
we find @xmath323 we look at the shape @xmath258 of the traveling wave solution for a step initial condition , which moves asymptotically at the velocity @xmath235 : @xmath324 using , we see that in the _ stable _ region @xmath325 one has @xmath326 we look for an exponential solution to this linearized equation : @xmath327 with @xmath328 , as @xmath329 .
( note that a periodic modulation of this exponential could occur as @xmath205 takes only discrete values . ) inserting into and using , one finds that the equation for @xmath33 is @xmath330 which is the same equation as for @xmath331 . to summarize , a _ positive _ solution to gives the exponential decay rate of the probability distribution of the distance between the two rightmost points , see , while a _ negative _ solution gives the coefficient @xmath332 governing the shape of the front @xmath154 in the stable region @xmath325 .
lalley s and sellke s theorem @xcite is @xmath333 where @xmath269 is the constant appearing @xcite in the large @xmath9 expansion of the function @xmath154 defined in @xmath334 and @xmath265 is defined in . in words ,
given the positions @xmath335 at time @xmath230 , there is a @xmath164 limit to the probability that the rightmost is on the left of @xmath336 which depends , obviously , on the @xmath335 and is as such a random variable .
as @xmath230 goes to infinity , this random variable converges almost surely to the gumble distribution around @xmath277 .
this result can be extended into the following : for any suitable function @xmath113 ( see section [ bbm - frame ] ) , one has @xmath337
\big| \{x_i(s)\}\big\rangle = \exp\big({-a z e^{-x - f[\phi]}}\big ) , \label{ls2}\ ] ] where @xmath160 $ ] is the delay function . for @xmath276 , reduces to .
we first give an outline of lalley s and sellke s proof applied to the case .
given the positions @xmath338 of the points at time @xmath230 , the system as time @xmath339 can be seen as a collection of independent branching brownian motions at time @xmath340 starting from the @xmath338 . therefore @xmath341
\big| \{x_i(s)\}\big\rangle = \prod_i h_\phi\big(x - x_i(s),t - s\big),\ ] ] where the product in the right hand side is made on all the points present at time @xmath230 .
we replace @xmath9 by @xmath342 , to center around the position of the front , and suppose @xmath0 large .
it is easy to see from bramson s formula that @xmath343 as @xmath0 becomes large , so that @xmath344 \big| \{x_i(s)\}\big\rangle = \prod_i h_\phi\big(m_{t - s}+2s + x - x_i(s)+{o}(1),t - s\big),\ ] ] and , using , @xmath345 \big| \{x_i(s)\}\big\rangle = \prod_i f\big(2s+x - x_i(s)+f[\phi]\big).\ ] ] we now take @xmath230 large .
of all the points present at time @xmath230 , the rightmost is around @xmath346 , see .
therefore , @xmath347 diverges for all @xmath12 .
using , @xmath345 \big| \{x_i(s)\}\big\rangle \simeq\exp\big(-\sum_i \big[a\big(2s+x - x_i(s)+f[\phi]\big)+b\big]e^{-2s - x+x_i(s)-f[\phi ] } \big).\ ] ] following lalley and sellke , we introduce the quantities @xmath348e^{-2s+x_i(s)},\ ] ] see , so that @xmath345 \big| \{x_i(s)\}\big\rangle \simeq\exp\left(- \left[a z_s + \big(ax+af[\phi]+b\big)y_s\right]e^{-x - f[\phi]}\right ) .
\label{lstls}\ ] ] finally , the most technical part of lalley s and sellke s proof is that @xmath349 and @xmath350 are martingales converging when @xmath351 to @xmath352 and @xmath353 respectively , which leads to .
we do not reproduce this part of the proof here as it does not concern our extension with the function @xmath113 and it works in exactly as in . in ( [ ls2 ] )
, the average is made on all the realizations with a given set @xmath335 of points at a large time @xmath230 but the only relevant quantity appearing in the generating function is @xmath350 .
one would obviously have reached the same result if one had conditioned by @xmath350 instead of by the @xmath335 .
furthermore , as @xmath350 converges quickly to @xmath265 , as illustrated on figure [ fig2bbm ] , we argue that conditioning by @xmath350 at a large time @xmath230 or directly conditioning by @xmath265 should be equivalent , hence ( [ lsint],[ls2int ] ) .
kolmogorov , a. , petrovsky , i. , piscounov , n. : tude de lquation de la diffusion avec croissance de la quantit de matire et son application un problme biologique .
tat moscou , a * 1*(6 ) , 125 ( 1937 ) | we show that all the time - dependent statistical properties of the rightmost points of a branching brownian motion can be extracted from the traveling wave solutions of the fisher - kpp equation .
we show that the distribution of all the distances between the rightmost points has a long time limit which can be understood as the delay of the fisher - kpp traveling waves when the initial condition is modified .
the limiting measure exhibits the surprising property of superposability : the statistical properties of the distances between the rightmost points of the union of two realizations of the branching brownian motion shifted by arbitrary amounts are the same as those of a single realization .
we discuss the extension of our results to more general branching random walks . |
the size distribution of main - belt asteroids ( mbas ) traces their collisional history since the formation of primitive planetesimals .
previous survey observations for small solar system bodies have revealed the size distribution of mbas down to sub - km size ( ivezi et al .
2001 , yoshida & nakamura 2007 , gladman et al .
2009 ) . understanding the collisional processes of mbas
provides insight into the initial size distribution and collisional and dynamical evolution of planetesimals in the inner protoplanetary disk .
several models for collisional and orbital evolution of mbas suggest a dynamical excitation event at the final stage of the planet formation @xcite .
this event produced several features seen in the current main belt , such as a high eccentricity / inclination asteroid population and the radial mixing of taxonomic classes .
the dynamical excitation also ejected most asteroids out of the main belt , depleting more than 99% of the primordial mass @xcite .
the relative velocities among asteroids were elevated during this phase . in the main belt ,
asteroid collisions mostly occurred at higher velocities than the present average velocity of @xmath64 km s@xmath0 @xcite .
the relative velocity between a body in the remnant main belt population and one in the ejected population from the main belt zone exceeded 10 km s@xmath0 @xcite .
however , the fragment size distribution and ejecta velocities of such high - velocity collisions are still unclear ; reproduction by laboratory experiments is difficult @xcite . in our approach to this problem , we focus on mbas with high inclination
. the mean relative velocity of highly - inclined mbas at collisions with other mbas can be as large as @xmath610 km s@xmath0 @xcite . comparing the size distributions between low - inclination ( low-@xmath7 ) and high - inclination ( high-@xmath7 ) mbas could provide meaningful insight into disruptions at high velocity .
for investigating high-@xmath7 mbas , several published asteroid catalogs are available , but their detection limits are no smaller than kilometer size .
we performed observations for sub - km high-@xmath7 mbas to obtain their size distribution .
the inclination distribution of the known mba population drops steeply at @xmath8@xmath9 , where @xmath7 is the inclination . the scarcity of high-@xmath7 mbas is due not only to the small numbers but also to an observational bias .
most previous mba surveys have searched intensively around the ecliptic , but high-@xmath7 mbas only pass through the ecliptic plane briefly , spending most of their time in high ecliptic latitudes .
high-@xmath7 bodies are more likely to be found at @xmath10 where @xmath11 is the heliocentric ecliptic latitude than at low ecliptic latitudes ( @xmath12 ) .
we surveyed high ecliptic latitudes to detect high-@xmath7 mbas .
ivezi et al .
( 2001 ) showed that the surface density of asteroids in @xmath13 is only @xmath610 @xmath14 of that in @xmath15 , where @xmath16 is the geocentric ecliptic latitude ( a factor of 1.52 smaller than @xmath11 at opposition for mbas ) . to find a number of high-@xmath7 mbas , a wide - field survey is indispensable .
we performed a wide - field survey for faint mbas using data derived from the 8.2-m subaru telescope .
in addition , efficient observation is necessary to collect large amounts of data within a limited amount of time .
we adopted a less usual procedure , observing each field only two times with an interval of about 20 minutes .
terai et al .
( 2007 ) developed a new reduction technique to find asteroids efficiently in such two - visit observation data sets .
observations were conducted on june 3 2008 ( ut ) with the subaru telescope at the summit of mauna kea in hawaii .
we used the subaru prime focus camera ( suprime - cam ) , a mosaic camera of ten ccd chips .
it covers a 34@xmath17 field - of - view with a pixel scale of 0.20@xmath18 and has a high sensitivity of @xmath19 = 25.6 mag in 5-minutes integration for a point source @xcite .
the observational fields are near r.a.(j2000 ) = , dec.(j2000 ) = corresponding to @xmath20 at opposition .
we selected 22 fields with no bright stars , two of which were located on the sloan digital sky survey ( sdss ) fields .
several stars cataloged in the sdss databasehttp://www.sdss.org/@xmath21 . ] were used for photometric calibration .
all data were acquired with the sdss @xmath2-band filter @xcite .
the exposure time was 4 minutes .
each field was taken twice with an interval of about 20 minutes .
the total observed area of the sky was @xmath65.6 deg@xmath22 in 4 hours .
the typical seeing size was 0.6 arcsec .
the detection limit range was @xmath2 = 23.824.2 mag for moving objects whose motion corresponded to mbas ( see section 4.2 ) .
we used data with a limiting magnitude of @xmath2 = 24.0 mag .
the available data includes 14 fields , 3.6 deg@xmath22 ( table 1 ) .
we also used archival suprime - cam data obtained from the subaru - mitaka - okayama - kiso archive , smoka . ]
@xcite , consisting of @xmath19-band data with a detection limit of @xmath23 23.8 mag for moving objects ( section 4.2 ) .
the selected data were taken where the ecliptic latitudes with respect to opposition were within @xmath24 , allowing reasonably accurate orbital estimations from the short arcs .
we avoided star crowding regions around the galactic plane .
each field was taken at least twice , at intervals of 5 minutes to 1 hour .
we searched a total of 28 fields ( table 2 ) .
the images were processed chip - by - chip using iraf produced by the national optical astronomy observatories ( noao ) .
the raw data were reduced with standard processes consisting of bias subtraction , flat - fielding , sky subtraction , distortion correction , and positional correction .
we also used sdfred , the software package for suprime - cam data reduction ( yagi et al .
2002 , ouchi et al .
2004 ) to subtract the sky background and correct for geometric distortion . detecting a moving object from two - visit data
is a three - step procedure : ( i ) masking fixed stars and galaxies in the field , ( ii ) detecting the remaining sources from the processed images , and ( iii ) removing non - object sources such as cosmic rays .
first , the fixed objects were masked with the image - processing technique developed by terai et al .
two types of composite images were made from raw images : one excluding smaller values of each pixel ( or " image ) and one excluding larger values of each pixel ( and " image ) . the or image included all objects in either of the raw images , whereas the and image consisted of only fixed objects .
the and image was used to make a mask image whose pixel values were replaced with 0 for light sources and 1 for background .
this was multiplied by the or image to mask the fixed objects .
a moving object was recognized as a pair of light sources in the final processed image ( see fig .
the remaining sources were detected with the sextractor software @xcite .
the detection threshold was set to 2@xmath25 , where @xmath25 is the root - mean - square value of the background . but a large number of false sources were also identified .
most were discriminated using criteria such a sharpness of the point spread function ( psf ) , velocity of motion , and direction of motion .
an object pair that fulfils all of these criteria is considered a moving object candidate .
finally , each candidate was inspected visually to determine whether or not it was in fact a moving object .
we searched for moving objects in our observation data of 14 fields and the archival data of 28 fields . in the total area of 9.0 deg@xmath22
, 656 moving objects were detected .
the apparent motion of each object was measured from the shift in position between two images .
figure 2 shows the apparent motion distribution of the detected objects .
most of the objects have motion of @xmath63040 arcsec hr@xmath0 along the ecliptic longitude , corresponding to mbas .
the smallest motion rate was 3 arcsec hr@xmath0 , equivalent to a trans - neptunian object , and the largest was 107 arcsec hr@xmath0 .
highly inclined asteroids exhibited large motions in both the ecliptic longitude and latitude .
the flux of each moving object was measured from the total intensity of pixels within a non - circular aperture .
we used a capsule - shaped aperture given by a trail of a circle with the triple - fwhm ( full width at half maximum ) diameter in length of the motion during the exposure time .
the apparent magnitude of an asteroid was estimated from the mean flux in the two images .
flux calibration was conducted using several stars listed in the sdss database for our observation @xmath2-band data and the landolt standard stars for the archival @xmath26-band data . the limiting magnitude for moving objects
was evaluated by artificial trails added to the raw images .
they were made by imbricate pseudo point sources with the same fwhm of stars in the image .
we used 100 artificial trails with brightness steps of 0.1 mag .
we regarded the brightness corresponding to detection probability of 80 % as the limiting magnitude , which for our observational data was @xmath2 = 24.0 mag .
this was transformed to 23.8 mag for the @xmath26-band @xcite .
we used the archival data when its limiting magnitude was larger than @xmath26 = 23.8 mag . within the limiting magnitude ,
the photometric error was less than 0.1 mag .
to determine the orbit of an asteroid , we need at least three observations at different epochs . in this study
, we could not obtain precise orbits for the detected moving objects because we observed a field only twice in a short arc of several dozen minutes .
instead , we assumed a circular orbit .
this enabled us to estimate the semi - major axis and inclination of a moving object from its sky motion ( bowell et al .
1990 , nakamura & yoshida 2002 ) .
these elements were calculated using the equations for geocentric motion rate of a moving object presented by jedicke ( 1996 ) with an eccentricity of zero .
an object was excluded as an inaccurate orbit if its inclination was calculated to be larger than 40@xmath27 ( 9 objects ) .
we examined the estimation accuracy using the monte carlo simulations in nakamura and yoshida ( 2002 ) .
we generated 10,000 pseudo - mbas with randomly distributed orbital elements .
the ranges of their semi - major axes ( @xmath28 ) , eccentricities ( @xmath29 ) , and inclinations ( @xmath7 ) were @xmath302.13.3 au , @xmath310.00.4 , and @xmath320@xmath2730@xmath27 , respectively .
the eccentricity of each orbit was given with a probability based on the actual eccentricity distribution of known mbas .
the true anomaly was set with a probability following kepler s second law .
the apparent velocity of motion was calculated for a hypothetical observation field from the formulae in jedicke ( 1996 ) .
then , we estimated the semi - major axis and inclination assuming a circular orbit from the velocity of motion .
the accuracy of each element was represented by the standard deviations of differences between the estimated value and the given value .
if the separation from opposition in ecliptic longitude was less than 15@xmath27 , the semi - major axis error was @xmath60.10.15 au for a semi - major axis range of 2.53.3 au .
the error was larger than 0.15 au for semi - major axes between 2.02.5 au .
the inclination accuracy depends strongly on inclination .
nakamura and yoshida ( 2002 ) mentioned the degradation in inclination accuracy for @xmath33 ( the random error of inclination is more than 5@xmath27 ) .
however , accuracy is improved in high ecliptic latitudes .
the inclination error is 3@xmath274@xmath27 at @xmath20 where most high-@xmath7 mbas were detected .
this is small enough to obtain precise size distributions for low-@xmath7 and high-@xmath7 mbas . the semi - major axis distribution of the detected moving objects is shown in figure 3(a ) .
the distribution is concentrated between [email protected] au , similar to what is known for in the main belt .
we regarded 616 objects within this zone as the mba sample .
the semi - major axis distribution of the sample with @xmath35 2.5 au was similar to the actual population , which had a local minimum at 2.8 au and an outer edge at 3.3 au , corresponding to the 5:2 and 2:1 mean - motion resonances ( mmr ) with jupiter , respectively .
in contrast , the distribution in @xmath36 2.5 au did not appear to replicate the known distribution well .
the inclination distribution is shown in figure 3(b ) .
the two local peaks at @xmath610@xmath27 and @xmath616@xmath27 are quite different from the known distribution in the main belt .
this is because many of the surveyed fields are situated at geocentric ecliptic latitudes ( @xmath16 ) of 11@xmath27 , 17@xmath27 , and 25@xmath27 , corresponding to @xmath37 7@xmath27 , 11@xmath27 , and 16@xmath27 , respectively .
this means that the highest probability mba detections occur at inclinations of around 7@xmath27 , 11@xmath27 , and 16@xmath27 ( see section 2.1 ) .
the distribution of semi - major axis vs. inclination is plotted in figure 4 .
most bodies were located between the secular resonance @xmath38 and the 2:1 mmr , which agrees with the distribution of known mbas .
the absolute magnitude of an asteroid is defined as the brightness observed at a hypothetical location of @xmath2 = 1 au and @xmath39 = 1 au with the sun , the asteroid , and earth in a straight line .
it is given by @xmath40 where @xmath41 is the absolute magnitude and @xmath42 is the apparent magnitude in a given wavelength band .
the apparent magnitude is reduced by the phase angle @xmath43 , the sun - asteroid - earth angle following the phase function given by @xmath44 where @xmath45 represents the gradient of the phase function @xcite .
we used @xmath45 = 0.15 , which is generally used for asteroids with unknown taxonomic types .
@xmath46(@xmath43 ) and @xmath47(@xmath43 ) are given by @xmath48 \quad \left ( i = 1,2 \right),\ ] ] where @xmath49 = 3.33 , @xmath50 = 1.87 , @xmath51 = 0.63 , and @xmath52 = 1.22 @xcite .
although these are @xmath53-band values , we applied them to @xmath2- and @xmath26-band data . the relationship between absolute magnitude and diameter ( @xmath54 ) is described by @xmath55 where @xmath56 is the magnitude of the sun , -27.29 mag in the @xmath26-band @xcite .
this corresponds to -27.07 mag in the @xmath2-band , derived from the color transformation equations in fukugita et al .
albedo is represented by @xmath57 , which can not be determined in this survey .
therefore , we used typical albedo values .
asteroids have different albedos among the taxonomic groups . in photometric observation studies
, asteroids are divided into two major groups , s - like and c - like asteroids @xcite .
we estimated the mean albedos of s - like and c - like mbas from the albedo catalog presented by the iras minor planet survey and supplemental iras minor planet surveyhttp://sbn.psi.edu / pds / resource / imps.html@xmath21 . ]
( imps and simps , respectively ; tedesco et al .
the cataloged asteroids were classified into s - like and c - like groups according to the spectral database presented by the second phase of the small main - belt asteroid spectroscopic surveyhttp://sbn.psi.edu / pds / resource / smass2.html@xmath21 . ]
( smassii ; bus and binzel 2002 ) and the small solar system objects spectroscopic surveyhttp://sbn.psi.edu / pds / resource / s3os2.html@xmath21 . ]
( s@xmath58os@xmath1 ; lazarro et al .
the s - like asteroids include the s - complex and the minor classes including ld- , d- , t- , v- , and o - types .
the c - like asteroids consist of the c- and x - complexes .
the mean value is @xmath59 = 0.19 @xmath3 0.09 in the s - like asteroids ( 353 objects ) and @xmath60 = 0.08 @xmath3 0.07 in the c - like asteroids ( 649 objects ) .
the population ratio of s - like to c - like asteroids varies with heliocentric distance .
yoshida and nakamura ( 2007 ) showed that the ratio for asteroids of @xmath61 0.6 km are 1:1 in the inner belt ( 2.02.6 au ) , 1:2.3 in the middle belt ( 2.63.0 au ) , and 1:4 in the outer belt ( 3.03.5 au ) .
we estimated the mean albedo in each of the three zones from the population ratio and mean albedos of the s - like and c - like mbas .
we adopted the mean albedos as ( @xmath59+@xmath60)/(1 + 1 ) = 0.14 for 2.0 @xmath62 2.6 au , ( @xmath59 + 2.3@xmath60)/(1 + 2.3 ) = 0.11 for 2.6 @xmath62 3.0 au , and ( @xmath59 + 4@xmath60)/(1 + 4 ) = 0.10 for 3.0 @xmath62 3.3 au .
although , as discussed in section 5.2.2 , the population of s - like mbas is small in high inclination , the difference is negligible with respect to mean albedos .
the absolute magnitude is converted to diameter using these albedo values and eq.(4 ) .
figure 5 shows the diameter distribution for the detected mbas .
most of the bodies were smaller than 1 km in diameter .
considering the photometric error of @xmath60.1 mag and semi - major axis error of @xmath60.1 au , the inaccuracy of diameter estimation is about 15% .
when the error on the albedo ( @xmath60.06 ) is included , the added inaccuracy is 15% .
the detected mba sample included some observational bias , which had to be removed .
the most remarkable bias was a decrease in detection probabilities for faint objects .
the limiting magnitude ( 24.0 mag in the @xmath2-band or 23.8 mag in the @xmath26-band ) corresponded to @xmath54 = 0.7 km at a semi - major axis of 3.3 au with a typical phase angle of 25@xmath27 and a typical albedo of 0.1 .
we regarded asteroids with @xmath61 0.7 km as the sample for estimating size distribution .
in addition , we excluded asteroids with @xmath36 2.5 au for two reasons . first , this zone contains only a small population of high-@xmath7 mbas and s - like asteroids are abundant in this region , unlike in the outer zone .
if these asteroids were included in the sample , the fraction of s - like asteroids would differ significantly from the low-@xmath7 to high-@xmath7 mbas .
s - like and c - like asteroid size distributions are supposed to differ ( ivezi et al .
2001 , yoshida & nakamura 2007 , wiegert et al .
an excess of s - like population in the low-@xmath7 mbas could cause a bias in comparing size distributions between low-@xmath7 and high-@xmath7 mbas .
this bias is discussed again in section 5.2.2 .
second , the estimated semi - major axis for an asteroid with @xmath36 2.5 au has a large error ( @xmath630.15 au ) . ultimately , 178 asteroids with @xmath61 0.7 km were selected in @xmath28 = 2.53.3 au .
the relationship between the sub - km asteroid population and inclination was examined with this unbiased sample .
figure 6 shows the fractions of the sub - km asteroids for each inclination bin of 5@xmath27 .
the fractions are almost constant ( @xmath60.4 ) between @xmath64 and @xmath9 , and significantly decrease beyond 15@xmath27 .
we found a population deficiency of sub - km mbas with @xmath65 .
figure 7 shows the cumulative size distributions ( csds ) for the low-@xmath7 sample with @xmath66 and high-@xmath7 sample with @xmath65 .
bin size is @xmath67 $ ] = 0.05 , indicating the uncertainty of the diameter estimation .
the csd is described by the power - law equation as @xmath68 where @xmath69 is the cumulative number of asteroids larger than a diameter @xmath54 and @xmath70 is the power - law index of the csd .
the single power law was fit to each csd with 0.7 km @xmath71 2 km by the least squares method .
the best - fit indexes of the csds were @xmath70 = 1.79 @xmath3 0.05 for the low-@xmath7 mbas and @xmath70 = 1.62 @xmath3 0.07 for the high-@xmath7 mbas .
we performed statistical tests to compare the two regressions .
the regression lines are described by @xmath72 , where @xmath73 , @xmath74 , and @xmath75 .
first , a @xmath76-test was used to check whether the two populations possessed homogeneous variances around the regression lines .
we defined the null hypothesis as h@xmath77 : @xmath78 , against the alternate hypothesis as h@xmath79 : @xmath80 , where @xmath81 is a residual variance and the suffix 1 and 2 stand for the low-@xmath7 and high-@xmath7 mbas , respectively .
the computed @xmath76-value was @xmath82 2.00 ( @xmath83 ) less than the critical value @xmath84=3.18 .
the test accepted that @xmath85 and @xmath86 were equal with the 0.05 level of significance .
next , we performed a @xmath87-test with h@xmath77 : @xmath88=@xmath89 against h@xmath79 : @xmath90 .
let the population regression coefficient for @xmath70 be normally distributed with mean @xmath91 and variance @xmath92 where @xmath93 and @xmath94 is the mean of @xmath95 ( @xmath7=1,2 ) . in the case of equal variances ,
we use the criterion @xmath96 where @xmath97 and @xmath98 are sums of squares from the regression lines , derived from @xmath99 ^ 2 $ ] ( @xmath7=1,2 ) , and @xmath100 and @xmath101 are degrees of freedom @xcite .
the @xmath87-value for this test was calculated to be @xmath87=1.99 greater than the critical value @xmath102=1.76 with 14 degree of freedom .
we rejected the null hypothesis and accepted @xmath90 at the 0.05 level of significance . in conclusion , for asteroids of at least 1 km in diameter , the high-@xmath7 mbas had a shallower csd than the low-@xmath7 mbas at the 95% confidence level .
we used the published asteroid catalogs to obtain the size distribution of mbas larger than those detected in the subaru data .
the csds for @xmath61 10 km were constructed using the asteroid orbital elements database ( astorb)http://www.naic.edu/@xmath103nolan / astorb.html@xmath21 .
] distributed by lowell observatory @xcite .
in addition , the sdss moving object cataloghttp://www.astro.washington.edu / users / ivezic / sdssmoc / sdssmoc.html@xmath21 . ]
( sdss moc ; ivezi et al .
2002 ) was used for diameter in the 210 km range to link the csds based on the subaru and astorb data .
the astorb lists the orbital parameters and absolute magnitudes of more than 457,000 mbas .
we regarded 453,439 asteroids as mbas for which the semi - major axis was between 2.1 au and 3.3 au , the perihelion was larger than 1.66 au , and the aphelion was smaller than 4.61 au @xcite .
the diameter and albedo of 2228 mbas were measured by iras observations in the imps / simps surveys @xcite .
they cover almost all mbas with @xmath61 35 km . for mbas of unknown albedo
, the mean albedos deduced in section 4.4 were used to estimate diameters .
figure 8(a ) shows the absolute magnitude distribution and csd of the mbas in the astorb .
although the detection limit was never determined because the asteroids were found by several surveys with various limiting magnitudes , it appears to be a little under 10 km in diameter as seen in the absolute magnitude distribution . for km - size asteroids ,
the sdss moc was employed .
this lists the astrometric and photometric data for moving objects brighter than 21.5 mag in the @xmath2-band .
the fourth released version of sdss moc consists of 470,000 moving objects observed up to march 2007 .
the orbital elements of most sdss moc asteroids have not been determined .
the semi - major axis and inclination of each moving object in the sdss moc were estimated from apparent motions assuming a circular orbit , as in the analysis of the subaru data ( see section 4.3 ) . for estimation accuracy
, we used only objects with a phase angle smaller than 15@xmath27 .
a total of 130,000 objects located in 2.03.3 au in the semi - major axis were defined as mbas .
the diameter and albedo of each sdss moc asteroid are also unknown .
the apparent magnitudes of the moving objects were measured almost simultaneously in five bands ( @xmath104 , @xmath105 , @xmath2 , @xmath7 , and @xmath106 ) .
ivezi et al .
( 2001 ) defined a color index @xmath107 using @xmath108 and @xmath109 colors as @xmath110 to divide asteroids into two major groups , red " asteroids with @xmath111 and blue " asteroids with @xmath112 .
we considered the red " and blue " asteroids as s - like and c - like asteroids , respectively .
the diameter was obtained in the manner discussed in section 4.4 .
the albedo was given as 0.19 for s - like asteroids and 0.08 for c - like asteroids .
figure 8(b ) shows the absolute magnitude distribution and csd in the sdss moc . according to the absolute magnitude distribution , the detection limit was @xmath113 mag , which corresponds to @xmath114 km at the outer edge of the main belt with the albedo of c - like asteroids .
finally , we constructed the continuous csds using the astorb for @xmath115 km , the sdss moc for 2 km @xmath116 10 km , and subaru data for 0.7 km @xmath116 2 km .
we used the only asteroids with @xmath28 = 2.53.3 au to set the same sample condition as the analysis of the subaru data in section 5.1.1 .
the csds from the sdss moc were scaled to the cumulative number of the astorb sample at @xmath54 = 10 km .
the csd curves derived from the astorb and sdss moc connected smoothly in both the low-@xmath7 and high-@xmath7 samples . likewise , the csds from the subaru data were scaled to the cumulative number of the sdss moc sample at @xmath54 = 2 km .
figure 9 shows the combined csds of low-@xmath7 ( @xmath66 ) and high-@xmath7 ( @xmath65 ) mbas .
the csd slope of each sample from 0.7 km to 50 km in diameter was @xmath70 = 2.17 @xmath3 0.02 for low-@xmath7 mbas and @xmath70 = 2.02 @xmath3 0.03 for high-@xmath7 mbas .
the high-@xmath7 mbas had a shallower csd than the low-@xmath7 mbas , at least down to @xmath54 = 0.7 km .
the difference in @xmath70 was not caused by the wavy patterns seen in figure 9 .
for confirmation we adjusted the high-@xmath7 csd to the low-@xmath7 csd at @xmath54 = 50 km , and its power - law index was transformed to the same value as the low-@xmath7 csd ( see fig 10 ) .
the adjusted csd of high-@xmath7 mbas agreed well with the low-@xmath7 csd over all sizes . to state the significance of this result
, we performed a @xmath87-test ( see section 5.1.1 ) with the null hypothesis h@xmath77 : a regression line of the csd difference , @xmath117 where @xmath118 , is not significant ( @xmath119 ) and the alternative hypothesis @xmath120 : the regression line is significant ( @xmath121 ) .
the @xmath87-value is @xmath122 16.2 much greater than @xmath123 ( 2.43 ) , which rejected the null hypothesis at the 99% confidence level .
this test indicated that the overall high-@xmath7 csd was definitely shallower than the low-@xmath7 csd .
collisional fragmentation is characterized by impact strength , @xmath124 .
@xmath124 is the critical specific energy to fragment the target and disperse half of its mass .
it depends on asteroid size , generally expressed as @xmath125 the @xmath124 curve has two different power laws with opposite signs . for small bodies , @xmath124 decreases with increasing diameter ( @xmath126 ) , in what is called the strength - scaled regime .
in contrast , strength increases with increasing diameter ( @xmath127 ) for large bodies , in what is called the gravity - scaled regime .
the two slopes join at the transition diameter ( @xmath128 ) of 0.11 km @xcite . obrien and greenberg ( 2003 ) showed that the power - law index of the csd with @xmath129 is determined by the power - law index of the @xmath124 curve in the gravity - scaled regime , @xmath130 . according to the collisional steady - state scenario ,
this relationship is described by @xmath131 in the present study , the overall csd slope of high-@xmath7 mbas ( @xmath70 = 2.02 ) was smaller than that of low-@xmath7 mbas ( @xmath70 = 2.17 ) , corresponding to @xmath130 = 0.91 and @xmath130 = 1.43 , respectively .
the most remarkable difference between the low-@xmath7 and high-@xmath7 mbas was collision velocity .
the mean relative velocity among mbas , most of which are low-@xmath7 bodies , is 3.8 km s@xmath0 @xcite .
in contrast , high-@xmath7 mbas often collide at high velocities , around 10 km s@xmath0 @xcite .
we propose that these hypervelocity collisions have a large @xmath130 .
however , such @xmath130 has not been determined through previous impact experiments nor through numerical simulations .
benz and asphaug ( 1999 ) performed impact simulations for basalt at two different collision velocities , 3 km s@xmath0 and 5 km s@xmath0 , and found that @xmath130 did not vary between the two velocities .
however , jutzi et al .
( 2010 ) , using impact simulations , reported that the @xmath130 of basalt targets for 5 km s@xmath0 ( @xmath130 = 1.29 ) was smaller than that for 3 km s@xmath0 ( @xmath130 = 1.33 ) .
these values correspond to @xmath70 = 2.06 and @xmath70 = 2.05 , respectively .
this difference is too small to be recognized through observations .
their results showed that @xmath130 does not depend strongly on collisional velocity enough to shift @xmath70 .
but the relationship between the @xmath124 curve and collision velocity , especially in the hypervelocity collisions around 10 km s@xmath0 , is still uncertain .
our study might present evidence for an increase in @xmath130 in hypervelocity collisions around 10 km s@xmath0 . examined qualitatively ,
it is possible that the different spatial distributions between taxonomic groups caused the discrepancy in the csd slopes between low-@xmath7 and high-@xmath7 mbas .
the population ratio of s - like to c - like asteroids is not uniform with inclination .
moth - diniz et al .
( 2003 ) showed that a smaller fraction of s - complex mbas with @xmath61 7 km appears in the high eccentricity / inclination regime . meanwhile , several observations have indicated different csd slopes for different taxonomic groups , mostly in low inclination ( ivezi et al .
2001 , yoshida & nakamura 2007 , wiegert et al . 2007 ) .
this difference in abundance between the taxonomic groups could induce an apparent decrease in the csd slope in high inclination .
we checked this effect using the sdss moc data .
the sdss asteroids were separated into s - like and c - like groups by @xmath132 color defined in eq.(6 ) .
the low-@xmath7 mbas included 6585 s - like asteroids and 9746 c - like asteroids .
the high-@xmath7 mbas include 1139 s - like asteroids and 2019 c - like asteroids .
the csds of the low-@xmath7 and high-@xmath7 mbas ( 2.5 au @xmath62 3.3 au ) were constructed in each group .
the power - laws for these csds were approximated for the 210 km in diameter range . among the s - like asteroids , the slope , named @xmath133 , was 2.66 @xmath3 0.04 for low-@xmath7 mbas and 2.26 @xmath3 0.03 for high-@xmath7 mbas . among the c - like asteroids , the slope , named @xmath134 , was 2.41 @xmath3 0.02 for low-@xmath7 mbas and 2.36 @xmath3 0.05 for high-@xmath7 mbas .
the @xmath133 value was significantly smaller in high inclination , while @xmath134 was similar for low-@xmath7 and high-@xmath7 mbas .
we evaluated the contribution of the s - like mbas to the decrease in @xmath70 of high-@xmath7 mbas .
model csds were compounded from artificial power - law csds of s - like and c - like mbas based on a population model .
the model was characterized by a population ratio of s - like mbas to c - like mbas and their csd slopes ( table 3 ) .
we compared indexes of fitted power laws to the model csds with those of the observed csds using two models . in model 1
, we explored the possibility that the small @xmath133 of high-@xmath7 s - like mbas was a major factor in the small @xmath70 of total high-@xmath7 mbas .
the population ratio of s - like asteroids was set to the same value as the sdss moc data , 0.67 ( 6585/9746 ) for low-@xmath7 mbas and 0.56 ( 1139/2019 ) for high-@xmath7 mbas .
the @xmath133 and @xmath134 values for high-@xmath7 mbas were set to match the low-@xmath7 mbas ( @xmath133=2.66 , @xmath134=2.41 ) .
the model csds showed @xmath135 2.50 for low-@xmath7 mbas and @xmath135 2.49 for high-@xmath7 mbas .
the csd slope of high-@xmath7 mbas was not as small as that of the observed csd .
consequently , these results show that a small ( compound ) slope @xmath70 in the high-@xmath7 mbas requires a smaller s - type slope @xmath133 for that region . in model 2
, we examined the possibility that the small s - like / c - like population ratio of high-@xmath7 mbas caused the small @xmath70 .
the test was performed assuming that high-@xmath7 mbas have the same population ratio as low-@xmath7 mbas .
the csd power - law indexes were set to match the sdss moc data , ( @xmath133 , @xmath134 ) = ( 2.66 , 2.41 ) for low-@xmath7 mbas and ( @xmath133 , @xmath134 ) = ( 2.26 , 2.36 ) for high-@xmath7 mbas .
the best - fit power - law indexes of the compounded csds were @xmath135 2.50 for low-@xmath7 mbas and @xmath135 2.32 for high-@xmath7 mbas .
the both of @xmath70 values agreed with the observed csds .
this indicates that the scarcity of s - like asteroids with high inclination had little influence on the csd slope .
we conclude that the shallow csd of the high-@xmath7 mbas was due to the small csd slope of s - like mbas with high inclination .
we suggest that @xmath130 values for high - velocity collisions ( around 10 km s@xmath0 ) of s - like asteroids are large , while @xmath130 values for similar c - like asteroids are not .
one possible cause is the variation in porosities between s - like and c - like asteroids .
c - type asteroids have porosities of 3560% , whereas s - type asteroids have porosities of around 30% @xcite .
jutzi et al .
( 2010 ) showed through impact simulations that @xmath130 is constant ( 1.22 ) between collision velocities of 3 km s@xmath0 and 5 km s@xmath0 for porous targets ( pumice ) . on the other hand , the @xmath130 values for 5 km s@xmath0 collisions ( 1.29 )
is smaller than 3 km s@xmath0 collisions ( 1.36 ) for the non - porous targets ( basalt ) .
porosity may have a resistance effect to @xmath130 variation with collision velocity .
note that these considerations include uncertainty .
the wavy structure of the csd may differ between s - like and c - like mbas . estimating the csd slopes from a narrow size range is not appropriate for comparing the overall slope @xmath70 .
indeed , previous surveys have presented a variety of results .
ivezi et al .
( 2001 ) showed @xmath70 = 3.00 @xmath3 0.05 for both groups with @xmath136 5 km , and @xmath70 = 1.20 @xmath3 0.05 for s - like asteroids and @xmath70 = 1.40 @xmath3 0.05 for c - like asteroids with @xmath137 5 km .
yoshida and nakamura ( 2007 ) showed @xmath70 = 2.44 @xmath3 0.09 for s - like asteroids with 1 km @xmath71 3 km and @xmath70 = 1.29 @xmath3 0.02 with 0.3 km @xmath71 1 km , and a single slope @xmath70 = 1.33 @xmath3 0.03 for c - like asteroids with 0.6 km @xmath71 7 km . the difference in csd between the s - like and c - like asteroids is still unclear .
dynamical processes , as well as collisional processes , affect the shape of the csd . obrien and greenberg ( 2005 ) showed via numerical simulations that the csd slope of mbas is reduced from @xmath620 km in diameter through the action of the yarkovsky effect and resonances .
the yarkovsky effect causes a drift in the semi - major axis , the rate of change of which increases with decreasing size down to 0.01 km in diameter .
asteroids caught in a strong resonance are removed from the main belt .
smaller asteroids are removed more quickly from the main belt .
although the yarkovsky effect is independent of inclination , several secular resonances appear at high inclination . within the heliocentric distance of 2.53.3 au , none of these resonances appear at @xmath66 . in contrast , the strong secular resonances of @xmath138 , @xmath38 , and @xmath139 lie in @xmath65 @xcite .
high-@xmath7 asteroids in this zone are subject to removal .
these secular resonances together with the yarkovsky effect could cause significant losses of high-@xmath7 mbas , possibly enough to diminish the csd slope .
this action increases the difference in @xmath70 between the low-@xmath7 and high-@xmath7 mbas at small sizes ( @xmath140 km ) .
however , as shown in figure 10 , the difference in @xmath70 is almost constant as the diameter decreases from @xmath141 km .
this is inconsistent with the above expectation .
hence , dynamical removal does not decrease @xmath70 in high inclination .
we performed wide - field observations of high-@xmath7 mbas of sub - km diameter using the 8.2 m subaru telescope .
the csds of low-@xmath7 and high-@xmath7 mbas were constructed using the subaru data and the astorb and sdss moc catalogs .
a summary of our main conclusions follows .
\(1 ) a smaller fraction of small asteroids ( 0.7 km @xmath71 1 km ) appear in high inclination .
the power - law index of the csd for the 0.72 km diameter range is 1.79 @xmath3 0.05 for low-@xmath7 mbas ( @xmath142 ) and 1.62 @xmath3 0.07 for high-@xmath7 ( @xmath143 ) mbas .
the sub - km asteroids at @xmath143 have a lower index than the low-@xmath7 sample at the 95% confidence level based in a statistical @xmath87-test .
\(2 ) the single power - law slope of the combined csd for diameter in the 0.750 km range is 2.17 @xmath3 0.02 for the low-@xmath7 mbas and 2.02 @xmath3 0.03 for the high-@xmath7 mbas .
the high-@xmath7 csd is shallower than the low-@xmath7 csd for the entire size range at the 99% confidence level .
this is not caused by the difference in the wavy pattern of the csds .
\(3 ) the csd of s - like asteroids has a small slope in high inclination , whereas that of c - like asteroids shows little variation in slope with inclination . through modeling
we showed that the shallow csd of the high-@xmath7 mbas is not caused by the spatial distribution of the taxonomic groups , but by the shallow csd of the s - like mbas with high inclination .
\(4 ) the difference in slope of csds between the low-@xmath7 and high-@xmath7 mbas is constant across @xmath144 km .
the shallow csd of the high-@xmath7 mbas is not the result of dynamical removal due to the yarkovsky effect and the secular resonances .
we suggest that the small @xmath70 of s - like mbas with high inclination is due to a collisional effect .
the possible explanation is that the @xmath124 curve has a large gravity - scaled regime @xmath130 slope under hypervelocity collisions ( around 10 km s@xmath0 ) .
asteroid collisions often occurred with such high velocities in the dynamical excitation phase in the final stage of planet formation .
we suppose that during this phase , mbas experienced oligopolistic collisional evolution ; small bodies were more easily disrupted relative to large bodies than at present .
this indication claims that the current evolutionary models for mbas should be modified .
this work was partly supported by the 21st century coe program : the origin and evolution of planetary systems " of the mext .
t. terai was supported by the grant - in - aid from japan society for the promotion of science ( 20 - 4879 ) .
baba , h. , yasuda , n. , ichikawa , s. , yagi , m. , iwamoto , n. , takata , t. , horaguchi , t. , taga , m. , watanabe , m. , ozawa , t. , & hamabe , m. 2002 , asp conf .
vol . 281 , astronomical data analysis software and systems xi , ed .
d. a. bohlender , d. durand , & t. h. handley ( san francisco , ca : asp ) , 298 benz , w. , & asphaug , e. 1999 , icarus , 142 , 5 bertin , e. , & arnouts , s. 1996 , , 117 , 393 bottke , w. f. , durda , d. d. , nesvorn , d. , jedicke , r. , morbidelli , a. , vokrouhlick , d. , & levison , h. f. 2005 , icarus , 179 , 63 bottke , w. f. , morbidelli , a. , jedicke , r. , petit , j .-
m . , levison , h. f. , michel , p. , & metcalfe , t. s. 2002 , icarus , 156 , 399 bottke , w. f. , nolan , m. c. , greenberg , r. , & kolvoord , r. a. 1994 , icarus , 107 , 255 bowell , e 2001 , the asteroid orbital elements database , astorb ( flagstaff : lowell observatory ) bowell , e. , hapke , b. , domingue , d. , lumme , k. , peltoniemi , j. , & harris , a. w. 1989 , asteroids ii , ed .
r. p. binzel , t. gehrels , and m. s. matthews ( tucson : university of arizona press ) , 524 bowell , e. , skiff , b. a. , wasserman , l. h. , & russel , k. s. 1990 , asteroids , comets , meteors iii , ed . c. i. lagerkvist , h. rickman , and b. a. lindblad ( uppsala : uppsala universitet ) , 19 britt , d. t. , yeomans , d. , housen , k. , & consolmagno , g. 2002 , asteroids iii , ed .
w. f. bottke jr . , a. cellino , p. paolicchi , and r. p. binzel ( tucson : university of arizona press ) , 485 bus , s. j. , & binzel , r. p. 2002 , icarus , 158 , 146 drilling , j. s. , & landolt , a. u. 2000 , allen s astrophysical quantities , ed .
arthur n. cox ( 4th ed . ; aip press ) durda , d. d. , greenberg , r. , & jedicke , r. 1998 , icarus , 135 , 431 farinella , p. , & davis , d. r. icarus , 97 , 111 fukugita , m. , ichikawa , t. , gunn , j. e. , doi , m. , shimasaku , k. , & schneider , d. p. 1996
, , 111 , 1748 gil - hutton , r. 2006 , icarus , 183 , 93 gladman , b. j. , davis , d. r. , neese , c. , jedicke , r. , williams , g. , kavelaars , j. j. , petit , j .-
m . , scholl , h. , holman , m. , warrington , b. , esquerdo , g. , & tricarico , p. 2009
, icarus , 202 , 104 ivezi , . , juri '
c , m. , lupton , r. h. , tabachnik , s. & quinn , t. ( the sdss collaboration ) 2002 , , 4836 , 98 ivezi , . , tabachnik , s. , rafikov , r. , lupton , r. h. , quinn , t. , hammergren , m. , eyer , l. , chu , j. , armstrong , j. c. , fan , x. , finlator , k. , geballe , t. r. , gunn , j. e. , hennessy , g. s. , knapp , g. r. , leggett , s. k. , munn , j. a. , pier , j. r. , rockosi , c. m. , schneider , d. p. , strauss , m. a. , yanny , b. , brinkmann , j. , csabai , i. , hindsley , r. b. , kent , s. , lamb , d. q. , margon , b. , mckay , t. a. , smith , j. a. , waddel , p. , & york , d. g. , for the sdss collaboration 2001 , , 122 , 2749 jedicke , r. 1996 , , 111 , 970 jedicke , r. , & metcalfe , t. s. 1998 , icarus , 131 , 245 jutzi , m. , michel , p. , benz , wi . , richardson , d. c. , 2010 , icarus , 207 , 54 kadono , t. , sakaiya , t. , hironaka , y. , otani , k. , sano , t. , fujiwara , t. , mochiyama , t. , kurosawa , k. , sugita , s. , sekine , y. , nishikanbara , w. , matsui , t. , ohno , s. , shiroshita , a. , miyanishi , k. , ozaki , n. , kodama , r. , nakamura , a. m. , arakawa , m. , fujioka , s. , & shigemori , k. 2010 , , 115 , e04003 kneevi , z. , milani , a. , farinella , p. , froeschle , ch .
, froeschle , cl .
1991 , icarus , 93 , 316 lazzaro , d. , angeli , c. a. , carvano , j. m. , mothe-diniz , t. , duffard , r. , & florczak , m. 2004 , icarus , 172 , 179 miyazaki , s. , komiyama , y. , sekiguchi , m. , okamura , s. , doi , m. , furusawa , h. , hamabe , m. , imi , k. , kimura , m. , nakata , f. , okada , n. , ouchi , m. , shimasaku , k. , yagi , m. , & yasuda , n. 2002 , , 54 , 833 moth - diniz , t. , carvano , j. m. , & lazzaro , d. 2003 , icarus , 162 , 10 .
nakamura , t. , & yoshida , f. 2002 , , 54 , 1079 obrien , d. p. , & greenberg , r. 2003 , icarus , 164 , 334 obrien , d. p. , & greenberg , r. 2005 , icarus , 178 , 179 ouchi , m. , shimasaku , k. , okamura , s. , furusawa , h. , kashikawa , n. , ota , k. , doi , m. , hamabe , m. , kimura , m. , komiyama , y. , miyazaki , m. , miyazaki , s. , nakata , f. , sekiguchi , m. , yagi , m. , & yasuda , n. 2004 , , 611 , 660 petit , j .-
m . , chambers , j. , franklin , f. , & nagasawa , m. 2002 , asteroids iii , ed . w. f. bottke jr .
, a. cellino , p. paolicchi , and r. p. binzel ( tucson : university of arizona press ) , 711 tedesco , e. f. , noah , p. v. , noah , m. , & price , s. d. 2002 , , 123 , 1056 terai , t. , itoh , y. , & mukai , t. 2007 , , 59 , 1175 vedder , j. d. 1998 , icarus , 131 , 283 welch , b. l. 1938 , biometrika , 29 , 350 wiegert , p. , balam , d. , moss , a. , veillet , c. , connors , m. , & shelton , i. 2007 , , 133 , 1609 yagi , m. , kashikawa , n. , sekiguchi , m. , doi , m. , yasuda , n. , shimasaku , k. , & okamura , s. 2002 , , 123 , 66 yoshida , f. , & nakamura , t. 2007 , , 55 , 1113 yoshida , f. , nakamura , t. , watanabe , j. , kinoshita , d. , yamamoto , n. , & fuse , t. 2003 , , 55 , 701 lcccc & s / c & @xmath133 & @xmath134 & @xmath70 + sdss moc + low-@xmath7 & 0.67 & 2.66 & 2.41 & [email protected] + high-@xmath7 & 0.56 & 2.26 & 2.36 & [email protected] + + model 1 + low-@xmath7 & 0.67 & 2.66 & 2.41 & 2.50 + high-@xmath7 & 0.56 & 2.66 & 2.41 & 2.49 + + model 2 + low-@xmath7 & 0.67 & 2.66 & 2.41 & 2.50 + high-@xmath7 & 0.67 & 2.26 & 2.36 & 2.32 + | we investigated the size distribution of high - inclination main - belt asteroids ( mbas ) to explore asteroid collisional evolution under hypervelocity collisions of around 10 km s@xmath0 .
we performed a wide - field survey for high - inclination sub - km mbas using the 8.2-m subaru telescope with the subaru prime focus camera ( suprime - cam ) .
suprime - cam archival data were also used .
a total of 616 mba candidates were detected in an area of 9.0 deg@xmath1 with a limiting magnitude of 24.0 mag in the sdss @xmath2 filter .
most of candidate diameters were estimated to be smaller than 1 km .
we found a scarcity of sub - km mbas with high inclination .
cumulative size distributions ( csds ) were constructed using subaru data and published asteroid catalogs .
the power - law indexes of the csds were 2.17 @xmath3 0.02 for low - inclination ( @xmath4 ) mbas and 2.02 @xmath3 0.03 for high - inclination ( @xmath5 ) mbas in the 0.750 km diameter range .
the high - inclination mbas had a shallower csd .
we also found that the csd of s - like mbas had a small slope with high inclination , whereas the slope did not vary with inclination in the c - like group .
the most probable cause of the shallow csd of the high - inclination s - like mbas is the large power - law index in the diameter
impact strength curve in hypervelocity collisions .
the collisional evolution of mbas may have advanced with oligopolistic survival during the dynamical excitation phase in the final stage of planet formation . |
since their discovery in 1977 , the @xmath4 resonances have provided a unique laboratory for the study of strong interactions ( qcd ) , while the @xmath5 resonance provides a clean source of @xmath6 mesons which are used to probe the weak interaction .
the analyses reported on here make use of approximately 20@xmath710@xmath8 @xmath3 , 9@xmath710@xmath8 @xmath1 , and @xmath9 @xmath2 decays recorded using the cleo iii detector at cesr .
the analyses also utilize 0.2 @xmath10 , 0.4 @xmath10 and 0.2 @xmath10 of data taken just below the @xmath3 , @xmath1 , and @xmath2 resonances ( _ ie . , continuum _ ) , respectively .
the @xmath4 resonances have provided a fertile testing ground for both lattice qcd and qcd - inspired potential models . of particular interest
are the total widths of these resonances , @xmath11 , which can be related to the muonic branching fraction , @xmath12 via @xmath13 , where the electronic decay width , @xmath14 , can be obtained from the integrated resonant cross - section for @xmath15 .
@xmath16 also enters into the determination of other @xmath4 branching fraction measurements , and therefore may present a limiting systematic uncertainty .
cleo is now in a position to provide a much improved measurement of @xmath12 @xcite .
the branching fraction @xmath12 can be extracted by measuring the efficiency - corrected ratio : @xmath17 , which is related to @xmath12 through @xmath18 .
dimuon events are selected by requiring exactly two nearly back - to - back charged tracks ( @xmath19 ) with energy @xmath20 in the range @xmath21 , one which must be detected in the muon system .
hadronic events are selected by requiring @xmath22 , with additional requirements on the energy in the calorimeter when @xmath23 .
the number of dimuon ( hadronic ) events surviving all analysis requirements are @xmath24 ( @xmath25 ) , @xmath26 ( @xmath27 ) , and @xmath28 ( @xmath29 ) for @xmath3 , @xmath1 , and @xmath2 , respectively . event selection and analysis requirement efficiencies
are determined from simulation and are 65% ( 96 - 98% ) for the dimuon ( hadronic ) event selection .
backgrounds are estimated using simulation and are below 3% for the dimuon samples and less than 1% for the hadronic samples .
the branching fractions are found to be : @xmath30 , @xmath31 and @xmath32 , where the first uncertainty is statistical and the second systematic .
the @xmath3 measurement is consistent with previous measurements , but the @xmath1 and @xmath2 are larger by 2 - 3 standard deviations .
these latter two measurements provide factor of 2 - 3 improvement in the precision to which these branching fractions are known while the @xmath3 result is by itself competitive with the world average .
measurements of the radiative decays of the @xmath4 resonances provide clues into the nature of the strong force and provides for unique tests of potential models and lattice qcd .
for example , the electric dipole transition rate , @xmath33 can be determined experimentally , and can in turn be used to extract the corresponding matrix element . the matrix elements and the mass splittings
can be used to learn about the contributing spin - orbit and tensor interactions @xcite . the events for this analysis @xcite use the same hadronic event selection as in the @xmath12 analysis .
radiative transitions are observed through their peaks in the inclusive photon energy spectrum .
photon candidates are required to be isolated from all charged particles and have a shower profile consistent with an electromagnetic shower .
the predominant source of background arises from @xmath34 decay .
application of a @xmath35 veto for low energy photons does not improve the signal significance due to accidental combinations where the signal photon and a random photon have an invariant mass near the @xmath35 mass .
consequently , the @xmath35 veto is not applied to the @xmath36 and @xmath37 analyses . on the other hand ,
a @xmath35 veto is applied to the @xmath38 analysis , where the photon energies are much higher and accidental combinatorial losses are much smaller . the inclusive photon spectrum for @xmath1 data is shown in fig . [ fig : y2s_1p ] .
the top figure shows the data ( points ) and overlayed are the fits to the data ( solid line ) and the background contribution ( dashed line ) .
the lower plot shows the background subtracted peaks along with the fit ( solid line ) and the contributions from the individual @xmath39 states ( dashed line ) .
the background is determined using a combination of continuum and @xmath3 data .
the analogous distributions for @xmath2 data are shown in fig .
[ fig : y3s_2p ] . in both cases ,
large signals are observed for each of the transitions .
the yields , efficiencies , branching fractions and photon energies are summarized in table [ tab : upsrad ] .
the branching fraction measurements are by themselves about a factor of two better than the world average values . using the inclusive photon spectrum , we also observe the rare transition , @xmath40 , which is found to have a branching fraction of @xmath41 .
this is the first measurement of this transition rate . [
cols=">,^,^,^,^",options="header " , ]
one of the longstanding problems in qcd has been to understand the production mechanism for charmonium in a glue - rich environment . to account for the large excess of charmonium observed at the tevtron ,
an additional color - octet mechanism @xcite was proposed whereby a @xmath42 pair is produced in a color - octet state , and then evolves non - perturbatively into a color - singlet .
the matrix element for the latter was determined as to account for the deficit .
it has been suggested that the study of @xmath43 production in three - gluon decays of @xmath4 provides a unique opportunity to test color - octet predictions .
the most dramatic expectation is a peaking of the @xmath43 momentum spectrum near the kinematic endpoint .
a second mechanism which can also produce @xmath43 in @xmath4 decay is the color - singlet process , @xmath44 , which is inherently soft and should yield two additional open charm particles .
events used in this analysis @xcite are required to pass standard hadronic event selection . in order to suppress background from radiative return ( @xmath45 )
, we reject events which have @xmath46 and either a detected photon with momentum greater than 3.75 gev / c , or an event momentum imbalance of larger than 2 gev / c .
candidates @xmath43 s are reconstructed in the @xmath47 and @xmath48 modes .
in addition to standard track selection criteria , muons are required to be detected in the muon system .
electrons are required to deposit energy in the calorimeter which is consistent with its measured momentum .
the invariant mass distributions for @xmath47 and @xmath48 candidates are shown in fig .
[ fig : mass_1s ] . the continuum contribution ( @xmath49 )
is estimated using data taken on and below the @xmath5 .
this contribution is added to the expected @xmath50 contribution , and the total subtracted from the on-@xmath3 data .
the resulting scaled momentum distributions for @xmath47 and @xmath48 in @xmath3 decays are shown in fig .
[ fig : x_spec ] .
the data are compared to predictions of the color - octet ( solid curve ) and color - singlet ( dashed curve ) .
the data are much softer than the color - octet predictions and moderately softer than the color - singlet predictions . in the latter case , the effects of fragmentation of the recoiling charm into @xmath51 , @xmath52 , @xmath53 , @xmath54 , etc
is not taken into account and is expected to further soften the spectrum . integrating the spectrum , we measure a branching fraction @xmath55 .
this measurement is consistent with predictions of both the color - octet and color - singlet models which predict branching fractions of @xmath56 and @xmath57 , respectively .
we have also made a first observation of the decay , @xmath58 ( see fig . [
fig : psi2s ] ) and found evidence for @xmath59 .
the branching fractions are measured relative to @xmath60 , and are found to be : @xmath61 , @xmath62 , @xmath63 , and @xmath64 at 90% confidence level .
the feed - down contributions to @xmath43 are about a factor of two higher than expected .
these measurements have the potential to shed some additional light on the production mechanisms of charmonium in glue - rich environments .
i gratefully acknowledge the effort of the cesr staff in providing us with excellent luminosity and running conditions .
i also thank the national science foundation and the u.s . dept . of energy for their support of this research . | using data collected using the cleo iii detector , we present recent results on decays of the @xmath0 resonances .
we report on three analyses .
they are : ( 1 ) improved measurements of the muonic branching fraction of the @xmath0 , ( 2 ) precision measurements of @xmath1 and @xmath2 photonic transitions , and ( 3 ) new measurements of @xmath3 decays to charmonium final states . # 1#2#3 0 ( 1s)_c0+x 1 ( 1s)_c1+x 2 ( 1s)_c2+x f 4s ( 4s ) 4 ( 4s ) |
in this paper we investigate the connection between a standard construction in algebraic geometry ( elementary modifications of bundles ) and a standard construction in the study of hyperplane arrangements ( the deletion / restriction operation ) . in the setting of line configurations in @xmath3
, it turns out that they are the same thing .
given a two bundle @xmath4 and line @xmath5 , if @xmath4 sits in a modification @xmath6 then understanding @xmath4 means understanding @xmath7 and @xmath8 ( here @xmath9 is the inclusion of @xmath5 in @xmath3 ) . for line configurations ,
the twist @xmath8 has a simple combinatorial meaning , and the long exact sequence in cohomology yields information about the arrangement .
in particular , it gives a bound on the castelnuovo - mumford regularity of the module of derivations tangent to @xmath2 . in 2
we give a quick review of the fundamental objects : syzygy modules and hyperplane arrangements .
the module of @xmath2 derivations is denoted @xmath1 , it consists of derivations of @xmath10 tangent to @xmath2 .
the syzygy module on the jacobian ideal of the singular hypersurface @xmath2 is a summand of @xmath1 , hence the connection . in 3 we discuss elementary modifications and castelnuovo - mumford regularity , and in 4 turn our attention to stability .
we close with an application of these results to terao s conjecture that the freeness of @xmath1 depends only on the combinatorics of the arrangement .
let @xmath11 $ ] and @xmath12 a codimension two ideal ; @xmath13 defines a map : @xmath14 let @xmath15 denote the sheaf associated to the module of syzygies on @xmath13 .
@xmath16 iff @xmath13 is saturated .
since @xmath15 is a second syzygy sheaf , @xmath15 is locally free .
let @xmath17 be the scheme defined by @xmath13 ; the following lemma follows from standard properties of chern classes : [ lem : cclass ] the chern classes of @xmath15 are @xmath18 , @xmath19 . in @xcite
, serre describes a method of constructing a rank two vector bundle @xmath20 on @xmath10 from a codimension two local complete intersection @xmath21 with ideal sheaf @xmath22 . if the determinant bundle of the normal bundle of @xmath21 extends to a bundle on @xmath10 : @xmath23 then there is a rank two bundle @xmath20 on @xmath10 with section @xmath24 , which induces the short exact sequence : @xmath25 the chern classes of @xmath20 are given by @xmath26 and @xmath27 . if @xmath17 is a local complete intersection and the bundle @xmath20 exists , then the bundles @xmath15 and @xmath20 are related by the exact sequence : @xmath28 let char @xmath29 and let @xmath30 be a reduced polynomial ; the role of @xmath13 will be played by the jacobian ideal of @xmath31 . for the remainder of the paper
we restrict our attention to the case where @xmath31 is a product of distinct linear forms , although many of the results can be generalized .
we begin with some facts about hyperplane arrangements ; for more information see orlik and terao @xcite .
a hyperplane arrangement @xmath32 is a finite collection of codimension one linear subspaces of a fixed vector space v. @xmath32 is _ central _ if each hyperplane contains the origin * 0 * of v. the intersection lattice @xmath33 of @xmath32 consists of the intersections of the elements of @xmath32 ; the rank of @xmath34 is simply the codimension of @xmath35 .
v is the lattice element @xmath36 ; the rank one elements are the hyperplanes themselves .
@xmath32 is called _ essential _ if rank @xmath37 dim @xmath4 .
henceforth , @xmath2 will be an _ essential _ , _ central _ three arrangement with @xmath38 ; i.e. a set of @xmath39 lines in @xmath40 .
the mbius function @xmath41 : @xmath42 is defined by @xmath43 the poincar polynomial @xmath44 it follows directly from the definitions above that for such an arrangement , @xmath45 let @xmath31 be a reduced polynomial defining @xmath2 and @xmath46 the jacobian ideal of @xmath31 .
the next lemma gives an easy proof of the main result of @xcite : [ lem : lci ] the jacobian ideal of a line arrangement in @xmath3 is a local complete intersection .
localization , the product rule , and euler s relation .
[ thm : compute1 ] if @xmath32 is a line arrangement defined by @xmath31 and @xmath15 is the syzygy bundle of @xmath46 , then @xmath47 where @xmath48 is the chern polynomial .
lemma [ lem : cclass ] implies that @xmath49 and by lemma [ lem : lci ] we have @xmath50 now use the identity : @xmath51 the motivation for the previous theorem is terao s celebrated freeness theorem @xcite .
let @xmath0 be an arrangement with defining polynomial @xmath52 $ ] ; the module of derivations tangent to @xmath2 is defined as : @xmath53 terao s theorem is that if @xmath54 is free , then the poincar polynomial factors as @xmath55 , where @xmath56 are the degrees of a set of homogeneous generators of @xmath54 .
if char @xmath29 then @xmath57 , where @xmath58 is generated by the euler derivation and @xmath59 is the module of syzygies on @xmath46 . henceforth , @xmath15 will be the sheaf associated to @xmath59 .
a triple of arrangements @xmath60 consists of an arrangement @xmath2 and choice of distinguished hyperplane @xmath61 such that @xmath62 and @xmath63 .
@xmath64 is called the deletion of @xmath2 and @xmath65 the restriction of @xmath2 with respect to @xmath66 .
of course , the invariants of the elements of a triple are closely related , and deletion - restriction is often a valuable tool for inductive proofs . for the module of derivations , we have : [ prop : ot]@xmath67proposition 4.45 of @xcite@xmath68 there is an exact sequence : @xmath69 orlik and terao give an example of a line arrangement for which the above sequence is not right exact ( see example 4.56 of @xcite - this corresponds to example i in the next section , with the role of @xmath66 played by @xmath70 ) .
recall the definition of an elementary modification ( see @xcite ) : let @xmath71 be a smooth variety , @xmath21 an effective divisor on @xmath71 , @xmath72 .
let @xmath4 be a rank two bundle on @xmath71 , @xmath5 a line bundle on @xmath21 , and suppose @xmath73 .
then the kernel @xmath7 of the map is also a rank two bundle on @xmath71 , with @xmath74 and @xmath75 .
[ thm : elem ] let @xmath2 be an arrangement of lines in @xmath3 . if @xmath76 is a triple , and @xmath77 , then the sequence of sheaves corresponding to proposition [ prop : ot ] @xmath67with euler derivations pruned off@xmath68 is an elementary modification , i.e. @xmath78 is exact .
this follows since @xmath79 now use the hirzebruch - riemann - roch theorem and serre vanishing to convert the chern polynomials to hilbert polynomials , and compute .
an important measure of the complexity of a coherent sheaf on @xmath10 is the _ castelnuovo - mumford regularity _ : a coherent sheaf @xmath20 on @xmath10 is @xmath80-regular @xmath81 if @xmath82 the exact sequence @xmath83 gives us a good way to bound the castelnuovo - mumford regularity of line arrangements .
[ thm : cmreg ] for a triple @xmath60 of line arrangements , @xmath84 follows from the long exact sequence in cohomology . for an arrangement on @xmath39 lines , @xmath85
this is tight for generic arrangements .
since the sheaf @xmath15 is reflexive , it corresponds to a bundle on @xmath3 and ( @xcite ) @xmath86 for an arrangement of @xmath39 lines , this means that the minimal free resolution of @xmath59 is : @xmath87 where the @xmath88 are at most @xmath89 and the @xmath90 are at most @xmath91 . in @xcite
, ziegler gives bounds on the degrees of generators for @xmath92 , which gives bounds on the generators of @xmath59 . for hypersurfaces with only isolated singularities , choudary and dimca @xcite
give a bound ; for a ( reduced , singular ) degree @xmath39 curve in @xmath3 , the regularity of @xmath59 is at most @xmath93 .
thus , for line arrangements , the bound above is better than existing results .
the entire free resolution of @xmath59 for a _ generic _ arrangement is given in @xcite and @xcite .
in @xcite , yuzvinsky gives a set of generators for a submodule of @xmath92 ; these generators are determined by @xmath94 .
when the third relation space vanishes , they actually generate the entire module ( this generalizes the results of ziegler mentioned earlier ) . even in the case of line configurations , there are examples ( e.g. the braid arrangement ) where this space does not vanish . however , the maximal number of generators of @xmath59 is bounded by @xmath91 ( which is attained by generic arrangements ) . for one proof of this ,
see jiang and feng @xcite , 4.2 .
finally , we note that derksen and sidman @xcite have recently obtained regularity bounds on @xmath1 for higher dimensional arrangements .
in this section , we consider the stability and jump loci of the bundle @xmath15 obtained from an arrangement of @xmath39 lines ; the point is that we can often construct stable bundles with prescribed jumping lines . for generic arrangements
these questions were studied by dolgachev and kapranov in @xcite,@xcite . we want to investigate @xmath15 when the arrangement is nongeneric ; the tool will be the short exact sequence of the last section .
we first recall a few standard results about vector bundles on @xmath10 , referring for proofs to the book of okonek , schneider and spindler @xcite .
let @xmath95 be a bundle on @xmath10 .
the slope of @xmath95 is defined as @xmath96 a key concept in the study of bundles on @xmath10 is _ stability _ :
a bundle @xmath95 on @xmath10 is stable if for all subsheaves @xmath97 with @xmath98 , @xmath99 and semistable if @xmath100 if @xmath95 is a stable two bundle on @xmath3 , then schwarzenberger @xcite showed that @xmath101 ( this was generalized by bogomolov @xcite ) .
thus , a necessary condition for stability of @xmath15 is that degree @xmath102 .
the degree of the jacobian ideal ranges from @xmath103 ( achieved for generic line arrangements ) to @xmath104 ( achieved for arrangements with @xmath91 lines through a point , and one other line in general position , this class of arrangements has @xmath105 ) . basically , as the degree of the jacobian ideal gets large , @xmath15 has less chance of being stable . for generic arrangements @xmath2 in @xmath10 , dolgachev and kapranov prove that the bundle of meromorphic one forms with logarithmic pole along @xmath2 is stable ( in the setting of generic line arrangements
this bundle is dual to a twist of @xmath15 , see @xcite ) , they also prove that the map which associates to a set of @xmath39 generic hyperplanes the corresponding bundle is generically injective if @xmath106 . since dim @xmath107 this map has no chance of having zariski dense image if the number of lines is large ( if @xmath108 dolgachev and kapranov show that it is ) .
first , a few definitions .
the normalization @xmath109 of a rank two bundle @xmath20 is @xmath110 , where @xmath9 is chosen so that @xmath111 , i.e. let @xmath20 be a rank two bundle on @xmath3 .
then @xmath112 , where @xmath113 if @xmath114 is even , and @xmath115 if @xmath114 is odd .
[ lem : norm ] a reflexive sheaf @xmath20 of rank two over @xmath116 is stable iff @xmath117 . if @xmath114 is even , then @xmath20 is semistable iff @xmath118 . if @xmath114 is odd , semistable and stable coincide .
[ thm : stab ] let @xmath2 be an arrangement of @xmath39 lines in @xmath3 , @xmath66 any line in @xmath2 , and @xmath60 the corresponding triple @xmath67notice that the restriction to @xmath66 ignores multiplicities , so @xmath119
. then 1 .
if @xmath39 is odd , then @xmath15 is stable if @xmath120 is stable and @xmath121 .
2 . if @xmath39 is odd , then @xmath15 is semistable if @xmath120 is semistable and @xmath122 .
if @xmath39 is even , then @xmath15 is stable if @xmath120 is semistable and @xmath123 .
the proofs are similar so we prove the first statement . since @xmath39 is odd , @xmath124 and @xmath125 . by lemma [ lem : norm ] @xmath15 is stable iff @xmath126 so it is necessary that @xmath127 .
the simplest sufficiency criterion is then that @xmath128 ; from the exact sequence @xmath129 we have that @xmath130 iff @xmath121 .
an iff criterion for stability can be formulated in the obvious fashion : @xmath15 is stable iff @xmath131 and the connecting map @xmath132 is an inclusion .
this can be computed in any particular instance , but is not a criterion which is easy to apply ; the point of theorem [ thm : stab ] is that if one adds a line @xmath66 to a configuration @xmath133 for which @xmath120 is stable , then as long as no more than ( roughly ) half of the lines become redundant on restricting to @xmath66 , the new configuration is also stable .
next , we study the jump loci of these bundles . by a theorem of grothendieck , any bundle on @xmath134 splits as a direct sum of line bundles ; in particular , for a rank two bundle @xmath4 on @xmath3 and line @xmath5 @xmath135 in the grassmannian of lines on @xmath136 , there is a nonempty , zariski open subset where @xmath137 is constant ; the complement of this subset is the jump locus @xmath138 of @xmath4 . by semicontinuity and the grauert - mlich theorem , we have for a normalized , semistable two bundle @xmath4 on @xmath3 @xmath139 we now describe the jump lines of @xmath15 .
the method works in general if we have a modification @xmath140 .
in , @xcite , barth proved that if @xmath4 is a semistable two bundle on @xmath10 with @xmath141 , then @xmath138 is purely one codimensional , and is supported on a divisor of degree @xmath142 . if @xmath143 , then in general @xmath4 has only a finite number of jump lines
; in @xcite hulek introduced the notion of a jump line of the second kind : @xmath144 and proved that if @xmath4 is a semistable two bundle on @xmath10 with @xmath145 , then @xmath146 is a curve of degree @xmath147 , and @xmath148 we begin by asking which of the lines of the arrangement are contained in @xmath149 . let @xmath150 be a triple with respect to a line @xmath66 , put @xmath151 .
[ thm : jloci1]let @xmath2 be an arrangement of @xmath39 lines , @xmath152 and @xmath153 .
if @xmath39 is odd , @xmath154 iff either @xmath155 or det @xmath156 ; where @xmath157 if @xmath39 is even , @xmath154 iff either @xmath158 or rank @xmath159 ; where @xmath160 we prove the first statement .
since @xmath161 and @xmath162 , we have an exact sequence : @xmath163 from the long exact sequence in cohomology , it is obvious that @xmath164 iff one of the two conditions of the theorem holds .
consider the following set of arrangements ( where @xmath70 is the line at infinity ) : arrangement i consists of the coordinate lines and the lines @xmath165 ; @xmath166 ( which can be proved using theorem 4.51 of @xcite ) so @xmath15 is semistable .
arrangements ii through v are obtained by adding ( successively ) the lines @xmath167 .
theorem [ thm : stab ] implies that the bundles associated to arrangements ii , iv , and v are stable and the bundle associated to arrangement iii is semistable .
we can compute the chern classes using theorem [ thm : compute1 ] : @xmath168 the normalized bundles fit into exact sequences : @xmath169 an easy application of the beilinson spectral sequence @xcite shows that a stable bundle with @xmath170 and @xmath171 must be @xmath172 , so for arrangement ii , the jump locus of @xmath15 is empty . for arrangement iii
, barth s theorem implies that the support of @xmath149 is of degree one ; by theorem [ thm : jloci1 ] @xmath173 and @xmath174 are jumping lines , hence @xmath149 is the line @xmath175 . for arrangement iv ,
theorem [ thm : jloci1 ] shows that @xmath174 is a jumping line .
in fact ( hulek , prop 8.2 ) a semistable two bundle @xmath4 with @xmath145 , @xmath176 has only a single jump line , so this is it ! finally , for arrangement v , barth s theorem implies that the jump locus is a cubic curve ; theorem [ thm : jloci1 ] gives six lines of the arrangement which are jump lines .
a computation shows that @xmath177 , @xmath178 , and @xmath179 are also jump lines , so we have nine points in @xmath180 , which unfortunately only impose eight conditions on cubics .
however , a final computation shows that the coordinate lines are not jump lines , which allows us to determine that the jump locus is a smooth cubic curve : @xmath181 in @xcite , dolgachev and kapranov show that for generic arrangements in any dimension , a line contained in one of the hyperplanes of the arrangement is a jumping line ; for line arrangements this also follows from theorem [ thm : jloci1 ] .
in fact , dolgachev and kapranov prove that for a generic arrangement with an odd number of lines , the points of the jump locus corresponding to the lines of the arrangement are singular points of fairly high multiplicity , which is an interesting contrast to the example above .
if @xmath182 , then since @xmath183 , we obtain an exact sequence : @xmath184 with @xmath185 torsion . if @xmath39 is odd , normalizing , restricting to @xmath5 , and taking cohomology shows that @xmath186 in fact , we can do better .
let @xmath5 be a line which is not a line of the arrangement .
think of the arrangement as @xmath133 , @xmath187 , and put @xmath188 . restricting to @xmath5 yields a long exact sequence @xmath189 in the previous theorem we studied a short exact sequence pruned from the right end of this type of complex ; pruning a short exact sequence from the left end yields exact sequences:@xmath190 when @xmath39 is odd ; in this case @xmath191 .
@xmath192 when @xmath39 is even ; in this case @xmath193 .
[ thm : jloci2 ] let @xmath2 be an arrangement of @xmath39 lines , @xmath5 a line with @xmath194 and @xmath153 .
if @xmath39 is odd , then @xmath154 iff either @xmath195 or det @xmath156 ; where @xmath196 if @xmath39 is even , then @xmath154 iff either @xmath197 or rank @xmath198 ; where @xmath199 notice that the three jump lines of arrangement @xmath4 which are not lines of the arrangement are characterized by the first condition of the theorem .
it seems reasonable to expect that the jump locus of the second kind is related to _ multiarrangements _ , about which very little is known ( see ziegler @xcite , or solomon - terao @xcite for recent progress ) .
we plan to return to this question in a later paper .
one of the major open conjectures in the study of hyperplane arrangements is the following : in @xcite , yuzvinsky proves that for a fixed intersection lattice the set of free arrangements is zariski open . in this section
, we show that the vector bundle viewpoint has implications for terao s conjecture .
as noted earlier , @xmath1 decomposes as @xmath200 , and the module @xmath59 has a minimal free resolution of the form ( with @xmath201 ) : @xmath202 if there is an @xmath203 , then the arrangement is a `` near pencil '' , and we ignore this case , since such an arrangement is supersolvable @xcite .
the results of 3 imply @xmath204 @xmath205 now fix an arrangement with intersection lattice @xmath206 ( which also fixes @xmath207 ) . for terao s
conjecture we ll be interested in the situation where @xmath208 the results on the jacobian ideal in 2 impose the additional conditions on the resolution : @xmath209 @xmath210 an easy computation shows the condition on @xmath211 is equivalent to @xmath212 there are constraints on any free resolution of the form above ; for example , if @xmath213 then @xmath214 this follows since any relation involves at least two generators , and must be a positive degree multiple of both .
so if a counterexample to terao s conjecture exists , it must be an integral solution of the above inequalities .
another constraint is the following , which follows easily by localization .
a quick check shows that @xmath216 .
this is the simplest example of a nonfree arrangement where @xmath207 factors .
failure of freeness can be explained using the addition - deletion theorem of @xcite ; lemma [ lem : bigmu ] provides another explanation .
we combine the results above into [ thm : constraints ] let @xmath2 be an arrangement on @xmath39 lines with intersection lattice @xmath206 , @xmath217 and @xmath218 . for @xmath219
a unique integral solution to the following inequalities implies that terao s conjecture holds for arrangements with intersection lattice @xmath206 . 1 .
the bound on the number of generators of @xmath59 : @xmath220 2 .
global geometric constraints on the jacobian ideal : @xmath221 @xmath212 3 .
regularity constraints : @xmath204 @xmath205 4 .
resolution constraints : @xmath222 5 . local geometric constraints : @xmath223 it is the smallest free arrangement which is not supersolvable . applying theorem [ thm : constraints ]
, we find that there are nineteen different numerical possibilities for the free resolution of @xmath59 . to shorten this list ,
we employ stability .
write @xmath224 if @xmath2 and @xmath225 have isomorphic intersection lattices , and let @xmath226 , @xmath227 be the bundles associated to arrangements @xmath2 and @xmath225 . call a split two bundle @xmath20 _ balanced _ if @xmath228 ; a rank two bundle which splits is semistable iff it is balanced .
it is obvious that the only free line arrangements which can be semistable are those with an odd number of lines , so for the remainder of this section we ll assume the number of lines is odd .
finally , note that the normalization of a balanced split two bundle is just @xmath229 ; so both chern classes of the normalized bundle are zero .
[ lem : stab ] suppose @xmath226 is balanced , and @xmath230 is a counterexample to terao s conjecture .
then @xmath227 is not semistable .
in particular , there must be a syzygy on the jacobian ideal of @xmath225 of degree @xmath231 .
once we ve fixed a lattice , we have also fixed the chern classes .
thus , a counterexample to terao s conjecture , if semistable , would have a jump locus of degree @xmath232 by barth s theorem , so a semistable counterexample would be a uniform bundle . but
a uniform two bundle on @xmath3 which does not split is @xmath233 ( @xcite ) , and these possibilities are excluded simply by considering the chern classes .
so we know that a counterexample must be unstable . now from lemma
[ lem : norm ] we see that @xmath234 , which implies the result about the syzygies .
[ thm : unstable ] suppose a free arrangement @xmath2 on @xmath39 lines has @xmath239 . a counterexample @xmath240 to terao s conjecture must have a syzygy of degree @xmath231 , and also a syzygy of degree @xmath241 .
[ thm : nonf ] we return to case where @xmath2 is the non - fano arrangement . of the nineteen numerically possible free resolutions for @xmath59 , one corresponds to the case where @xmath2 is free .
theorem [ thm : unstable ] allows us to rule out fifteen of the other possibilities .
thus , there are only three numerical types of resolution possible if @xmath2 is not free : if there are only two quadratic first syzygies , as in the last two cases , there can be at most one linear second syzygy , so these resolutions can not occur .
the first possibility actually does occur as the free resolution for an arrangement on seven lines with poincar polynomial @xmath216 - it is the free resolution for the arrangement in example @xmath245 .
however , it can not be the free resolution for the non - fano arrangement . to see this ,
start out with arrangement @xmath13 from last section , and apply theorem [ thm : cmreg ] twice .
the regularity can be at most four , whereas this resolution has regularity five .
hence , any arrangement combinatorially equivalent to the non - fano arrangement must be free .
remark : theorem 4.51 of @xcite and a similar sequence of subarrangements can also be used to obtain this result .
g. jiang , j. feng , _ an algorithm to produce the structure sequence of an arrangement _ , singularities and complex geometry , q. lu , s.s.t yau , a. libgober , eds .
, ams / ip studies in advanced mathematics , * 5 * ( 1997 ) , 83 - 92 . | associated to a projective arrangement of hyperplanes @xmath0 is the module @xmath1 , which consists of derivations tangent to @xmath2 .
we study @xmath1 when @xmath2 is a configuration of lines in @xmath3 . in this setting
, we relate the deletion / restriction construction used in the study of hyperplane arrangements to elementary modifications of bundles .
this allows us to obtain bounds on the castelnuovo - mumford regularity of @xmath1 .
we also give simple combinatorial conditions for the associated bundle to be stable , and describe its jump lines .
these regularity bounds and stability considerations impose constraints on terao s conjecture
. [ section ] [ defn0]proposition [ defn0]conjecture [ defn0]theorem [ defn0]lemma [ defn0]corollary [ defn0]example = 4pt |
fifty years ago sandage ( 1953 ) discovered a unique class of stars , the blue stragglers ( bss ) . in the color - magnitude diagrams ( cmd ) of clusters ,
these stars appear brighter and bluer than the main sequence turn - off .
they appear in the cmd of open and globular clusters ( gc ) , as well as in the field , and their brightnesses and colors are consistent with evolutionary models for stars more massive than the cluster ( or population ) turn - off .
to explain how stars more massive than the cluster s turn - off stars could still be present in gcs , or in old populations , several scenarios have been proposed .
after years of debate ( for a review , see bailyn 1995 ) , the consensus has settled on stellar collisions or binary mass transfer and mergers as the most likely events responsible for the formation of bs stars .
bs populations have a great diversity of properties .
for instance , their specific frequencies and mean brightnesses vary greatly .
it is likely that these differences are due to differences in the clusters dynamical histories .
however , to date , attempts to correlate bs population photometric properties with other cluster parameters have failed to determine the formation scenarios most likely to apply ( e.g. , piotto et al .
2004 ) . as a result ,
the information the bs population contains about the history of its cluster s dynamics remains inaccessible due to our ignorance of the bs formation channels .
critical bs properties which have not been fully investigated due to the difficulty of the observations are their masses and rotation rates . whether the bss are the result of a collision or a binary merger , they are expected to be more massive than the main sequence turn - off mass .
however , models of the two scenarios do not predict bss which are systematically different , with the exception that masses larger than twice the turn - off mass for the cluster are likely to be the result of 3-body collisions .
the fact that bss are more centrally - concentrated than subgiants , and , by inference , than to stars , already suggested that bss are more massive than the clusters turn - off mass ( nemec & harris 1987 ) .
spectroscopic evidence that bs masses are larger than the turn - off mass for the cluster was previously obtained by shara et al .
( 1997 ) , who measured the mass of one of the most luminous bs in the core of 47 tuc , showing that it has twice the cluster s turn - off mass .
this determination was also confirmed by ouellette ( 2000 ) in a distance - independent way , by comparing stellar temperatures and gravities with evolutionary calculations .
saffer et al . ( 2002 ) carried out a spectroscopic analysis on 6 bss in ngc 6397 , also deriving masses higher than the turn - off mass for 5 out of 6 objects .
masses larger than the cluster turn - off were also reported for 4 bss by gilliland et al .
( 1998 ) , based on their pulsation frequency measurements .
however the distribution of the bs masses remains an open question since the samples analyzed were invariably small . since mass predictions do not yet allow us to differentiate between formation scenarios , one can hope that determining masses for a large enough sample will reveal correlations with other stellar and cluster properties .
rotation rates for stars formed either through collisions or binary mergers are also expected to be higher than normal .
simulations of collision mergers ( e.g. , sills et al . 2002 , 2005 ) ,
show merger product to be rapidly rotating .
simulations of main sequence binary mergers are lacking , but it is likely that these systems also inherit considerable angular momentum .
however , rotation rates measured for bss in open cluster m67 appear to be _ lower _ than those for comparable main - sequence stars ( mathys 1991 ) .
shara et al .
( 1997 ) determined the stellar rotation for the 47 tuc bs bss19 to be a high @xmath2 = 155 km s@xmath3 .
theoretically , rapidly - rotating bss are expected to slow down : sills et al .
( 2005 ) argue that a magnetically - locked accretion disk or wind can draw angular momentum from the star and slow it down .
however , matt & pudritz ( 2005 ) have recently argued that this might not be the case .
de marco et al . ( 2004 ) suggested that low mass disks might be present around some bss , but porter & townsend ( 2005 ) showed that the same data can also be interpreted by accounting for the distortion of the stellar shape due to rotation .
it is clear that a great deal of work remains to be done both observationally and theoretically . with all this in mind
, we obtained hubble space telescope ( hst ) spectroscopic observations of a sample of bss drawn from four different clusters with different central concentration ( 47 tucanae [ 47 tuc ; ngc 104 ] , m 3 [ ngc5272 ] , ngc 6752 and ngc 6397 ) , to determine spectroscopic masses and rotation rates . in section [ sec : observations ] we describe our hst observations .
the details of the spectroscopic data reduction and extraction are given in section [ sec : datareduction ] . in section [ sec : photometry ] we present the stellar photometry , while in section [ sec : identification ] we use the photometry to classify the analyzed stars . in section
[ sec : stellarparameters ] we derive the stellar parameters , followed by an assessment of the uncertainties in section [ sec : uncertainties ] .
stellar masses and rotation rates are presented in sections [ sec : masses ] and [ sec : rotation ] , respectively . in section
[ sec : disk ] we describe the spectroscopic indicators of circumstellar disks or rapid rotation , while in section [ sec : individualobjects ] individual stars are discussed . a comparison with stellar evolutionary models
is presented in section [ sec : evolutionarytracks ] , followed by our summary in section [ sec : conclusions ] .
in order to obtain a sizable sample of bss , we selected three clusters , rich in bss : 47 tuc ( ngc 104 ) , m 3 ( ngc 5272 ) and ngc 6752 .
we also re - analyzed data for the cluster ngc 6397 , in which 6 bss and one hb star were previously identified and analyzed by saffer et al .
these clusters span a range in metallicity and central star density .
details of the clusters are presented in table [ tab : clusterproperties ] .
the cmds of these clusters are presented in figures [ fig : cmd_ngc104_bv ] to [ fig : cmd_ngc6397_uv ] , where the bss can be clearly seen at positions brighter and bluer than the main sequence turn - off .
spectra of stars in the gcs 47 tuc , m 3 and ngc 6752 were acquired through hst proposal go-8226 ( pi : shara ) , using the space telescope imaging spectrograph ( stis ) between april 1999 and may 2000 , with a few frames acquired in march 2001 .
low ( grating g430l ) and intermediate ( grating g430 m ) resolution spectroscopy was obtained using the [email protected] arcsec@xmath6 slit .
this setup afforded 5.5- resolution spectra in the range 3200 - 5600 and 0.56-resolution spectra in the range 3800 - 4100 .
the slit width was chosen to allow some uncertainty in the target acquisition , in view of the fact that more than one target was to be observed per slit .
several slit positions were used for each cluster .
each slit position was chosen so as to maximize the number of bss that fell in the slit .
hst / fos ( faint object spectrograph ) spectroscopy of 6 bss and one hb star in ngc 6397 ( proposal go-6697 ; pi : de marchi ) was acquired on october 5 , 1996 using the g400h ( wavelength range 32404822 ) and g570h ( wavelength range 45746872 ) gratings with the 0.5-arcsec aperture .
this setup afforded 2.82 and 4.09- resolution for the g400h and g570h gratings , respectively ( using the nominal 3.07 and 4.45 per diode for the g400h and g570h gratings , respectively , and the size of the post - costar full width at half maximum [ fwhm ] of 0.92 diodes for the 0.5-arcsec aperture ) . in table [ tab : observations ] we present a log of the observations .
hst / stis image names are listed in column 1 followed , in column 2 , by our own stellar identification names .
ra & dec ( column 3 ) and pixel positions ( column 4 ) were determined on the images displayed in figs .
[ fig:47tuc_image ] to [ fig : ngc6397_image ] .
the slit positions ( column 5 ) are the y - pixel positions of the stellar spectra on the stis images . in column 6
we list distances from the respective clusters centers .
finally , point - spread - function ( psf ) photometry carried out on wide field and planetary camera 2 ( wfpc2 ) images is presented in columns 7 to 10 ( described in detail in section [ sec : photometry ] ) .
the entire go-8226 and go-6697 datasets were downloaded from the hst archive at the canadian astronomy data center ( cadc ) in march 2001 , with on - the - fly " calibration and processed ( sepinsky et al .
2002 ) to reduce the number of cosmic ray hits which were not eliminated by the automatic pipeline calibration . despite the 0.043-arcsec per pixel spatial resolution of the stis / hst , spectroscopy of gc stars
is affected by severe blending , so that spectral extraction is a real challenge and a source of uncertainty .
the size of the extracting aperture is a compromise between minimizing contamination from other stars ( width of the aperture @xmath75 pixels ) and insuring the full sampling of the spatial psf ( width of the aperture between 8 and 12 pixels ) .
the psf shape is wavelength dependent , with more and more stellar flux residing in the psf s wings for longer wavelengths .
because of this , small apertures tend to extract spectra which are systematically bluer ( less flux falls within the aperture at longer wavelengths ) . in fig .
[ fig : disk_spatialcut ] we show spatial cuts of the stis frames containing four stars from two clusters taken at three wavelength positions , corresponding to the central wavelengths of the johnson @xmath8 , @xmath9 and @xmath10 filter band passes .
the limits of the apertures used to extract the 4 stars , as well as the adopted background level are also shown . in three out of four cases
the star is blended with a very faint neighbor which is not bright enough , nor close enough to contribute substantial flux . in the fourth case , the neighbor is brighter , but again , not close enough to constitute a problem . in fig .
[ fig : aperture_test ] we present star n5272 - 11 , extracted using three aperture widths : 2 pixels , 4 pixels and 8 pixels .
as already described , the flux in the smaller two apertures is bluer and somewhat distorted .
the large 8-pixel aperture allows flux from one faint neighbor to contribute .
it was found that apertures larger than 8 pixels increase contamination but the shape of the spectrum remains unchanged .
hence , 8-pixel apertures are large enough to not incur the psf problem , but small enough to not incur severe blending .
overall , we maintain that we have carried out the best reduction and extraction that can be carried out with these data .
a better reduction and extraction demands the use of software that can deconvolve blended spectra . during the analysis of the data ,
great attention has been put into checking the individual extractions before attributing an uncertainty to the derived parameters .
we have carefully tracked the uncertainties and propagated them into our results ( section [ sec : uncertainties ] and table [ tab : results ] ) .
in order to determine spectroscopic masses one has to scale the stellar atmosphere model fluxes to observed ( de - reddened ) stellar fluxes .
if the spectra are photometric , one can rely on the spectroscopy alone .
this is not the case for spectra acquired in crowded regions , or when there is a risk that the star is not centered on the slit .
for this reason we scaled our models ( section [ sec : stellarparameters ] ) not to the spectral fluxes but to stellar photometry from archival wfpc2 images .
we did not expect bss to show significant variability , but some did .
this is in itself an important result of this work .
with hindsight , one should always take photometric observations within a short time of the spectroscopic ones , since parameters derived from spectra taken at a particular variability phase should not be mixed with the brightness derived from photometry taken at a different phase .
variable stars will be discussed in sections [ sec : identification ] and [ sec : uncertainties ] .
psf photometry was carried out on a subsample of all the available archival hst / wfpc2 images of the four clusters .
a photometric reduction using all the available images will be undertaken elsewhere .
the data ( table [ tab : photometry ] ) were obtained from the mast database with on - the - fly " calibrations .
the photometric reductions were performed using the daophot / allframe software ( stetson 1987 ) .
briefly , allframe determines the brightness of each star on each frame with point - spread - function fitting at the same time .
this reduction process increases the accuracy of the photometry by fixing the position of each star across all the frames which increases the accuracy of de - blending the stars .
we calibrated the photometry to the vegamag system ( wfpc2 data handbook ) in order to use the fewest corrections when scaling the atmosphere models ( section [ sec : stellarparameters ] ) to the observations .
the uncertainty of the photometric magnitudes is estimated to be 0.02 mag for the @xmath11 and @xmath12 filters , 0.03 mag for the @xmath13 filter and 0.04 mag for the @xmath14 one .
a calibration to any standard " system ( e.g. , johnson ) includes many additional corrections because the wfpc2 filters are not strictly standard " . to convert the wfpc2 magnitudes listed in table [ tab : observations ] to the johnson system
, one can use the transformations of dolphin ( 2000a , b ) .
the observed stis / vega @xmath14 , @xmath13 , @xmath11 and @xmath12 magnitudes obtained from averaging _ all _ images from table [ tab : photometry ] are listed in table [ tab : observations ] ( columns 7 to 10 ) .
stellar atmosphere models were scaled to the @xmath11photometric magnitudes to obtain radii , luminosities and spectroscopic masses . for variable stars ,
it is not appropriate to obtain radii by scaling models to photometric magnitudes obtained at a different time from the spectra . for variable stars we determine radii , luminosities and masses by scaling the models to the spectra . in section [ sec : uncertainties ] we will discuss in detail the uncertainties in this method .
the analyzed stars were identified on wfpc2 images by using the iraf task _ siapier _ to position the stis slit on the image and match stars along the slit to spectra in the stis ccd images . for the fos observations we relied on the coordinates provided in the image header . in figs .
[ fig:47tuc_image ] to [ fig : ngc6397_image ] we present portions of @xmath11 images containing the stars thus identified , which we label according to the scheme presented in table [ tab : observations ] .
ra ( to the nearest tenth of second ) and dec ( to the nearest tenth of arc - second ) , as well as pixel coordinates ( to the nearest pixel ) for all analyzed stars were read off the wfpc2 images displayed using ds9 ( joye & mandel 2003 ) and are also listed in table [ tab : observations ] .
all stars in the instability strip are susceptible to stellar pulsations .
variable hb stars ( rr lyrae ) have by far the largest brightness amplitudes .
they are typically in the temperature range @xmath1560007300 k and have periods between 0.4 and 1 day .
variable pulsating main sequence stars fall into the @xmath16 scuti or sx phoenicis classes .
these tend to pulsate with smaller amplitudes ( @xmath150.1 mag ) , although they can occasionally vary more ( 0.4 mag ) and they have periods between 0.04 and 0.2 days .
the typical temperatures for main sequence pulsations are marginally hotter than those of hb stars , and are in the range @xmath157000 - 8200 k ( from allen s astrophysical quantities , 2000 ) .
stars can obviously also vary if they are in eclipsing binary systems . to identify which ones of our stars are variable , we adopted four methods .
( i ) stars that vary by more than 0.2 mag over multi - epoch same - filter wfpc2 images ( stars n104 - 8 , n5272 - 8 , 9 , 12 , 14 , 15 , n6397 - 5 and 6 in table [ tab : results ] ) .
this method does not detect all variables , since it depends on when the individual frames were acquired .
( ii ) stars whose colors clearly vary when combining @xmath14 , @xmath13and @xmath11 magnitudes from images taken at different dates ( stars n104 - 1 , 3 , 6 , n5272 - 8 , 9 , 12 , 14 , 15 , 18 , 20 , n6752 - 17 , n6397 - 1 , 2 , 5 , 6 , and 7 ) .
( iii ) stars whose low- and intermediate - resolution spectra have different flux levels ( stars n104 - 8 ,
n5272 - 13 ) and , finally , ( iv ) stars whose spectral and photometric fluxes and colors are very different ( stars n104 - 2 , n5272 - 13 , n6752 - 5 , 9 , n6397 - 1 , and 2 ) , but the difference is inconsistent with other causes ( explained in detail in section [ ssec : errvar ] ) . stars which satisfied any of these four criteria are considered variable and are marked v " in table [ tab : results ] and discussed further in section [ sec : uncertainties ] . the core of 47 tuc was surveyed for variables by edmonds et al .
none of our 47 tuc stars matches the positions of the variables found in that study , despite the fact that , based on the criteria listed above , five of the 9 stars are variable .
( we note in passing that the positions quoted by edmonds et al . [ 1996 ; their table 2 ] are with reference to star e in guhathakurta et al .
1992 , but the position of that star quoted in that paper [ their table 3 : 00@xmath1721@[email protected] 72@xmath2021@[email protected] [ b1950.0 ] which maps to 00@xmath1724@[email protected] 72@xmath2004@[email protected] [ j2000.0 ] ] does not correspond to any star on the j2000.0 coordinate system of our ccd frame [ e.g. , j6ll01yiq ] . this is not unexpected due to the positional uncertainty of the dss1 system .
we therefore re - downloaded their dataset to find that there is a coordinate shift .
their star e corresponds to a bright star in j2000.0 position 00@xmath1724@[email protected] 72@xmath2004@[email protected] . ) for 47 tuc and ngc 6397 we constructed @xmath13 @xmath11 vs. @xmath11 cmds from images acquired on september 1 , 1999 and march 6 - 7 , 2001 , respectively ( figs.[fig : cmd_ngc104_bv ] and [ fig : cmd_ngc6397_bv ] ) . for m 3
we constructed a @xmath11@xmath12 vs. @xmath11 cmd from observations acquired on may 14 , 1998 ( fig.[fig : cmd_ngc5272_vi ] ) .
for ngc 6752 we constructed a @xmath14@xmath11 vs. @xmath11 cmd from observations acquired on march 22 , 2001 ( fig.[fig : cmd_ngc6752_uv ] ) .
these same - epoch " cmds were constructed to minimize the variability and it is on them that we classified our stars as bs , hb ( ehb for extreme hb ) , and turn - off ( to ; ato for above turn - off " ; column 3 in table [ tab : results ] ) .
star n104 - 3 has no clear classification based on its cmd position and we mark it with a ? " in table [ tab : results ] . to confirm the classifications we also constructed cmds with different color combinations using all the available photometry ( table [ tab : photometry ] ; figs.[fig : cmd_ngc104_uv ] , [ fig : cmd_ngc5272_uv ] , [ fig : cmd_ngc6752_bv ] and [ fig : cmd_ngc6397_uv ] ) .
cluster membership could not be determined using the radial velocity of the absorption lines of the stars in our sample , because we can not derive a sufficiently precise wavelength calibration from our spectra .
the reason for this is that the stis spectra are taken through the wide 0.5-arcsec slit and a star s wavelength is a function of its exact position in slit .
we can , however , assess that foreground contamination is close to zero
. from the total number of stars in the guide star catalogue ( version ii ; 170,000,000 to a limiting photographic j magnitude of 19.5 [ @xmath23 ) we can estimate that there is @xmath151 star arcmin@xmath24 on average in the sky .
all our stars are two or more magnitudes brighter .
furthermore , close to the high galactic latitudes characteristic of our four gcs ( 28@xmath20@xmath2572@xmath20 ) there are often zero .
following the referee s suggestion , we also used the calculations by robin et al .
( 1996 ) to estimate the number of foreground stars in the direction of the four gcs . using the gcs galactic coordinates , a distance smaller than 10 kpc , and restricting the stellar types to the apparent @xmath10 magnitude and color ranges of the stars in our list .
the simulated number of foreground stars is zero for solid angles comparable to the angles subtended by the gc cores ( typically 1 arcmin@xmath6 ) . even enlarging the solid angle to 10 deg@xmath6
the number of predicted stars with the required characteristics was only about 10 for 47 tuc and m 3 and zero for the other two clusters .
we therefore conclude that , although we can not be certain that every one of our stars belong to its cluster , it is very unlikely that even 1 star per cluster is a foreground contaminant .
all the selected spectra were de - reddened using @xmath26 values for their home clusters obtained from the literature ( @xmath26 = 0.032 , 0.016 , 0.056 and 0.18 mag for 47 tuc , m 3 , ngc 6752 and ngc6397 , respectively ; see table [ tab : clusterproperties ] ) and fitted by synthetic spectra convolved with gaussian profiles to mimic the instrumental resolution ( fwhm=5.46 and 0.56 , for the stis g430l and g430 m grating , respectively [ 2-pixel resolution ] , and 2.82 and 4.09 for the g400h and g570h fos gratings , respectively ) .
the model spectra were calculated with the hubeny & lanz ( 1995 ) spectrum synthesis code synspec , based on local thermodynamic equilibrium ( lte ) as well as on non - lte ( nlte ) model atmospheres .
we started by calculating a grid of synthetic spectra for the parameter range , 5000 @xmath27 25000 k , @xmath28 ( lowest gravities only for the coolest models ) , and metallicities , [ fe / h][email protected] ( for 47 tuc ; vandenberg 2000 ) , @xmath291.5 ( for m 3 and ngc 6752 ) , and @xmath291.97 ( for ngc 6397 ; harris [ 1996 ] quotes 1.57 , 1.56 and 1.95 , for m 3 , ngc 6752 and ngc 6397 , respectively , similar to the values used ) , using lte model atmospheres calculated with kurucz atlas9 program ( kurucz 1993 ) . to achieve this
, we selected the `` odfnew '' grids of model atmospheres with [ fe / h]=@xmath291 , @xmath291.5 , and @xmath292 from the kurucz web site and used them to calculate detailed spectra with the synspec code , scaling solar abundances for all elements heavier than helium to the three selected metallicities .
the helium and @xmath30-elements abundances were y = he / h=0.24 ( mass fraction ) , and [ @xmath30/h]=0.3 for all clusters , following the prescription of bergbusch & vandenberg ( 2001 ) .
we can not measure micro - turbulence from our data , so we adopted a value of 2 km s@xmath3 which is a good compromise from previous studies of f and g stars ( e.g. , edvardsson et al . 1993 ) . in figs .
[ fig : col - col1 ] and [ fig : col - col2 ] , we show mono - chromatic color - color diagrams including colors for all the lte synthetic spectra as well as the de - reddened stars ( table [ tab : col - col ] ) . we have defined new , narrow - band magnitudes which we call [ 3600 ] , [ 4200 ] , and [ 5450 ] , which are constructed by averaging the theoretical and the observed fluxes in 20-wide spectral bands centered at 3600 , 4200 , and 5450 ( these are the johnson bandpass centers , except for [ 4200 ] for which we used 4200 instead of 4400 , thereby avoiding @xmath314388 , a line which is sensitive to the adopted helium abundance in the hottest models ) , and converting them to magnitudes using the same zero - points than johnson ubv magnitudes ( bessel et al . 1998 ) : @xmath32 = -2.5 \log f_{3600}-21.0 - 0.77\ ] ] @xmath33 = -2.5 \log f_{4200}-21.0 - 0.12\ ] ] @xmath34 = -2.5 \log f_{5450}-21.0\ ] ] compared to the more familiar johnson colors , these quasi - monochromatic colors have the advantage that they are more sensitive to the continuum spectrum shape from which we derive the stellar parameters , while the large breadth of the johnson bandpasses include many additional spectral lines that might be poorly matched by these initial models ( for instance , because of inadequate abundances of some chemical species ) .
we used these diagrams to derive initial parameters for all stars .
most stars in our sample have colors consistent with the models , that is , they are located within the model locii
. there are , however , a few stars whose colors do not match any of the models . these stars will be discussed in detail in section [ sec : disk ] . for each star , we then recalculated a lte atlas9 model atmosphere and the detailed emergent spectrum using the stellar parameters derived from a comparison of stellar and model colors . if the synthetic spectrum did not match the observations to a high degree of accuracy , we calculated new models until agreement was reached ( often mismatches between model and observed colors were due to low snr in the faintest stars ) .
the fitting process was carried out by eye , using the paschen continuum slope , the balmer jump , and the balmer line wings from the higher resolution spectroscopy .
note that their dependence on the basic parameters changes from hot to cool stars ( as seen in the color - color diagrams , figs .
[ fig : col - col1 ] and [ fig : col - col2 ] ) .
accordingly , the accuracy of the derived parameters varies across the parameter space , and this is reflected in the attributed uncertainties ( see table [ tab : results ] and section [ sec : uncertainties ] ) . finally , we computed nlte model atmospheres with the code tlusty , version 198 ( hubeny & lanz 1995 ) .
tlusty assumes a static atmosphere , plane - parallel geometry , hydrostatic and radiative equilibria .
tlusty allows for explicit departures from lte for an extensive , user - selected set of species and levels .
the present nlte model atmospheres explicitly incorporate hydrogen , helium , carbon , and iron .
table [ tab : atom ] lists the detail of the atomic makeup of the model atmospheres . for fe i and ii we have 38 and 35 super - levels , respectively .
iron line blanketing is treated using opacity distribution functions ( hubeny & lanz 1995 ) .
additional sources of opacity include h@xmath35 and h@xmath36 .
the micro - turbulent velocity was maintained at 2 km s@xmath3 , and convection was neglected .
the abundance mix is chosen as for the lte models .
for the hotter models , we found that the predicted lte and nlte model spectra are quite similar , indicating that departures from lte remain limited .
the essential nlte effects are small changes in the predicted fluxes and in the @xmath37 colors , i.e. , in the amplitude of the balmer jump . for a calculation with t@xmath38=9000 k and @xmath39=3.3 the lte model is 3% brighter at @xmath10 but 6% brighter at @xmath8 , making the balmer jump slightly shallower .
this effect is alleviated for hotter calculations : for a t@xmath38=18000 k and @xmath39=4.3 the lte atmosphere is 2% brighter at @xmath10 and 5% brighter at @xmath8 .
the different treatment of opacities is likely the cause of the different @xmath10 fluxes , while departures from lte are more likely the cause of the relative changes in the balmer jump .
these changes translate into slightly altered values of the radii and masses obtained when using lte and nlte atmospheres and , while they are smaller than the random uncertainty ( section [ sec : uncertainties ] ) the use of lte , rather than nlte , stellar atmosphere models would have introduced a _
systematic _ source of error .
we should point out at this stage that tlusty is a model atmosphere code originally designed to model hot stellar atmospheres .
we found that model convergence was often hard to achieve for cool models ( @xmath40 @xmath41 8000 k ) with the current version .
the basic physical reason behind this difficulty is that hydrogen is partly ionized in these atmospheres , resulting in a strong , highly non - linear coupling between radiative equilibrium and the hydrogen equilibrium ( h , h@xmath42 , h@xmath36 , and h@xmath35 ) .
spurious results might be derived from models that are not truly converged .
for instance , a poor conservation of the total flux would yield an incorrect temperature stratification and result in an erroneous predicted emergent spectrum .
we exercised special care when converging nlte cool models , carrying out additional iterations and changing the default limit between the integral and the differential form of the radiative equilibrium equation ( see hubeny & lanz 1995 ) to ensure a satisfactory conservation of the total flux . after obtaining @xmath39 and @xmath40 from the spectral fits ( columns 4 and 5 in table [ tab : results ] ) ,
the stellar radii were obtained by scaling each model s flux ( @xmath43 ) to the de - reddened @xmath11 psf photometric magnitudes or , for variable stars , to the spectroscopic fluxes at 5550 ( @xmath44 ) . from the scaling factor
, the radius can be obtained using : @xmath45 where @xmath44 , @xmath43 , @xmath46 and @xmath47 are the observed ( de - reddened ) and theoretical stellar fluxes , the stellar radius and the distance from earth , respectively . from the radii and effective temperatures , the stellar luminosities ( @xmath48 )
were obtained from the boltzmann formula : @xmath49 where @xmath50 is the stefan - boltzmann constant , while from the radii and gravities , we obtained the stellar spectroscopic masses : @xmath51 where @xmath52 is the gravitational constant .
radii , luminosities and spectroscopic masses thus obtained are listed in columns 6 to 8 of table [ tab : results ] .
while the metallicity of the stars is kept at the adopted values , the strength of the theoretical he i lines is clearly too large to reproduce these lines in the high resolution spectra of stars with @xmath53 12000 k and was therefore reduced to y = he / h=0.014 ( by mass ) .
this behavior has been previously reported by , e.g. , behr et al .
( 1999 ) . except for the stars discussed in sections [ sec : disk ] and [ sec : individualobjects ]
, model fits to all of the analyzed stars are excellent .
four such fits ( one per cluster ) are presented in figs .
[ fig : ngc104_spectrum ] , [ fig : ngc5272_spectrum ] , [ fig : ngc6752_spectrum ] and [ fig : ngc6397_spectrum ] . fits to all other stars are available only electronically .
in order to assess the errors on the derived spectroscopic masses , we must carefully consider all of the sources of uncertainty .
first , there are _ random _ errors on @xmath39 and @xmath0 , which derive from the sensitivity of the synthetic spectra to changes in these two quantities as well as the snr of the data .
second , we have _ random _ errors on the factors to scale the models to the stellar photometry ( or spectroscopy for variable stars ) .
these depend on the uncertainty in the photometry and the snr of the spectroscopy . for variable stars , model fluxes
are not scaled to the photometry , but to the spectroscopy .
these stars are therefore not corrected for _ systematic _ errors due to stellar blending and slit losses .
stellar blending also adds an _ additional _ source of systematic uncertainty to the derived effective temperatures . in what follows
we assess all sources of error and propagate the uncertainties into our derived quantities .
errors on @xmath39 and @xmath0 are due to sensitivity of the synthetic spectra to these quantities as well as the snr of the data . at low @xmath0 ,
i.e. , @xmath77000 k , balmer line widths as well as the balmer jump are not very sensitive to gravity and the value of @xmath39 can not be obtained to an accuracy better than @xmath150.10 dex , but more typically 0.15 dex ( or worse in some cases ; see table [ tab : results ] ) .
it is important to emphasize that while there will be a minimum @xmath54 value for which the fit is best , the formal error of the procedure is an underestimate of the true uncertainty .
the effective temperature is determined from the slope of the paschen continuum ( 4200 - 5600 ) and its uncertainty is typically @xmath15100 k , with some higher values ( table [ tab : results ] ) . for higher effective temperatures , it is possible to obtain more accurate values of gravity ( 0.10 dex typically ) , while the temperature estimates remain similarly constrained .
the sensitivity of @xmath39 and @xmath0 to abundance changes as large as [ fe / h]=0.2 dex is very low .
we therefore conclude that our adopting a constant cluster metallicity has negligible repercussion on the determination of @xmath39 and @xmath0 .
the uncertainty on the ratio @xmath44/@xmath43 ( eq . 1 ) is a combination of the uncertainty on the @xmath11 band photometry of 0.02 mag i(or the spectrophotometry for variable stars 0.15 mag ) and the uncertainty on the synthetic model fluxes . the latter is dominated by the uncertainty in the temperature . for a 100 k change in @xmath0 , the model flux changes by 7.5% at 6000 k , and by 1.3% at 13000 k ( or @xmath55 for @xmath0@xmath76500 k , @xmath56 for 6500@xmath7@xmath0@xmath710000 k and @xmath57 for @xmath0@xmath5810000 k ) . for every star
, we therefore derived the absolute error on @xmath43 ( @xmath59 ) from the @xmath0 value , the absolute error on @xmath0 ( @xmath60@xmath0 ) and the value of @xmath61 . from the error on the @xmath44/@xmath43 ratio and a 10% error on the distances ( listed in table [ tab : clusterproperties ] )
( this was chosen following the recent assessment by gratton et al .
2003 ) , we derive an error on the radii by uncorrelated error propagation . to determine the error on the stellar luminosities , however , we need to consider that the errors on @xmath46 and @xmath0 are correlated , because , as explained above , the uncertainty on @xmath0 conditions the uncertainty on @xmath43 and hence on @xmath46 .
we therefore must use the error propagation formula for correlated errors : @xmath62 the cross - term is due to the fact that the errors on @xmath46 and @xmath0 are not independent .
@xmath63 is negative since @xmath46 and @xmath0 are inversely proportional .
we set its value at 0.5 which is intermediate between zero ( no error correlation ) and 1 ( maximum correlation ) .
the uncertainty on @xmath46 and @xmath64 propagates into the mass according to the formula for uncorrelated uncertainty . in the absence of stellar photometry , our determinations of radii
, luminosities and masses would be affected by spectral blending ( which would make the spectra brighter and alter their color ) and slit losses ( which would make the spectra fainter , but not alter their color ) . for non variable stars we scaled the model spectra to photometric fluxes , thereby removing the uncertainty on the brightness , though not on the color , due to blending and slit losses .
variable stars , which are not scaled to the photometry , are not corrected in this way . to determine the impact of stellar blending and slit losses on our dataset , all observed stellar spectra were convolved with the wfpc2 filter profiles ( the iraf routine _ calcphot _ was used with the vega system ;
the latest vega magnitudes were used : _ stsdas _ version 3.3 ) and the spectro - photometric magnitudes thus obtained were compared with the photometric values from table [ tab : observations ] .
( since the red end of the observed spectra is slightly blue - ward of the red end of the f555w bandpass , we actually used the synthetic spectra scaled to the observed spectra at 5540 , to obtain the f555w spectrophotometric magnitudes . )
the difference between the photometric and spectrophotometric @xmath11 magnitudes for all stars is listed in table [ tab : results ] , both as flux ratios ( column 9 ) and as magnitude differences ( column 10 ) . in columns 11 and 12
we list the differences between spectrophotometric and photometric colors , @xmath16(@xmath13@xmath11 ) and @xmath16(@xmath14@xmath11 ) , which we will use to assess additional uncertainty affecting our choice of temperatures .
magnitudes derived from spectra can be brighter or dimmer than the corresponding photometry and the spectral colors can be bluer or redder than the photometric ones .
brighter spectra with bluer , redder , or similar colors than the photometry can be the result of spectral blending .
brighter spectra with the same colors as the photometry can be due to an eclipsing binary with a faint companion .
dimmer spectra with bluer or redder colors than the photometry can be due to pulsational variability .
dimmer spectra with the same color , probably mean that the star was on the edge of the slit . in what follows we will take variable stars out of the discussion , as for these , spectroscopy and photometry was not taken at the same time , and the the values of temperature and gravity obtained from the spectra can not be used together with the apparent brightness obtained from the photometry .
variable stars will be treated later .
although by scaling to the photometric fluxes we remove the source of uncertainty due to blending or slit losses it is instructive to compare photometry and spectrophotometry . in the absence of any systematic effect , f555w magnitude and color differences are expected to be distributed around zero with a standard deviation of the order of the random errors .
the random error on the magnitude differences is 0.15 mag , determined by adding in quadrature the 0.02 mag error on the @xmath11photometry and an average error of 0.15 mag on the spectrophotometry , due to the snr of the spectra .
the random errors on the color differences are 0.22 mag for @xmath65 and 0.25 mag for @xmath66 , and were calculated by adding in quadrature the uncertainties on the @xmath11 and @xmath13 ( @xmath14 ) photometry and the expected average errors on the equivalent spectrophotometric magnitudes of 15% in @xmath13 and @xmath11 ( 20% in @xmath14 ) estimated from the snr of the data . in fig .
[ fig : scaling_hist ] we plot a histogram of all the @xmath11 magnitude differences for the 34 non - variable stars listed in table [ tab : results ] ( column 10 ) .
the mean and standard deviation are 0.18 and 0.29 mag , respectively ( with a sigma clipping of 2 ) .
the mean of the color differences ( columns 11 and 12 in table [ tab : results ] ) are @xmath67=0.02 mag and @xmath68 = 0.14 mag .
if photometry and spectro - photometry were only affected by random errors we would expect the mean of the histogram in fig .
[ fig : scaling_hist ] to be zero , with a standard deviation of the order of the random error expected on the values ( 0.15 mag ) .
the fact that the mean is less than zero means either that the predominant systematic uncertainty is stellar blending ( which makes the spectra brighter ) or that there is a systematic difference in the methods used to convert flux values and @xmath11 bandpass counts to magnitudes .
( the statistics are not very different if the comparisons are carried out on stis and fos datasets separately , from which we conclude that the reason for the shift is not in the spectroscopic instruments , but rather in the calibrations carried out in the photometry and spectrophotometry . )
the means of the color difference columns @xmath16(@xmath13@xmath11 ) and @xmath16(@xmath14@xmath11 ) are not the same ( 0.02 and 0.14 mag , respectively ) .
this suggests that there is an offset in the transformations used to convert fluxes to magnitudes in the photometry and/or in the spectrophotometry , which introduces a color term .
this offset might therefore also be the reason for the fact that the histogram in fig .
[ fig : scaling_hist ] does not peak at zero .
the standard deviation of the magnitude differences ( 0.29 mag ) is about twice as large as the random error ( 0.15 mag ) , reinforcing the fact that systematic as well as random uncertainties are at play . to assess which stars are most affected by blending or slit losses
, we selected all stars with magnitude differences outside the range [email protected] mag and whose color differences were outside the ranges [email protected] mag ( [email protected] mag for @xmath16(@xmath14@xmath11 ) ) .
a total of 16 stars out of 34 non - variable stars have magnitude differences outside the range . of these ,
six stars have magnitude differences @xmath7(0.180.15 ) mag , indicative of blending and we mark them with : " . of these 6 stars ,
one ( n104 - 5 ) has a color difference ( @xmath16(@xmath13 @xmath11 ) ) outside the range expected from random error alone , and we mark it with : : " and raise its @xmath0 uncertainty to 300 k since , although this star s blended flux was corrected by scaling the model to the photometry , its color was affected as well .
ten stars have magnitude differences @xmath58(0.18 + 0.15 ) mag , indicative of slit losses .
although slit losses are not expected to alter the color of the star we note that two of these ten stars ( n5272 - 10 and n6752 - 10 ) have color differences outside the range expected for random errors alone and they too are given a larger temperature uncertainty and marked : : " in table [ tab : results ] .
six stars ( n104 - 2 , n5272 - 13 , n6752 - 5 , 9 , n6397 - 1 , and 2 ) have magnitude differences consistent with random uncertainties , but _
differences outside the range expected from random uncertainties alone .
this is likely to be due to variability , as discussed in section [ sec : identification ] ( variability detection criterion [ iv ] ) . for the stars known to be variable ( variability detection criteria [ i - iv ] in section [ sec : identification ] ) , we calculated their radii , luminosities and spectroscopic masses scaling the models to the spectra instead of the photometry , since , as explained , scaling to a photometric flux determined at a different time from the spectroscopy is inappropriate .
however , since we concluded earlier in this section , that there is a 0.18 mag offset between photometry and spectrophotometry , ( corresponding to a factor of 0.80 ) , _ due to calibration _ , scaling models to spectra will lead to masses that are 20% larger on average .
we therefore account for this offset in the derivation of the radii , luminosities and spectroscopic masses of the variable stars , by multiplying these quantities by 0.80 . in the assessment of the uncertainty of a spectroscopic analysis such as ours
it is also instructive to compare results with those of other similar analyses .
here we compare our results for ngc 6397 with those obtained by saffer et al .
( 2002 ) .
the gravity values they obtained from an lte analysis similar to the one reported here are either in agreement ( for two out of seven objects ) , 0.1 dex lower ( for three objects ) or 0.2 dex higher ( for the remaining two objects ) .
their effective temperature values are always within 5% of those determined here , except for ngc 6397 - 2 , for which they are 12% higher .
their photometry , which , like in our study , is used to scale the fos spectroscopy , is systematically brighter than ours ( by about 0.7 mag ) .
the masses obtained here agree with those of saffer et al .
( 2002 ) for n6397 - 4 and 7 , while the others are between 0.20 and 0.47 @xmath1 lower , due to our dimmer photometric fluxes .
this reinforces what we will state again in section [ sec : masses ] : the spectroscopic masses are to be trusted only in a statistical sense .
in table [ tab : results ] we list the spectroscopic masses derived in the present analysis , along with their uncertainties . in table [ tab : masses ] we list ( weighted ) mean masses for the four clusters and the three stellar types along with their variance .
the weights are constructed using the masses and their absolute errors : @xmath70 , where @xmath71 are the mass values and @xmath72 are their absolute errors , calculated by averaging the positive and negative error bars listed in table [ tab : results ] .
the variance ( @xmath73 ) gives an idea of the error on the mean masses .
considering the entire non - variable sample , the bss have masses which are marginally larger ( [email protected] @xmath1 ) than the hb masses ( [email protected] @xmath1 ) , but are definitely larger than the to masses ( [email protected] @xmath1 )
. looking at the individual cluster results , it becomes clear that the reason for the similar bs and hb mean masses is an anomaly in m 3 for which the mean ( non - variable ) bs mass ( [email protected] @xmath1 ) is smaller than the mean ( non - variable ) hb mass ( [email protected] @xmath1 ) .
for ngc 6752 , the only other cluster for which the comparison can be done , the opposite is true , as expected .
we later show that it is likely that some of the m 3 hb stars are actually bss and this might explain the high mean masses . from table [ tab : masses ] we can also see that the mean ( non - variable ) spectroscopic mass for ngc 6397 is the largest ( [email protected] @xmath1 ) , followed by ngc 6752 ( [email protected] @xmath1 ) , 47 tuc ( [email protected] @xmath1 ) and m 3 ( [email protected] @xmath1 ) . the mean @xmath11 brightness of our four bs samples ( scaled to the distance of 47 tuc ) are 15.80 , 15.59 , 16.00 and 16.01 mag for ngc 6397 , ngc 6752 , 47 tuc and m 3 , respectively .
masses do scale with bs magnitude ( see below ) so the fact that the bss in ngc 6397 ( and not in ngc 6752 ) are by far the most massive might have something to do with this cluster s high central density ( table [ tab : clusterproperties ] ) .
this might indicate that the gc central concentration plays a role on the bs masses . including variable stars in the means
improves the statistics , although the variable star spectroscopic masses might be affected by worse uncertainty .
the mean bs , hb and to masses are not dissimilar , in particular the ngc 6397 mean bs mass is still the highest ( 1.23 @xmath1 ) even if its mean scaled @xmath11 brightness ( 15.50 mag ) is the same as that of ngc 6752 ( 15.51 mag ; the mean bs @xmath11 brightnesses of m 3 and 47 tuc are both 15.90 mag and their mean masses are 1.03 and 0.99 @xmath1 , respectively ) . in fig .
[ fig : teff_vs_mass ] we plot the spectroscopic masses for all 55 stars as a function of effective temperature ( left ) and scaled @xmath11 brightness ( right ) .
the mean error on the masses is indicated on the top - left panel .
the latter plot gives a measure of the relative brightness of the stars once all stars are scaled to the distance of 47 tuc .
as is expected , the bss masses correlate with their effective temperature and brightness although the scatter is large .
a pearson @xmath63 test returns a correlation factor of 0.28 ( 1.0 being perfect correlation ) for effective temperature vs. mass and 0.22 , for brightness vs. mass ( where the negative sign results from large numbers expressing faint brightnesses ) .
if stars n6752 - 9 and n6397 - 4 are eliminated from the @xmath0mass plot the correlation index becomes 0.59 , indicating a much better correlation .
these two stars have mass uncertainties of 0.39 and 0.24 @xmath1 , respectively , hence their low masses , which do not follow the trend , are unlikely to be due to a particularly large error . the to star masses also correlate with @xmath0 ( 0.17 ) and with @xmath11 ( 0.30 ) .
the to star spectroscopic masses sit at the low end of the bs mass range , as expected .
some to stars have masses larger than the turn - off mass , in particular for 47 tuc .
hb star masses are not expected to show a marked correlation with @xmath0 , and they are not predicted to have masses larger than the turn - off mass .
their masses are expected to be around 0.6 @xmath1(see figure 9 in moehler et al .
surprisingly , we find that 4 hb stars have masses significantly larger than the turn - off mass ( n5272 - 14 , 17 , n6752 - 5 and n6397 - 7 ) . on the @xmath39- @xmath0 plane
only one of these ( n6397 - 7 ) is consistent with the hb locus , one is just below it ( n5272 - 17 ) , while two are well below it ( n5272 - 14 and n6752 - 5 ) , in the region populated by the bss .
looking at the cmds in figs .
[ fig : cmd_ngc104_bv ] to [ fig : cmd_ngc6397_uv ] , n5272 - 14 and n5272 - 17 have colors which are also consistent with being bright bss , although on the @xmath14@xmath11 vs. @xmath11 plane n5272 - 14 appears in the rr - lyrae region .
n6752 - 5 and n6397 - 7 appear on the hb of their clusters , although , once again , they could also be bright bss .
parts of the hb and bs loci , in fact , overlap .
this is also the case for the @xmath39- @xmath0 plane where evolutionary track for @xmath152.5 @xmath1 grazes the hb locus .
to measure stellar rotation spectroscopically requires high resolution spectroscopy of metal lines .
it is inappropriate to rely on fits to the troughs of the balmer lines to measure stellar rotation , except to impose broad limits .
small changes in the model parameters result in significant differences in the troughs of the balmer lines .
in addition , the troughs of balmer lines are affected by poor snr more than the continuum .
relying on a a handful of data points around these regions to determine stellar rotation will likely lead to erroneous results .
although the resolution , snr and spectral range of our spectroscopy are not ideal for determining stellar rotation , we can impose some upper limits and determine some values , using our stis intermediate resolution spectra and the following technique . the best - fitting stellar atmosphere model spectrum ( pre - convolved with the instrumental profile ) ,
is convolved with gaussian profiles of increasing fwhm and then with poisson noise so as to reproduce the continuum snr of the observations .
each of these model spectra is then compared to the data .
for some stars a model is found for which the fit is clearly worse .
this will provide an upper limit to the stellar rotation above which the model does not fit the data . for other stars , convolving models to broaden the lines beyond the instrumental effect improves the fit , and a best value of @xmath2 can therefore be determined .
three types of spectral lines present in our spectra can be used : weak iron and other trace metal lines , the ca ii k line ( or h and k for those cooler stars where h@xmath74 is weak compared to ca ii h ) and , finally , the balmer line wings .
weak iron and other metal lines can be used for spectra with low effective temperature and reasonable snr .
however these lines are never resolved , hence only their troughs can be used .
these are affected by noise as well as by the uncertainty on the abundance and temperature .
some spectra , like those of n104 - 6 and n104 - 7 have models that reproduce well most of these metal lines . in such cases convolving the model with a gaussian with
fwhm@xmath58100 km s@xmath3 clearly deteriorates the fits . in these cases
, we can argue that @xmath2 is below this value . on the other hand ,
if metal lines are not well fit anyway ( whether with or without rotation convolution , e.g. n104 - 5 ) , it is inappropriate to use them to determine a limit for the stellar rotation . the ca ii k line ( or h and k lines ) can also be used .
this line is best at intermediate temperatures , since for high temperatures it is too weak and unresolved , and at low temperatures it is very broad and similar to the balmer lines ; for intermediate temperatures its wings and , to an extent even its trough , can be used .
this line , too , can be affected by abundance and temperature uncertainties .
the balmer lines troughs , as already mentioned , are not reliable to measure rotation , but can still be used to impose broad limits . finally , for stars hotter than about 10000 k , he i lines can be used .
these are ideal since they are always resolved and their shapes vary differently with abundance and rotation . in fig .
[ fig : rotation ] we present the intermediate resolution spectra of n5272 - 11 , one of the disk stars described in section [ sec : disk ] . the high snr of this spectrum and its weak , yet resolved ca ii k line makes the determination of the rotation upper limit more precise : ( 70@xmath75 ) km s@xmath3 .
the wings of the line , which are less affected by small temperature and abundance differences , tell us that the fit is already satisfactory with no rotation , however the wings can also tolerate a rotation of 70 km s@xmath3and in fact the trough is better fit in this way .
higher values of the rotation make the fit progressively worse until at 110 km s@xmath3 both wings and trough no longer fit . in fig .
[ fig : disk ] we present a similar plot for the ehb star n5272 - 17 .
this is the same plot as in de marco et al .
it shows that the atmosphere model fits the ca ii k line wings only when convolved with a gaussian with fwhm=200 km s@xmath3 . in table
[ tab : rotation ] we present values ( and limits ) of @xmath2 obtained with this method .
these results are the most extensive ever set of measurements and limits for bs rotation velocities .
five bss ( n104 - 4 , n5272 - 13 and 15 , n6752 - 11 and 18 ) have reasonably well measured @xmath2 ( 120@xmath76 , 100@xmath6920 , 225@xmath6950 , 50@xmath6920 , 50@xmath6920 km s@xmath3 , respectively ) .
the sample s median and average @xmath2 are 100 and 109 km s@xmath3 , respectively .
five other bss ( n104 - 7 and 8 , n6752 - 4 , 9 and 14 ) have only upper limits to their @xmath77 ( 120 , 120 , 100 , 25 and 50 km s@xmath3 , respectively ) . for a randomly oriented set of rotating stars , the average value of @xmath78 is 0.64 .
thus the five bss with measured @xmath2 suggest an average rotation velocity of @xmath79 km s@xmath3 . from values of radii and masses of these stars
, we can calculate their breakup speeds to be between 250 and 350 km s@xmath3(with the exception of star n5272 - 15 which , with a very high mass of 3.74 @xmath1 , has a breakup speed of @xmath15600 km s@xmath3 .
this star s mass is discussed in section [ sec : individualobjects ] ) . even taking the mean @xmath2 at face value , it is hard to tell whether this value is more in agreement with bss deriving from a collision or a binary merger .
both scenarios predict that the resulting star should have a large angular momentum , and it should rotate very fast , especially when it starts to contract following the post - formation expansion . even if one scenario predicted a slower rotation rate than the other , one would still have to contend with the various mechanisms that are likely to slow the initially - fast - rotating star down .
the often - cited paper by leonard and livio ( 1995 ) predicts that collision mergers puff up enough to develop a convective zone that would slow the star down , while binary mergers coalesce less violently and expand less , such that no convective layer is expected and can not therefore lose angular momentum efficiently .
however this is in contradiction with the recent calculations of sills et al .
( 2005 ) , who predict that the product of an off - axis collision does not develop a convective layer so the leonard and livio mechanism can not operate .
sills et al .
( 2005 ) do however show that a disk , or even a magnetically locked wind can slow the star down , therefore providing an alternative slow - down mechanism for collision mergers .
it appears that there is quite a bit of work still to be done in modeling bss .
on the other hand these models need to be guided by observations and the only way to make headway observationally is to carry out an analysis similar to the one carried out here on a very large dataset spanning several gcs as well as open clusters and the field .
in particular , field bss are fundamental since they are highly unlikely to derive from collisions .
field bss should therefore display characteristics which should make them stand aside from the rest of the bss .
higher resolution and snr spectra should be obtained , however , since only then can rotation rates be determined in a meaningful way .
we should also note that all six @xmath2 values for hb stars ( all of which are ehb stars ) , are @xmath8050 km s@xmath3 , including one at 200@xmath6950 km s@xmath3 .
values these high are not common for hb stars ( e.g. , behr 2003 ) .
field ehb stars are called sub - dwarf o and b stars .
their rotation rates are by and large found to have small values ( @xmath2@xmath75 km s@xmath3 ; napiwotzki 2001 ) as is the case for the ehb stars .
recently , lanz et al . ( 2004 ) and ahmad et al . (
2004 ) have found values of the order of 100 km s@xmath3 for sdob stars .
the latter study has determined that their target sdb stars is actually a short - period double - lined spectroscopic binary , and that this is the reason for the broad absorption lines and the large derived values of @xmath2 .
this , they argue , could also explain the high rotation rates found by lanz et al .
it is therefore possible that this is the explanation for our rapidly - rotating ehb stars .
if this were the case a more careful look should be given to these stars as their parameters would be affected by the companion .
two of the six ehb stars with large rotation rates are among the four objects which are suspected of being bss . in section
[ sec : disk ] we return to the issue of rapidly - rotating ehb and hb stars .
de marco et al . ( 2004 ) showed that for 6 stars out of our sample ( marked in table [ tab : results ] ) no reasonable fit can be achieved . in all cases
the model that fits the continuum redward of the balmer jump ( the paschen continuum ) has a corresponding continuum blue - ward of the balmer jump ( the balmer continuum ) which is too bright .
the discrepancy was evident at the 5@xmath50 level and could not be reconciled with any change in model parameters . on the color - color diagram of figs .
[ fig : col - col1 ] and [ fig : col - col2 ] the six exceptional stars fall leftward of the region occupied by the models . for these stars
the [ 3600]-[4200 ] color ( the slope of the paschen continuum ) is too blue ( hot ) for the size of their balmer jump , as parametrized by the [ 4200]-[5450 ] color . here
we identify two additional stars , ngc5272 - 11 and 16 , where the same discrepancy , though to a lesser extent , is identified .
de marco et al . ( 2004 ) excluded a purely instrumental problem as four of the stars in their group were observed with the fos while the others were observed with the stis .
extraction problems that can affect the shape of the spectrum such as background subtraction , tracing and the aperture size were investigated in great detail , but were not responsible for the appearance of these spectra ( see also section [ sec : datareduction ] ) .
an incorrect cluster reddening assumption does not solve the problem either .
blending or binarity was also excluded as the cause of the discrepancy since these effects would push stars to the upper side of the color - color diagram .
no two - temperature model , such as a disk or equatorial enhancement , can account for the discrepancy . as a result of the impossibility of reconciling the data with the models , de marco et al .
( 2004 ) conjectured that a disk of partly ionized gas is responsible for the absorption of balmer continuum photons .
absorption models demonstrated that the column densities of material needed to reconcile the data with the stellar models are quite modest , and that making reasonable assumptions for the possible radii of these disks ( @xmath150.1 au ) , their masses are quite low , of the order of 2@xmath510@xmath81 m@xmath82 .
we refer the reader to that paper for further details and to our conclusions in section [ sec : conclusions ] .
a closer inspection revealed that two additional stars might have similar disks , although the hydrogen column densities should be lower than for the six stars in de marco et al .
these two additional stars are hb stars , as are two of the six stars in de marco et al .
in section [ sec : masses ] we have already noted that some hb stars , including two of the six disk stars of de marco et al .
( 2004 ) are likely to be bright bss . however , the two additional hb stars with possible disks detected here appear to be bona fide hb stars ( with low masses , and locations on the cmd and the @xmath39@xmath0 plane consistent with hb stars ) .
it is therefore unclear at this point what the presence of disks might mean in the context of bs and other gc stars .
recently , porter and townsend ( 2005 ) modeled the anomalous balmer jump of n5272 - 17 and showed that they can reproduce the data with a rotating star whose shape is made oblate by the centrifugal force .
the rotation rates needed for the effect to explain the data depend on whether the star is viewed pole- or equator - on , but there is a solution for the values of @xmath2 determined for n5272 - 17 and n6752 - 11 ( table [ tab : rotation ] ) .
their balmer line fits are not of the highest quality , but they ascribe that to not having optimized the overall values of temperature and gravity ( they took the stellar parameters of de marco et al . [
2004 ] as their starting values ) .
there is however a remaining issue . in fig .
[ fig : disk ] we show the ca ii k line region of our intermediate resolution spectrum of n5272 - 17 .
this is the same figure as on the upper right - hand panel in fig . 2 of de marco et al .
the ca ii k line shows three components , a broad , shallow one , interpreted as the stellar , rotationally - broadened component , a narrow one at the same wavelength , due to the putative disk , and a narrow one shifted by the gc s heliocentric velocity and interpreted as arising from the interstellar medium .
if disks are not present , it might be hard to explain the narrow stellar ca ii k line component .
on the other hand the snr of the data is not high enough to resolve the issue conclusively .
if porter & townsend ( 2005 ) are right , and stellar rotation is solely responsible for the anomalous balmer jumps , then this confirms that four hb stars in our sample are fast rotators .
two ( n5272 - 17 and n6397 - 7 ) are , as we already explained , likely to be bright bss , but the remaining two , ( n5272 - 11 and 16 ) are likely to be bona fide hb stars .
fast - rotating hb stars would be an interesting finding in itself , since rotation this high might indicate a binary interaction in the star s past , leading us to wonder the extent to which binarity influences the evolution of the horizontal branch .
in this section we report the highlights of a subset of our objects which presented particular challenges during the fitting process or which show other peculiarities . _
n104 - 3 : _ this star is variable and hard to classify without photometric observations taken at the same time .
nonetheless it is likely to be a bs or a to star , rather than an hb star . on the basis of its spectrum alone it appears to be a bs .
its mass , obtained by scaling the model to the spectrum is however only 0.42 @xmath1 . _
n104 - 5 : _ in this high snr spectrum , all of the metal lines are overestimated by the model .
this is also the case for the ca ii k line which is broader in the model than in the observations . by convolving the model with a gaussian profile with fwhm=80 km
s@xmath3 the overall fit to the metal lines is slightly improved .
the ca ii k line fit , however , remains poor .
rotation does not appear to be responsible for the discrepancy between the intermediate resolution spectra and the model .
rather , we conclude that there is a temperature inconsistency between the intermediate resolution spectrum , where a higher model temperature ( by about 300 k ) would achieve a better fit , and the low resolution spectrum , where a higher temperature would misrepresent the paschen continuum ( gravity has little effect in this temperature regime ) .
a lower abundance would help to fit better the intermediate resolution spectrum , although the abundance change needed would be too large and inconsistent with the cluster abundance . _
n104 - 8 : _ the continuum level of the intermediate resolution spectrum for this bs is about 20% brighter than for the low resolution spectrum .
no reason was found from the data reduction to explain the discrepancy , since the flux levels of the low and intermediate resolution spectra of stars on the same slit are consistent .
this might point to an inherent photometric variability of n104 - 8 .
this star was marked as having a high probability of being a variable on the basis of the photometry described in section [ sec : photometry ] and [ ssec : errvar ] . as a bs
, this star might be a @xmath16 scuti or a sx phoenicis pulsating variable , or it might be an eclipsing binary
. we also found that the balmer lines of the low resolution spectrum are narrower than for the intermediate resolution one . _
ngc104 - 9 : _ the spectrum of this bs suffers from the same problem as n104 - 5 , although to a lesser extent . here
too we decided that improving the fit to the metal lines by including rotation is not the correct interpretation of the mismatch between data and model . _
n5272 - 6 : _ this ehb star must be a slight blend of two stars , since it is impossible to fit perfectly the balmer continuum once the paschen continuum is well fit .
since the gravity can be determined by the balmer line widths in its high - snr intermediate resolution spectrum , we have retained this spectrum in the sample and assigned a larger error to the temperature . _
n5272 - 13 : _ as is the case for ngc104 - 8 , this bs s intermediate resolution spectrum is brighter than the low resolution spectrum by 7% . _
n5272 - 15 : _ this bs star has the largest spectroscopic mass encountered ( [email protected]@xmath1 ) .
it was identified as a variable because its magnitude determined from individual images varied by just above 0.20 mag . as a result
we did not scale the stellar atmosphere model flux ( @xmath43 ) to the photometry , but used instead the spectroscopic flux scaled by 0.8 as explained in section [ sec : uncertainties ] .
one could suppose that the spectrum was blended with that of another star , making @xmath44 artificially bright , and the mass artificially large .
we note however , that even a blend with a star of equal brightness would only make the mass twice as large .
such a blend is excluded on the basis of the imaging , where we see only one star close to n5272 - 15 which has only @xmath1525% of its brightness and its flux was partly - to - mostly excluded from the spectral extraction of n5272 - 15 by the extraction procedure .
a blend with a very bright star is also highly improbable since the most likely candidate would be a red giant and this would be obvious in the spectrum .
the only blend that could feasibly have augmented the stellar flux is one with another bs or an hb star almost coincident in position with n5272 - 15 .
alternatively this star is a main sequence foreground star . with a smaller value of the distance @xmath47
, its radius and spectroscopic mass would also be smaller . as explained in section [
sec : identification ] we do not expect our sample of hot stars to be contaminated , but without radial velocity measurements , the possibility can not be ruled out . _
n6752 - 2 : _ it is likely that a small amount of contamination affects this hb star . when its paschen continuum is fitted and the gravity constrained by the intermediate resolution spectrum , the balmer continuum is a little too bright to be fitted by the model .
this is the reason why in the color - color diagram in fig .
[ fig : col - col2 ] the star has the colors of a 17000-k object , while the balmer continuum is better fit with 18000 k. since we do not know whether the contaminating star is bluer or redder , we adopt an intermediate temperature and increase the uncertainty . _
n6397 - 4 : _ this bs has a peculiarly low mass ( [email protected] @xmath1 ) .
its high effective temperature ( @xmath0=13000 k ) and gravity ( @xmath39= 4.2 ) position it on the hb locus ( fig .
[ fig : loggvsteff_n6397 ] ) .
its position on the cmd ( fig .
[ fig : cmd_ngc6397_bv ] ) is slightly blue - ward of the zams .
we therefore wonder whether this star could actually be an ehb star .
stellar evolutionary models were calculated for a range of main sequence masses for each of the assumed cluster metallicities ( listed in table 1 ) .
in addition to the assumed metallicities , an alpha - element over - abundance relative to iron of [ @xmath30/fe]=+0.30 was assumed ( carney 1996 ) .
the opacities and equation of state used in the calculations were those described in vandenberg et al .
evolutionary tracks and isochrones were created using these models and the techniques described in bergbusch & vandenberg ( 1992 ) .
for the three metal - poor clusters ( m 3 , ngc 6752 and ngc 6397 ) an absolute age of 14 gyr was assumed , whereas an age 12 gyr was assumed for the more metal - rich 47 tuc : these ages were accepted based on the agreement of the shape of the isochrones and the shape of the cluster fiducial on the sub - giant branch , turn - off , and upper main - sequence . while assuming an age for a cluster based on morphological considerations is problematic
due to uncertainties in abundances , opacities , and color - temperature transformations
the exact cluster age is not critical to the results of this study . in fig .
[ fig : cmd_ngc104_bv ] , [ fig : cmd_ngc5272_vi ] , [ fig : cmd_ngc6752_bv ] and [ fig : cmd_ngc6397_bv ] , we over - plot theoretical isochrones , zero - age main sequence ( zams ) locii and evolutionary tracks on the @xmath13@xmath11 vs. @xmath11 , @xmath11@xmath12 vs. @xmath11 and @xmath14@xmath11 vs. @xmath11cmds of the four clusters .
as previously observed , the bss tend to reside redward of the zmas .
ouellette & pritchet ( 1999 ) and ouellette ( 2000 ) simulated different synthetic bs populations using a range of assumptions for their creation scenarios . by comparing the position of these synthetic populations to the real populations in several cluster they deduced the most likely bs formation scenario .
they concluded that a 60:40 mix of bss formed through collisions involving a turn - off star and a lower mass main sequence star and bss formed by equal - mass collisions was responsible for the bss in 47 tuc , while an 80:20 mix of the same two populations was responsible for the bss of ngc 6397 .
they also concluded that the bss are not unevolved stars , in that their position on the cmd is well above the position of the zams .
sills et al .
( 2000 ) carried out similar simulations .
they reported that the bs stars in 47 tuc must have formed in a burst that stopped several giga - years ago since in this gc bright bss are lacking .
the zams , theoretical evolutionary tracks for several initial masses , and the theoretical locus of hb stars ( from the calculations of vandenberg et al .
2000 ) , are plotted in figs .
[ fig : loggvsteff_47tuc ] to [ fig : loggvsteff_n6397 ] .
as is the case for the @xmath13@xmath11 vs @xmath11 plane , most of the bss lie above the zams , in a region of lower gravity .
this is unlikely to be due to high rotation rates since not even the largest rotation rates possible ( the largest breakup speed for these stars is between @xmath15250 and 350 km s@xmath3 ) would lower the gravities of these stars by the needed amount . in table
[ tab : rotation ] we were able to mostly exclude rotation rates that high .
the conclusion of shara et al .
( 1997 ) and saffer et al .
( 2001 ) that the stars analyzed ( which were re - analyzed in the present work ) fall on the zams was due to their use of an incorrect zams , located at gravities far smaller than the gravities appropriate for the metallicities in their clusters .
all of the bss analyzed in this study have gravities lower than expected for main sequence stars , in agreement with their positions on their clusters cmds , which is well above the main sequence .
we find in fact that most of the bss reside _ well above _ not only the zams but the _ terminal - age _ main sequence as well , and are in the region of the hertzsprung gap .
stars formed by collision or merger binary scenarios are expected to be puffed - up but should contract back to the main sequence on one or two thermal timescales , of order 5@xmath510@xmath83 yr .
the main sequence lifetimes of these stars should be much longer .
this is a major puzzle at the heart of the bs phenomenon .
there is a mild correlation between the spectroscopic masses and the masses deduced from a comparison with evolutionary tracks , in that the to stars cluster near tracks of lower mass compared to the bss .
overall , the masses that would be derived from a comparison with the evolutionary tracks would be slightly larger than those determined from the spectroscopy .
ouellette ( 2000 ) found this to be the case also for 5 stars in ngc 6397 analyzed spectroscopically by rodgers & roberts ( 1995 ) , who found masses in the range 0.62 - 1.15 @xmath1 .
the masses obtained via a comparison with evolutionary tracks are in the range 1.03 - 1.15 @xmath1 , although the mass estimates from these two methods agreed in the case of bss-19 ( @xmath84 @xmath1 ) in 47 tuc , which was analyzed spectroscopically by shara et al .
( 1997 ) . some hb stars in m 3 and ngc 6752 fall above the hb locus in the @xmath39 and @xmath0 plane ( figs .
[ fig : loggvsteff_m3 ] and [ fig : loggvsteff_n6752 ] ) .
moehler et al . ( 2003 ) find that hb stars with temperatures above 12000 have gravities lower than predicted , similarly to our hb stars in ngc 6752 .
moehler et al . (
2003 ) do not reach a satisfactory explanation for the discrepancy , although they suggest it might have something to do with their using _ scaled - solar _ model atmospheres ( i.e. , atmospheres calculated with solar abundances but whose level populations were scaled to the cluster metallicity before the spectrum was computed ) .
we , however , use model atmospheres calculated with the correct abundances , hence can not use their suggestion to explain our discrepancy . at the high temperature end ,
the gravities of terminal - age hb stars are considerably lower than those of zero - age hb stars ( dorman 1993 ) , possibly presenting us with a solution for the low - gravities determined for our hot hb stars .
our cool hb stars n5272 - 8 , 9 and 12 have gravities well below the model prediction . at these temperatures zero - age and terminal - age locii
occupy virtually the same position in the @xmath39-@xmath0 plane , which means that our determined low gravities are not readily explained .
the fact that these three stars are variable rr lyrae stars does not readily present us with an explanation for their low gravities . as already discussed in section [ sec : masses ] ,
hb stars n5272 - 14 and n6752 - 5 have gravities which are distinctly higher than the hb locus and more consistent with high mass bss .
their masses are well above the turn - off value ( 1.01 and 1.57 @xmath1 , respectively ) and their cmd positions , although not canonically that of bss , are consistent with high mass bss .
we briefly summarize our results as follows : 1 .
bs spectroscopic masses are larger than the turn - off mass for their cluster and are larger for the bss than for either the to or hb stars .
four out of 13 hb stars have masses significantly larger than the main sequence turn - off for their home clusters .
these stars could be high - luminosity bs .
bs spectroscopic masses correlate both with their effective temperatures and luminosities , although the scatter is large .
2 . a comparison with evolutionary models leads to larger bs masses than the spectroscopic masses obtained via a stellar atmosphere analysis .
3 . _ all
_ bss have @xmath39 values lower than expected for ms stars .
their positions on the cmd , as well as @xmath39@xmath0 diagram is consistent with stars on the hertzsprung gap .
this is inconsistent with evolutionary timescales expected for stars out of thermodynamic equilibrium .
we emphasize that this is a major discrepancy with current theory that must be resolved by any successful theory of bs stars .
five bss have mean @xmath2 of 100 km s@xmath3 , suggesting a mean value for the bs rotation rate of @xmath15160 km s@xmath3 .
this is below the breakup speed for these stars , but rapidly rotating nonetheless .
de marco et al . ( 2004 ) suggested that 4 bss and 2 ehb stars have spectra consistent with low - mass circumstellar disks .
two of the disk stars " discussed by de marco et al .
are hb stars based on their cmd position .
however , based on their high spectroscopic masses and position on the @xmath39
@xmath0 they might actually be bss .
two additional disk stars " are found here
. however these are both ehb stars based on their cmd locations , @xmath39 @xmath0diagram locations as well as on the ground of their low masses .
the fact that the presence of disks is suggested around hb stars as well as bss might point to a wider role for disks and binarity in gcs .
recently , porter and townsend ( 2005 ) showed that the data are also consistent with a star whose rotation rate makes it oblate .
the rotation rates needed are of the order of those found in this study .
although the shape of the ca ii k line still argues for a disk being present , it is clear at this time that more data are necessary to determine whether disks are present around these stars .
6 . for some ( mostly cool ) hb stars
we determine @xmath39 which are lower than the hb locus .
we do not have a convincing explanation for this .
we are confident , however that this is not a systematic effect that affects the whole sample .
od would like to acknowledge financial support from janet jeppson asimov .
mms acknowledges nasa support via hst grant go-8226 .
data for this paper was obtained at the canadian astronomy data centre , which is operated by the herzberg institute of astrophysics , national research council of canada .
we thank alison sills for useful discussions and the anonymous referee for suggestions that improved the paper .
ahmad , a. , jeffery , c.s .
, & fullerton , a.w . , 2004 , a&a , 418 , 275 allen s astrophysical quantities , 4th edition , arthur n. cox editor , 2000 .
springer - verlag , new york .
bailyn , c.d .
, 1995 , ara&a , 33 , 133 behr , b.b .
, 2003 , apjs , 149 , 67 behr , b.b . , cohen , j.c . ,
mccarty , j.k . , and djorgovski , s.g .
, 1999 , apj letters , 517 , 135 bessell , m.s . , castelli , f. , plez , b. , 1998 , a&a , 333 , 231 bergbusch p.a .
, & vandenberg , d.a . , 1992 ,
apjs , 81 , 163 carney , b. , 1996 , pasp , 108 , 900 davies , m.b . , piotto , g. , & de angeli , f. , 2004 , mnras , in press ( astro - ph/0401502 ) de marco , o. , lanz , t. , ouellette , j.a . ,
zurek , d. , shara , m.m . , 2004 , apjl , dolphin , a.e . ,
2000a , pasp , 112 , 1383 dolphin , a.e .
, 2000b , pasp , 112 , 1397 dorman , b. , rood , r.t . ,
& oconnell , r.w . , 1993 , apj , 419 , 596 edmonds , p.d . , gilliland , r.l . , guhathakurta , p. , petro , l.d . ,
saha , a. , & shara , m.m . , 1996 ,
apj , 468 , 241 edvardsson et al .
1993 , a&a , 275 , 101 ferraro , f.r .
, fusi pecci , f. , bellazzini , m. , 1995 , a&a , 294 , 80 ferraro , f.r . , sills , a. , rood , r.t . ,
paltrinieri , b. , & buonanno , r. , 2003 , apj , 588 , 464 gilliland , r.l . , bono , g , edmonds , p.d . ,
caputo , f. , cassisi , s. , petro , l.d .
, saha , a. shara , m.m . , 1998 , apj , 507 , 818 gratton , r.g . ,
bragaglia , a. , carretta , e. , clementini , g. , desidera , s. , grundahl , f. , & lucatello , s. 2003 , a&a , 408 , 529 guhathakurta , p. , yanni , b. , schneider , d.p . , & bahcall , j.n . , 1992 ,
aj , 104 , 1790 harris , w.e . , 1996 ,
aj , 112 , 1487 hubeny , i. , & lanz , t. , 1995 , apj , 439 , 875 joye , w.a . , & mandel , e. , 2003 , asp conf .
295 , 498 .
kaluzny , j. , krzeminski , w. , mazur , b. , wysocka , a. , stepien , k. , 1997 , acta astronomica , 47 , 249 kurucz , r.l . , 1993 , model atmospheres " , ycat.6039 , origonally published in 1979 , apjs , 40 , 1 lanz , t. brown , t.m . , sweigart , a.v . ,
hubeny , i. , & landsman , w.b . , 2004 ,
apj , 602 , 342 leonard , p.j.t .
& livio , m. , 1995 , apj letters , 447 , 121 mathys , g. , 1991 , a&a , 245 , 467 matt , s. , pudritz , r.e . , 2005 , mnras , 356 , 167 moehler , s. , landsman , w.b . , sweigart , a.v . , grundahl , f. , 2003 , a&a , 405 , 135 napiwotzki , r. , edelmann , h. , heber , u. , karl , c. , drechsel , h. , pauli , e .-
m . , christlieb , n. , 2001 , a&a letters , 378 , 17 nemec , j.m . , & harris , h.c . , 1987 , apj , 316 , 172 ouellette , j.a . , 2000 ,
thesis ( phd ) .
university of victoria ( canada ) .
ouellette , j.a . ,
pritchet , c.j . , 1998 , aj , 115 , 2539 piotto , g. , de angeli , f. , king , i.r . ,
djorgovski , s.g . ,
bono , g. , cassisi , s. , meylan , g. , recio - blanco , a. , rich , r.m . , davies , m.b .
, 2004 , apjl , 604 , 109 porter , j.m . , &
townsend , r.h.d , 2005 , in press ( astro - ph0503437 ) robin , a.c . ,
haywood , m. , creze , m. , ojha , d.k .
, & bienayme , o. , 1996 , a&a , 305 , 125 rodgers , a.w . , & roberts , w.h . , 1995 , aj , 109 , 264 sabbi , e. , ferraro , f.r . , sills , a. , rood , r.t
. , 2004 , apj , 617 , 1296 saffer , r.a . , sepinski , j.f . ,
de marchi , g. , livio , m. , paresce , f. , shara , m. m. , zurek , d. , 2002 , aspc , 263 , 157 sandage , a.r .
, 1953 , aj , 58 , 61 sepinsky , j.f . ,
saffer , r.a .
, shara , m.m . ,
zurek , d. , 2002 , aas , 201 , 07.14 shara , m.m . , saffer , r.a . ,
livio , m. , 1997 , apj letters , 489 , 59 sills , a. , bailyn , c.d . , edmonds , p.d . ,
gilliland , r.l . , 2000 , apj , 535 , 298 sills , a. , adams , t. , davies , m.b . , bate , m.r . , 2002 , mnras , 332 , 49 sills , a. , adams , t. , davies , m.b . , 2005 , mnras , in press .
stetson , p.b .
1987 , pasp , 99 , 191 vandenberg , d.a . , 2000 , apjs ,
129 , 315 vandenberg , d.a . , swenson , f.j .
; rogers , f.j . , iglesias , c.a . ,
alexander , d.r .
, 2000 , apj , 532 , 430 lllll cluster & 47 tucanae & m 3 & ngc 6752 & ngc 6397 + alias & ngc 104 & ngc 5272 & & + ra ( j2000 ) & 00@xmath17 24@xmath18 05.@xmath1919 & 13@xmath17 42@xmath18 11.@xmath1923 & 19@xmath17 10@xmath18 51.@xmath1978 & 17@xmath17 40@xmath18 41.@xmath1936 + dec(j2000 ) & -72@xmath20 04@xmath8549.@xmath869 & + 28@xmath20 22@xmath8531.@xmath866 & -59@xmath20 58@xmath8554.@xmath867 & -53@xmath20 40@xmath8525.@xmath863 + @xmath87 ( mag ) & 13.37 & 15.05 & 13.25 & 12.36 + @xmath26 ( mag ) & 0.032 & 0.016 & 0.056 & 0.18 + @xmath88fe / h@xmath89 $ ] & -0.83 & -1.60 & -1.54 & -1.97 + @xmath90 ( mag ) & -9.42 & -8.93 & -7.73 & -6.63 + rv ( km s@xmath3 ) & -18.7 & -147.6 & -27.9 & 18.9 + c & 2.03 & 1.84 & 2.50c & 2.50c + r@xmath91(arcmin ) & 0.40 & 0.55 & 0.17 & 0.05 + @xmath92(@xmath1 pc@xmath93 ) & 4.81 & 3.51 & 4.91 & 5.68 + turn - off mass ( @xmath1 ) & 0.87 & 0.78 & 0.78 & 0.77 + n(bs ) & 61 & 137 & 28 & 21 + lcclcccccc + o5gx01010 & n104 - 1 & 00@xmath1724@xmath1806.@xmath1978 72@xmath2004@xmath8546.@xmath9454 & 1993 2212 & 581 & 6.64 & 17.508 & 17.77&17.38 & 17.08 + o5gx01010 & n104 - 2 & 00@xmath1724@xmath1806.@xmath1977 72@xmath2004@xmath8548.@xmath9452 & 2032 2207 & 603 & 5.44 & 17.760 & 17.96&17.36 & 16.76 + o5gx01010 & n104 - 3 & 00@xmath1724@xmath1806.@xmath1974 72@xmath2004@xmath8549.@xmath9462 & 2054 2206 & 641 & 5.37 & 17.795 & 17.76&17.71 & 17.81 + o5gx01010 & n104 - 4 & 00@xmath1724@xmath1806.@xmath1968 72@xmath2004@xmath8549.@xmath9490 & 2061 2211 & 648 & 4.88 & 16.806 & 17.07&16.64 & 16.23 + o5gx01010 & n104 - 5 & 00@xmath1724@xmath1806.@xmath1965 72@xmath2004@xmath8551.@xmath9427 & 2088 2209 & 675 & 4.72 & 16.017 & 16.04&15.78 &
+ o5gx02010 & n104 - 6 & 00@xmath1724@xmath1807.@xmath1900 72@xmath2004@xmath8532.@xmath9467 & 1715 2235 & 784 & 19.49 & & 17.78&17.17 & + o5gx03010 & n104 - 7 & 00@xmath1724@xmath1804.@xmath1971 72@xmath2004@xmath8552.@xmath9437 & 2137 2382 & 646 & 4.57 & 16.219 & 16.39&15.96 & + o5gx05010 & n104 - 8 & 00@xmath1724@xmath1806.@xmath1954 72@xmath2004@xmath8534.@xmath9454 & 1759 2251 & 283 & 16.75 & & 15.84&15.50 & + o5gx06010 & n104 - 9 & 00@xmath1724@xmath1805.@xmath1981 72@xmath2004@xmath8551.@xmath9430 & 2100 2285 & 706 & 1.00 & 15.879 & 15.90&15.61 & + + o5gx08010 & n5272 - 1 & 13@xmath1742@xmath1811.@xmath1975 + 28@xmath2022@xmath8546.@xmath9415 & 740 916 & 348&16.78&18.55 & & 18.55 & 17.86 + o5gx08010 & n5272 - 2 & 13@xmath1742@xmath1811.@xmath1957 + 28@xmath2022@xmath8543.@xmath9413 & 702 917 & 424&12.94&18.67 & & 18.64 & 18.00 + o5gx08010 & n5272 - 3 & 13@xmath1742@xmath1811.@xmath1950 + 28@xmath2022@xmath8541.@xmath9481 & 686 918 & 456&10.42&18.61 & & 18.55 & 17.90 + o5gx08010 & n5272 - 4 & 13@xmath1742@xmath1811.@xmath1947 + 28@xmath2022@xmath8540.@xmath9480 & 675 922 & 477&11.28&18.95 & & 18.95 & 18.34 + o5gx08010 & n5272 - 5 & 13@xmath1742@xmath1810.@xmath1986 + 28@xmath2022@xmath8529.@xmath9408 & 533 933 & 758&4.32&18.20 & & 18.16 & 17.65 + o5gx08010 & n5272 - 6 & 13@xmath1742@xmath1812.@xmath1904 + 28@xmath2022@xmath8552.@xmath9410 & 811 907 & 211&16.61&15.95 & & 16.21 & 16.08 + o5gx09010 & n5272 - 7 & 13@xmath1742@xmath1811.@xmath1935 + 28@xmath2022@xmath8528.@xmath9457 & 571 987 & 691&4.31&17.16 & & 17.00 & 16.53 + o5gx09010 & n5272 - 8 & 13@xmath1742@xmath1811.@xmath1955 + 28@xmath2022@xmath8523.@xmath9434 & 546 1040 & 806&10.02&16.08 & & 15.71 & 15.12 + o5gx10010 & n5272 - 9 & 13@xmath1742@xmath1811.@xmath1964 + 28@xmath2022@xmath8526.@xmath9433 & 577 1030 & 423&8.65&16.02 & & 15.71 & 15.28 + o5gx10010 & n5272 - 10 & 13@xmath1742@xmath1811.@xmath1949 + 28@xmath2022@xmath8525.@xmath9461 & 560 1020 & 487&8.26&19.22 & & 19.23 & 18.67 + o5gx10010 & n5272 - 11 & 13@xmath1742@xmath1812.@xmath1987 + 28@xmath2022@xmath8531.@xmath9447 & 707 1108 & 126&21.08&15.86 & & 15.87 & 15.71 + o5gx10010 & n5272 - 12 & 13@xmath1742@xmath1812.@xmath1953 + 28@xmath2022@xmath8530.@xmath9444 & 684 1095 & 177&18.69&15.96 & & 15.59 & 14.84 + o5gx11010 & n5272 - 13 & 13@xmath1742@xmath1812.@xmath1931 + 28@xmath2022@xmath8533.@xmath9442 & 689 1054 & 529&16.21&17.52 & & 17.44 & 16.99 + o5gx11010 & n5272 - 14 & 13@xmath1742@xmath1812.@xmath1932 + 28@xmath2022@xmath8528.@xmath9440 & 651 1088 & 628&15.81&15.91 & & 15.64 & 15.33 + o5gx12010 & n5272 - 15 & 13@xmath1742@xmath1810.@xmath1980 + 28@xmath2022@xmath8531.@xmath9453 & 547 912 & 529&4.21&17.07 & & 17.08 & 17.02 + o5gx13010 & n5272 - 16 & 13@xmath1742@xmath1810.@xmath1980 + 28@xmath2022@xmath8555.@xmath9473 & 734 758 & 186&17.25&15.91 & & 16.15 & 15.98 + o5gx13010 & n5272 - 17 & 13@xmath1742@xmath1810.@xmath1957 + 28@xmath2022@xmath8558.@xmath9472 & 738 716 & 103&15.49&15.94 & & 16.14 & 15.996 + o5gx13010 & n5272 - 18 & 13@xmath1742@xmath1812.@xmath1909 + 28@xmath2022@xmath8541.@xmath9443 & 724 986 & 643&15.65&17.84 & & 17.27 & 16.41 + o5gx14010 & n5272 - 19 & 13@xmath1742@xmath1810.@xmath1951 + 28@xmath2022@xmath8537.@xmath9405 & 565 848 & 303&9.65&18.91 & & 18.94 & 18.32 + o5gx14010 & n5272 - 20 & 13@xmath1742@xmath1811.@xmath1925 + 28@xmath2022@xmath8531.@xmath9410 & 582 961 & 527&1.84&18.54 & & 18.57 & 18.08 + o5gx14010 & n5272 - 21 & 13@xmath1742@xmath1811.@xmath1963 + 28@xmath2022@xmath8527.@xmath9486 & 588 1021 & 650&8.04&18.25 & & 18.17 & 17.64 + + o5gx15010 & n6752 - 1 & 19@xmath1710@xmath1851.@xmath1989 59@xmath2058@xmath8558.@xmath9435 & 831 791 & 713&5.98&17.42 & 17.65 & 17.25 & + o5gx15010 & n6752 - 2 & 19@xmath1710@xmath1851.@xmath1956 59@xmath2058@xmath8555.@xmath9469 & 868 796 & 785&9.73&14.61 & 15.67 & 15.82 & + o5gx16010 & n6752 - 3 & 19@xmath1710@xmath1851.@xmath1935 59@xmath2058@xmath8555.@xmath9475 & 880 787 & 926&10.71&16.80 & 17.09 & 16.67 & + o5gx16010 & n6752 - 4 & 19@xmath1710@xmath1851.@xmath1970 59@xmath2058@xmath8558.@xmath9449 & 842 782 & 850&6.88&16.09 & 16.37 & 15.93 & + o5gx17010 & n6752 - 5 & 19@xmath1710@xmath1849.@xmath1993 59@xmath2058@xmath8555.@xmath9433 & 967 724 & 512&19.79&14.20 & & 15.00 & + o5gx17010 & n6752 - 6 & 19@xmath1710@xmath1850.@xmath1932 59@xmath2058@xmath8557.@xmath9409 & 932 728 & 554&16.31&17.13 & 17.38 & 16.91 & + o5gx17010 & n6752 - 7 & 19@xmath1710@xmath1852.@xmath1915 59@xmath2059@xmath8506.@xmath9492 & 763 736 & 571&3.65&17.18 & 17.54 & 17.09 & + o5gx17010 & n6752 - 8 & 19@xmath1710@xmath1852.@xmath1926 59@xmath2059@xmath8507.@xmath9443 & 753 737 & 904&3.93&16.87 & 17.10 & 16.71 & + o5gx17010 & n6752 - 9 & 19@xmath1710@xmath1852.@xmath1950 59@xmath2059@xmath8508.@xmath9448 & 732 740 & 972&5.19&15.27 & 15.26 & 15.00 &
+ o5gx18010 & n6752 - 10 & 19@xmath1710@xmath1852.@xmath1932 59@xmath2059@xmath8508.@xmath9488 & 741 729 & 544&5.36&17.07 & 17.42 & 17.02 & + o5gx18010 & n6752 - 11 & 19@xmath1710@xmath1851.@xmath1998 59@xmath2059@xmath8506.@xmath9499 & 772 727 & 604&4.32&14.06 & 14.92 & 15.03 &
+ o5gx18010 & n6752 - 12 & 19@xmath1710@xmath1851.@xmath1903 59@xmath2059@xmath8502.@xmath9443 & 857 719 & 772&9.78&14.14 & & 14.10 & + o5gx18010 & n6752 - 13 & 19@xmath1710@xmath1850.@xmath1995 59@xmath2059@xmath8501.@xmath9497 & 864 720 & 784&10.30&17.38 & 17.63 & 17.21 &
+ o5gx19010 & n6752 - 14 & 19@xmath1710@xmath1852.@xmath1921 59@xmath2059@xmath8504.@xmath9428 & 776 759 & 414&1.16&15.61 & 15.61 & 15.50 & + o5gx19010 & n6752 - 15 & 19@xmath1710@xmath1852.@xmath1991 59@xmath2059@xmath8509.@xmath9409 & 704 754 & 552&7.13&16.30 & 16.35 & 16.09 & + o5gx20010 & n6752 - 16 & 19@xmath1710@xmath1853.@xmath1983 59@xmath2058@xmath8559.@xmath9401 & 713 877 & 741&12.11&15.31 & & 16.87 & + o5gx21010 & n6752 - 17 & 19@xmath1710@xmath1853.@xmath1917 59@xmath2058@xmath8540.@xmath9455 & 867 992 & 355&23.89&17.62 & 17.14 & 17.33 & + o5gx21010 & n6752 - 18 & 19@xmath1710@xmath1851.@xmath1959 59@xmath2058@xmath8541.@xmath9469 & 953 908 & 591&22.53&15.86 & 15.86 & 15.75 &
+ + y3fj010 ft & n6397 - 1 & 17@xmath1740@xmath1841.@xmath1969 53@xmath2040@xmath8527.@xmath941 & 548 456 & &2.43 & 14.89&14.83&14.67&14.32 + y3fj0109 t & n6397 - 2 & 17@xmath1740@xmath1841.@xmath1974 53@xmath2040@xmath8528.@xmath944 & 564 429 & &1.39 & 14.97&14.62&14.50&14.10 + y3fj010ht & n6397 - 3 & 17@xmath1740@xmath1841.@xmath1988 53@xmath2040@xmath8530.@xmath944 & 579 380 & &1.67 & 14.34&14.67&14.62&14.44 + y3fj010ct & n6397 - 4 & 17@xmath1740@xmath1842.@xmath1911 53@xmath2040@xmath8531.@xmath947 & 570 326 & &3.57 & 14.87&15.27&15.37&15.12 + y3fj010mt & n6397 - 5 & 17@xmath1740@xmath1842.@xmath1902 53@xmath2040@xmath8533.@xmath944 & 609 315 & &4.77 & 14.70&14.61&14.57&14.26 + y3fj010pt & n6397 - 6 & 17@xmath1740@xmath1842.@xmath1925 53@xmath2040@xmath8534.@xmath940 & 589 274 & &6.12 & 14.74&14.63&14.56&14.34 + y3fj0104 t & n6397 - 7 & 17@xmath1740@xmath1840.@xmath1992 53@xmath2040@xmath8526.@xmath945 & 638 577 & &8.97 & 14.90&13.79&13.85&13.54 + cccc + f336w & 08/07/1999 & 160.0 & 1 + f336w & 08/07/1999 & 200.0 & 6 + f336w & 09/07/1999 & 300.0 & 1 + f336w & 09/07/1999 & 400.0 & 6 + f336w & 10/07/1999 & 600.0 & 1 + f336w & 10/07/1999 & 700.0 & 6 + f336w & 11/07/1999 & 900.0 & 7 + f336w & 13/07/2001 & 400.0 & 6 + f336w & 13/07/2001 & 1000.0 & 5 + f439w & 01/09/1995 & 7.0 & 1 + f439w & 01/09/1995 & 50.0 & 2 + f555w & 01/09/1995 & 1.0 & 1 + f555w & 01/09/1995 & 7.0 & 1 + f555w & 28/10/1999 & 1.0 & 3 + f555w & 28/10/1999 & 120.0 & 2 + f555w & 03/07/1999 & 160.0 & 6 + f555w & 28/10/1999 & 20.0 & 18 + f814w & 03/07/1999 & 160.0 & 5 + + f336w & 25/04/1995 & 70.0 & 2 + f336w & 25/04/1995 & 600.0 & 16 + f336w & 25/04/1995 & 800.0 & 4 + f555w & 14/05/1998 & 3.0 & 2 + f555w & 14/05/1998 & 60.0 & 21 + f555w & 14/05/1998 & 100.0 & 4 + f555w & 14/05/1998 & 400.0 & 6 + f555w & 14/05/1998 & 500.0 & 2 + f814w & 14/05/1998 & 3.0 & 2 + f814w & 14/05/1998 & 100.0 & 12 + f814w & 14/05/1998 & 140.0 & 4 + f814w & 28/04/1999 & 100.0 & 4 + + f336w & 22/03/2001 & 20.0 & 1 + f336w & 22/03/2001 & 260.0 & 3 + f439w & 17/08/1994 & 70.0 & 1 + f439w & 17/08/1994 & 700.0 & 2 + f555w & 22/03/2001 & 0.2 & 1 + f555w & 22/03/2001 & 30.0 & 3 + + f336w & 06/03/1996 & 10.0 & 2 + f336w & 06/03/1996 & 80.0 & 2 + f336w & 06/03/1996 & 400.0 & 28 + f336w & 06/03/1996 & 500.0 & 6 + f336w & 06/03/1996 & 700.0 & 1 + f439w & 07/03/1996 & 10.0 & 2 + f439w & 07/03/1996 & 80.0 & 2 + f439w & 07/03/1996 & 400.0 & 16 + f439w & 07/03/1996 & 500.0 & 4 + f555w & 06/03/1996 & 1.0 & 1 + f555w & 06/03/1996 & 8.0 & 1 + f555w & 06/03/1996 & 40.0 & 6 + f555w & 04/04/1999 & 1.0 & 2 + f555w & 04/04/1999 & 8.0 & 2 + f555w & 04/04/1999 & 40.0 & 24 + f814w & 06/03/1996 & 1.0 & 1 + f814w & 06/03/1996 & 8.0 & 1 + f814w & 06/03/1996 & 40.0 & 2 + f814w & 04/04/1996 & 1.0 & 2 + f814w & 04/04/1996 & 8.0 & 2 + f814w & 04/04/1996 & 40.0 & 24 + llllrrrllrrr + 1&v & to & 3.80 @xmath69 0.15 & 6250 @xmath69 150 & 1.9 @xmath69 0.2 & 4.9 @xmath69 1.3 & 0.84 ( 0.46 - 1.50 ) & 0.40 & -0.99 & 0.02 & + 2&v & to & 3.85 @xmath69 0.15 & 7040 @xmath69 100 & 2.2 @xmath69 0.3 & 10 @xmath69 3 & 1.24 ( 0.67 - 2.22 ) & 0.19 & -1.83 & -0.22 & -0.08 + 3&v & ? & 4.20 @xmath69 0.30 & 7650 @xmath69 250 & 0.9 @xmath69 0.2 & 2.2 @xmath69 1.0 & 0.42 ( 0.13 - 1.27 ) & 0.63 & -0.50 & 0.17 & + 4 & & bs & 3.85 @xmath69 0.15 & 6420 @xmath69 100 & 1.8 @xmath69 0.2 & 4.8 @xmath69 0.9 & 0.83 ( 0.47 - 1.43 ) & 0.69 & -0.40 & 0.10 & 0.17 + 5 & : : & bs & 3.00 @xmath69 0.15 & 6210 @xmath69 300 & 2.8 @xmath69 0.4 & 11 @xmath69 3 & 0.29 ( 0.15 - 0.54 ) & 0.53 & -0.70 & 0.27 & 0.31 + 6&v & to & 4.10 @xmath69 0.20 & 6500 @xmath69 100 & 1.6 @xmath69 0.2 & 4.2 @xmath69 1.2 & 1.22 ( 0.58 - 2.49 ) & 0.57 & -0.61 & -0.18 & + 7 & & bs & 4.05 @xmath69 0.15 & 6840 @xmath69 100 & 2.1 @xmath69 0.2 & 8.9 @xmath69 1.6 & 1.88 ( 1.07 - 3.21 ) & 1.18 & 0.18 & -0.01 & 0.03 + 8&v & bs & 4.25 @xmath69 0.10 & 8000 @xmath69 150 & 1.3 @xmath69 0.2 & 6.1 @xmath69 1.8 & 1.09 ( 0.64 - 1.80 ) & 1.70 & 0.57 & -0.10 & + 9 & & bs & 3.35 @xmath69 0.20 & 6810 @xmath69 100 & 2.5 @xmath69 0.3 & 12 @xmath69 2 & 0.51 ( 0.26 - 0.99 ) & 0.80 & -0.25 & 0.10 & 0.12 + + 1 & & to & 3.60 @xmath69 0.20 & 6000 @xmath69 100 & 2.1 @xmath69 0.2 & 5.2 @xmath69 1.0 & 0.66 ( 0.33 - 1.28 ) & 0.86 & -0.17 & & 0.20 + 2 & & to & 3.40 @xmath69 0.15 & 6250 @xmath69 100 & 1.8 @xmath69 0.2 & 4.5 @xmath69 0.8 & 0.31 ( 0.17 - 0.53 ) & 0.99 & -0.01 & & 0.00 + 3 & : : & to & 4.00 @xmath69 0.10 & 6650 @xmath69 300 & 1.7 @xmath69 0.2 & 4.7 @xmath69 0.8 & 0.99 ( 0.64 - 1.52 ) & 1.20 & 0.19 &
& -0.29 + 4 & & to & 3.00 @xmath69 0.15 & 5750 @xmath69 300 & 1.9 @xmath69 0.3 & 3.6 @xmath69 0.9 & 0.14 ( 0.07 - 0.25 ) & 0.34 & -1.17 & & 0.30 + 5 & & bs & 3.60 @xmath69 0.10 & 6500 @xmath69 100 & 2.1 @xmath69 0.2 & 6.8 @xmath69 1.2 & 0.62 ( 0.40 - 0.95 ) & 0.97 & -0.03 & & 0.15 + 6 & & ehb & 3.60 @xmath69 0.10 & 9750 @xmath69 700 & 2.4 @xmath69 0.2 & 44 @xmath69 11 & 0.80 ( 0.51 - 1.24 ) & 0.90 & -0.11 & & 0.20 + 7 & & bs & 3.40 @xmath69 0.10 & 6500 @xmath69 100 & 3.5 @xmath69 0.4 & 20 @xmath69 4 & 1.14 ( 0.73 - 1.74 ) & 1.05 & 0.05 & & 0.05 + 8&v & hb & 2.00 @xmath69 0.15 & 5750 @xmath69 100 & 7.8 @xmath69 0.9 & 59 @xmath69 14 & 0.22 ( 0.12 - 0.39 ) & 0.89 & -0.12 & & + 9&v & hb & 2.20 @xmath69 0.10 & 6100 @xmath69 150 & 7.1 @xmath69 0.8 & 63 @xmath69 16 & 0.30 ( 0.18 - 0.47 ) & 0.82 & -0.21 & & + 10 & : & to & 4.00 @xmath69 0.15 & 7000 @xmath69 100 & 1.1 @xmath69 0.1 & 2.4 @xmath69 0.4 & 0.42 ( 0.24 - 0.71 ) & 0.42 & -0.94 & & 0.08 + 11 & & ehb & 3.30 @xmath69 0.10 & 9000 @xmath69 100 & 3.0 @xmath69 0.3 & 54 @xmath69 10 & 0.67 ( 0.43 - 1.02 ) & 0.88 & -0.14 & & 0.20 + 12&v & hb & 2.00 @xmath69 0.20 & 5750 @xmath69 100 & 8.1 @xmath69 0.9 & 63 @xmath69 15 & 0.24 ( 0.12 - 0.47 ) & 0.95 & -0.06 & & + 13&v & bs & 2.60 @xmath69 0.15 & 7050 @xmath69 100 & 4.7 @xmath69 0.6 & 49 @xmath69 12 & 0.32 ( 0.18 - 0.57 ) & 0.19 & -1.82 & & 0.40 + 14&v & hb & 3.85 @xmath69 0.10 & 7200 @xmath69 100 & 2.0 @xmath69 0.3 & 9.3 @xmath69 2.5 & 1.01 ( 0.60 - 1.63 ) & 5.18 & 1.79 & & + 15&v & bs & 4.40 @xmath69 0.10 & 8150 @xmath69 100 & 2.0 @xmath69 0.4 & 16 @xmath69 6 & 3.74 ( 2.01 - 6.54 ) & 0.78 & -0.27 & & + 16 & & ehb & 3.60 @xmath69 0.10 & 9750 @xmath69 100 & 2.4 @xmath69 0.2 & 47 @xmath69 9 & 0.85 ( 0.55 - 1.29 ) & 0.81 & -0.23 & & 0.21 + 17 & & ehb & 3.80 @xmath69 0.10 & 10000 @xmath69 200 . & 2.4 @xmath69 0.2 & 49 @xmath69 9 & 1.29 ( 0.83 - 1.96 ) & 0.89 & -0.13 & & 0.22 + 18&v & bs & 4.10 @xmath69 0.10 & 8800 @xmath69 300 & 1.3 @xmath69 0.2 & 8.7 @xmath69 3.0 & 0.75 ( 0.42 - 1.27 ) & 1.12 & 0.12 & & + 19 & : & to & 3.80 @xmath69 0.20 & 6400 @xmath69 100 & 1.5 @xmath69 0.2 & 3.4 @xmath69 0.6 & 0.53 ( 0.27 - 1.02 ) & 0.62 & -0.51 & & 0.07 + 20&v & bs & 3.80 @xmath69 0.15 & 6600 @xmath69 300 & 2.2 @xmath69 0.3 & 8.2 @xmath69 2.2 & 1.12 ( 0.60 - 2.02 ) & 0.48 & -0.81 & & + 21 & & bs & 3.40 @xmath69 0.10 & 6500 @xmath69 100 & 2.1 @xmath69 0.2 & 6.7 @xmath69 1.2 & 0.39 ( 0.25 - 0.59 ) & 1.05 & 0.05 & & 0.04 + + 1 & & to & 3.40 @xmath69 0.20 & 6250 @xmath69 100 & 1.5 @xmath69 0.2 & 3.0 @xmath69 0.6 & 0.20 ( 0.10 - 0.39 ) & 0.89 & -0.12 & 0.10 & 0.09 + 2 & & ehb & 4.30 @xmath69 0.15 & 18000 @xmath69 500 & 0.7 @xmath69 0.1 & 44 @xmath69 8 & 0.35 ( 0.20 - 0.60 ) & 0.89 & -0.13 & 0.09 & 0.11 + 3 & & ato & 3.80 @xmath69 0.20 & 6300 @xmath69 100 & 1.9 @xmath69 0.2 & 4.9 @xmath69 0.9 & 0.81 ( 0.41 - 1.56 ) & 0.86 & -0.16 & 0.05 & 0.06 + 4 & & bs & 3.40 @xmath69 0.20 & 6250 @xmath69 100 & 2.7 @xmath69 0.3 & 10 @xmath69 2 & 0.68 ( 0.34 - 1.31 ) & 0.89 & -0.12 & 0.06 & 0.13 + 5 & v & ehb & 3.70 @xmath69 0.10 & 7250 @xmath69 100 & 2.9 @xmath69 0.3 & 21 @xmath69 5 & 1.57 ( 0.99 - 2.45 ) & 0.80 & -0.24 & & 1.12 + 6 & & to & 3.80 @xmath69 0.15 & 7250 @xmath69 100 & 1.2 @xmath69 0.1 & 3.6 @xmath69 0.7 & 0.34 ( 0.20 - 0.59 ) & 0.81 & -0.23 & -0.16 & 0.07 + 7 & & to & 3.80 @xmath69 0.10 & 6500 @xmath69 150 & 1.4 @xmath69 0.1 & 3.3 @xmath69 0.6 & 0.48 ( 0.31 - 0.74 ) & 0.99 & -0.01 & 0.02 & 0.11 + 8 & & ato & 4.00 @xmath69 0.15 & 6250 @xmath69 100 & 1.9 @xmath69 0.2 & 5.0 @xmath69 0.9 & 1.35 ( 0.77 - 2.33 ) & 1.35 & 0.32 & 0.12 & -0.03 + 9 & v & bs & 4.20 @xmath69 0.10 & 13400 @xmath69 150 & 1.2 @xmath69 0.1 & 42 @xmath69 10 & 0.85 ( 0.53 - 1.32 ) & 0.90 & -0.11 & -0.29 & -0.90 + 10 & : : & to & 3.40 @xmath69 0.25 & 6500 @xmath69 300 & 1.4 @xmath69 0.1 & 3.0 @xmath69 0.5 & 0.17 ( 0.08 - 0.37 ) & 1.02 & 0.02 & -0.18 & -0.22 + 11 & & bs & 3.50 @xmath69 0.05 & 9000 @xmath69 150 & 2.9 @xmath69 0.3 & 49 @xmath69 9 & 0.98 ( 0.71 - 1.33 ) & 0.76 & -0.29 & & 0.25 + 12 & & ehb & 4.20 @xmath69 0.10 & 15800 @xmath69 100 & 1.1 @xmath69 0.1 & 68 @xmath69 13 & 0.71 ( 0.46 - 1.09 ) & 0.97 & -0.04 & 0.03 & 0.06 + 13 & & to & 4.10 @xmath69 0.15 & 7100 @xmath69 100 & 1.2 @xmath69 0.1 & 3.4 @xmath69 0.6 & 0.69 ( 0.40 - 1.18 ) & 1.12 & 0.12 & 0.11 & 0.39 + 14 & & bs & 3.70 @xmath69 0.10 & 7250 @xmath69 100 & 2.3 @xmath69 0.2 & 13 @xmath69 2 & 0.99 ( 0.63 - 1.50 ) & 1.76 & 0.61 & 0.18 & 0.18 + 15 & & bs & 4.30 @xmath69 0.05 & 8600 @xmath69 100 & 1.3 @xmath69 0.1 & 7.9 @xmath69 1.4 & 1.19 ( 0.86 - 1.61 ) & 0.58 & -0.59 & -0.09 & -0.01 + 16 & & ehb & 5.20 @xmath69 0.15 & 24500 @xmath69 500 & 0.3 @xmath69 0.0 & 35 @xmath69 6 & 0.63 ( 0.36 - 1.07 ) & 0.97 & -0.04 & & 0.11 + 17&v & to & 3.80 @xmath69 0.15 & 6650 @xmath69 100 & 1.5 @xmath69 0.2 & 3.7 @xmath69 0.9 & 0.49 ( 0.27 - 0.86 ) & 0.58 & -0.60 & & -0.09 + 18 & & bs & 4.10 @xmath69 0.10 & 9000 @xmath69 100 & 1.4 @xmath69 0.1 & 11 @xmath69 2 & 0.87 ( 0.56 - 1.33 ) & 0.97 & -0.04 & 0.01 & 0.09 + + 1&v & bs & 3.70 @xmath69 0.10 & 8600 @xmath69 100 & 2.0 @xmath69 0.2 & 20 @xmath69 5 & 0.76 ( 0.48 - 1.18 ) & 0.89 & -0.13 & 0.10 & 0.47 + 2&v & bs & 3.80 @xmath69 0.10 & 8450 @xmath69 100 & 2.3 @xmath69 0.3 & 25 @xmath69 6 & 1.25 ( 0.78 - 1.95 ) & 0.85 & -0.17 & 0.11 & 0.46 + 3 & & bs & 4.40 @xmath69 0.10 & 11700 @xmath69 200 & 1.5 @xmath69 0.2 & 38 @xmath69 7 & 2.10 ( 1.35 - 3.20 ) & 0.88 & -0.14 & 0.08 & 0.10 + 4 & & bs & 4.20 @xmath69 0.10 & 13000 @xmath69 300 & 1.0 @xmath69 0.1 & 24 @xmath69 4 & 0.55 ( 0.35 - 0.84 ) & 0.82 & -0.21 & 0.13 & 0.07 + 5&v & bs & 4.20 @xmath69 0.10 & 10200 @xmath69 300 & 1.7 @xmath69 0.2 & 28 @xmath69 7 & 1.68 ( 1.05 - 2.61 ) & 0.87 & -0.15 & -0.06 & + 6&v & bs & 4.00 @xmath69 0.10 & 10000 @xmath69 100 & 1.7 @xmath69 0.2 & 26 @xmath69 6 & 1.07 ( 0.67 - 1.67 ) & 0.88 & -0.14 & 0.02 & + 7&v & hb & 3.70 @xmath69 0.10 & 10500 @xmath69 200 & 2.4 @xmath69 0.3 & 63 @xmath69 15 & 1.07 ( 0.67 - 1.66 ) & 0.83 & -0.20 & 0.10 & + [ tab : results ] lccc + 1 & 15.99 & -0.35 & -0.03 + 2 & 15.07 & -0.19 & -0.24 + 3 & 16.69 & -0.09 & -0.35 + 4 & 15.80 & -0.33 & -0.07 + 5 & 14.84 & -0.17 & -0.02 + 6 & 16.09 & -0.37 & -0.10 + 7 & 15.64 & -0.29 & -0.19 + 8 & 15.58 & -0.05 & -0.44 + 9 & 15.00 & -0.11 & -0.20 + + 1 & 17.88 & -0.45 & -0.07 + 2 & 18.05 & -0.50 & -0.09 + 3 & 18.10 & -0.44 & -0.21 + 4 & 17.32 & -0.48 & 0.00 + 5 & 17.64 & -0.33 & -0.20 + 6 & 15.46 & 0.00 & -0.69 + 7 & 16.60 & -0.31 & -0.18 + 8 & 15.35 & -0.26 & -0.02 + 9 & 15.25 & -0.13 & -0.15 + 10 & 17.72 & -0.29 & -0.30 + 11 & 15.33 & 0.23 & -0.67 + 12 & 15.28 & -0.24 & -0.04 + 13 & 15.40 & 0.17 & -0.40 + 14 & 16.91 & -0.21 & -0.32 + 15 & 16.28 & -0.05 & -0.51 + 16 & 15.31 & 0.05 & -0.70 + 17 & 15.43 & 0.07 & -0.70 + 18 & 17.12 & 0.07 & -0.59 + 19 & 17.87 & -0.44 & -0.12 + 20 & 17.18 & -0.41 & -0.17 + 21 & 17.65 & -0.42 & -0.17 + + 1 & 16.42 & -0.47 & -0.12 + 2 & 13.94 & -0.88 & -0.80 + 3 & 15.26 & -0.32 & -0.14 + 4 & 15.83 & -0.43 & -0.13 + 5 & 14.20 & -0.07 & -0.38 + 6 & 16.07 & -0.14 & -0.36 + 7 & 16.42 & -0.38 & -0.22 + 8 & 16.35 & -0.47 & -0.11 + 9 & 13.57 & -0.50 & -0.76 + 10 & 16.30 & -0.41 & -0.31 + 11 & 13.40 & -0.73 & -0.80 + 12 & 13.36 & 0.30 & -0.65 + 13 & 16.52 & -0.39 & -0.19 + 14 & 15.58 & -0.05 & -0.39 + 15 & 14.88 & 0.05 & -0.58 + 16 & 14.76 & -1.14 & -0.88 + 17 & 15.10 & 0.11 & -0.62 + 18 & 16.08 & -0.31 & -0.19 + + 1 & 13.61 & 0.17 & -0.60 + 2 & 13.41 & 0.16 & -0.57 + 3 & 13.05 & -0.27 & -0.70 + 4 & 13.56 & -0.39 & -0.77 + 5 & 13.32 & 0.06 & -0.70 + 6 & 13.40 & 0.08 & -0.68 + 7 & 12.43 & -0.01 & -0.71 + lcccccc & + 47 tuc & 0.95 & 0.42(4 ) & & & & + m 3 & 0.72 & 0.31(3 ) & 0.56 & 0.56(6 ) & 0.90 & 0.31(4 ) + ngc 6752 & 1.01 & 0.41(5 ) & 0.61 & 0.61(7 ) & 0.59 & 0.41(3 ) + ngc 6397 & 1.73 & 0.43(2 ) & & & &
+ all & 1.04 & 0.17(14 ) & 0.58 & 0.22(13 ) & 0.79 & 0.25(7 ) + & + 47 tuc & 0.99 & 0.37(5 ) & 1.09 & 0.52(3 ) & & + m 3 & 1.03 & 0.24(7 ) & 0.56 & 0.56(6 ) & 0.74 & 0.24(8 ) + ngc 6752 & 0.99 & 0.35(6 ) & 0.59 & 0.59(8 ) & 0.87 & 0.35(4 ) + ngc 6397 & 1.27 & 0.26(6 ) & & & 1.07 & 0.65(1 ) + all & 1.07 & 0.14(24 ) & 0.65 & 0.20(17 ) & 0.81 & 0.19(13 ) + llcccc + n104 - 1 & to & @xmath7140 & &&
+ n104 - 2 & to & @xmath770 & &&
+ n104 - 4 & bs & 120@xmath97 &
&& + n104 - 6 & to & @xmath7120 & &&
+ n104 - 7 & bs & @xmath7120 & && + n104 - 8 & bs & @xmath750 & && + + n5272 - 6 & ehb & & 100@xmath6950 & & + n5272 - 11&ehb & & 70@xmath6930 & & + n5272 - 13&bs & 100@xmath6920 & @xmath7150 & @xmath7200 & + n5272 - 15&bs & & 225@xmath6950 & @xmath7300 & + n5272 - 16&ehb & & 100@xmath6930 & & + n5272 - 17&ehb & & 200@xmath6950 & & + + n6752 - 1 & to & & @xmath7200 & & + n6752 - 2 & ehb & & & @xmath7100 & @xmath7100 + n6752 - 3 & ato & @xmath775 & @xmath7150 & @xmath7150 &
+ n6752 - 4 & bs & @xmath7100 & @xmath7100 & @xmath7200 & + n6752 - 5 & ehb & @xmath750 & 150@xmath6930 & @xmath7300 &
+ n6752 - 6 & to & & & @xmath7100 & + n6752 - 8 & ato & & @xmath7100 & @xmath7250 & + n6752 - 9 & bs & & @xmath725 & @xmath7200 & + n6752 - 11&bs & @xmath750 & 50@xmath6920 & @xmath7200 & @xmath750 + n6752 - 12&ehb & & & @xmath7200 & 50@xmath6920 + n6752 - 13&to & & & @xmath750 & + n6752 - 14&bs & & @xmath750 & & + n6752 - 16&ehb & & & & @xmath7100 + n6752 - 17&to & & @xmath7100 & & + n6752 - 18&bs & & 50@xmath6920 & @xmath7200 & + | we present a spectroscopic analysis of hst / stis and fos low- and intermediate - resolution spectroscopy of 55 stars in four globular clusters ( 47 tucanae , m 3 , ngc 6752 , and ngc 6397 ) .
stars hotter than @xmath0= 5750 k and with a signal - to - noise ratio larger than 15 were analyzed with non - local thermodynamic equilibrium model atmospheres , and values for their effective temperatures and gravities were obtained . using photometric fluxes
, we also obtained radii , luminosities and spectroscopic masses .
twenty - four stars in our sample are blue stragglers ( bs ) .
their photometric colors and magnitudes place these bss above and redward of the clusters zero - age main sequence : this is consistent with the gravities we find for these stars , which are lower than zero - age main sequence gravities .
a comparison with stellar evolutionary tracks shows that almost all of our bss are in the hertzsprung gap .
this is contrary to theory , because of the short timescale expected for stars in this evolutionary phase .
mean bs masses are 1.04 @xmath1 for 14 non - variable stars , or 1.07 @xmath1 counting all 24 bss in our sample .
for the non - variable stars the mean bs mass for individual clusters are 1.73 , 1.01 , 0.95 and 0.72 @xmath1 for ngc 6397 , ngc 6752 , 47 tuc and m 3 , respectively . adding the variable stars ( which improves the statistics but increases the uncertainty ) the mean masses become 1.27 , 1.05 , 0.99 and 0.99 @xmath1 , respectively .
though there is considerable scatter , the bs spectroscopic masses correlate with both effective temperature and brightness of the stars , as expected .
the bss in the mean non - variable turn - off star mass ( 0.58 @xmath1 ) is significantly below the values determined for the bss and below the main sequence turn - off mass .
the mean non - variable horizontal - branch ( hb ) star mass is higher than expected ( 0.79 @xmath1 ) .
in particular , several hb stars have masses well above the main sequence turn - off mass .
some of these hb stars are suspected of actually being bss , since most of them reside at ambiguous locations on the cmd making them prone to misclassification .
values and limits to the stellar rotation rates ( @xmath2 ) are imposed by fitting weak metal lines , the ca ii k line wings , or the helium lines for the hotter stars .
five bss with reasonably constrained rotations show average and median @xmath2 values of 109 and 100 km s@xmath3 , respectively , suggesting @xmath4 km s@xmath3 .
at least some gc bss are are very rapid rotators , but this information can not yet constrain their origin as stellar collision or binary mergers , because of the lack of clear theoretical predictions .
six extreme hb stars have rotation rates @xmath2 between 50 and 200 km s@xmath3 , which are high for these stars and might indicate a binary origin .
de marco et al . found that four bss and two hb star in our sample have balmer jumps which are too large for the effective temperatures implied by the slopes of their paschen continua .
two additional hb stars are now identified in the current study as having the same feature . for these stars ,
the presence of a disk of partly ionized material is suspected , although high stellar rotation rates could also partly explain the data . |
the rigidity of a crystallographic bar - joint framework @xmath0 in the euclidean spaces @xmath1 with respect to periodic first order flexes is determined by a finite matrix , the associated periodic rigidity matrix . for essentially generic frameworks of this type in two dimensions there is a deeper combinatorial characterisation which is a counterpart of laman s characterisation of the infinitesimal rigidity of generic placements of finite graphs in the plane . see ross @xcite . for related results and characterisations of other forms of periodic infinitesimal rigidity see borcea and streinu @xcite , connelly , shen and smith @xcite , malestein and theran @xcite , owen and power @xcite , power @xcite and ross , schulze and whiteley @xcite .
there is also extensive literature in condensed matter physics concerning the nature and multiplicity of low energy oscillations and rigid unit modes ( rums ) for material crystals in three dimensions . in this case
bloch s theorem applies and the excitation modes are periodic modulo a phase factor .
the set of phase factors , or , equivalently , the set of reduced wave vectors for the modes , provides what may be viewed as the rum spectrum of the crystal .
see dove et al @xcite , giddy et al @xcite and wegner @xcite for example . in owen and power @xcite and power
@xcite the rum spectrum was formalised in mathematical terms as a subset @xmath2 of the @xmath4-torus @xmath5 , or as an equivalent subset of @xmath6 , which arises from a choice of translation group @xmath7 .
this set of multi - phases is determined by a matrix - valued function @xmath3 on @xmath5 with the value at @xmath8 providing the corresponding periodic rigidity matrix for @xmath7 . in the present article we move beyond periodicity and
consider infinitesimal flexes of a crystallographic bar - joint framework which are almost periodic in the classical sense of bohr .
such flexes are independent of any choice of translation group and so are intrinsic to @xmath0 as an infinite bar - joint framework .
it is shown that @xmath0 is almost periodically infinitesimally rigid if and only if for some choice of translation group it is periodically infinitesimally rigid and the corresponding rum spectrum is the minimal set @xmath9 .
more generally , we show how the almost periodic infinitesimal flexes of @xmath0 are determined in terms of the matrix function @xmath3 .
an ongoing interest in the analysis of low energy modes in material science is to quantify the implications of symmetry and local geometry for the set of rum wave vectors .
see for example kapco et al @xcite where the phenomenon of extensible flexibility is related to the maximal symmetry and minimal density forms of an idealised zeolite crystal framework . here
the term extensive flexibility corresponds to a maximal rigid unit mode spectrum @xmath10 .
we show that for crystal frameworks whose rum spectrum decomposes into a finite union of linear components there is a corresponding vector space decomposition of the almost periodic flex space .
the flexes in these subspaces are periodic in specific directions associated with certain symmetries of the crystallographic point group .
in section [ examples ] we give a small gallery of crystal frameworks which display a variety of periodic and almost periodic flexibility properties .
a bar - joint framework in the euclidean space @xmath1 is a pair consisting of a simple undirected graph @xmath11 and an injective map @xmath12 . a _ ( real ) infinitesimal flex _ of @xmath13 is a field of velocities , or velocity vectors , @xmath14 assigned to the joints @xmath15 such that for every edge @xmath16 , @xmath17 if the above condition holds for all pairs @xmath18 then @xmath19 is a _ trivial _ infinitesimal flex of @xmath13 . for convenience
we let @xmath20 denote the set of open line segments @xmath21 with @xmath16 .
a _ crystal framework _
@xmath22 is a bar - joint framework @xmath13 for which there exist finite subsets @xmath23 and @xmath24 and a full rank translation group @xmath7 such that @xmath25 @xmath26 the pair @xmath27 is called a _ motif _ for @xmath0 .
the elements of @xmath28 are called _ motif vertices _ and the elements of @xmath29 are called _
motif edges_. the translation group @xmath7 is necessarily of the form @xmath30 where @xmath31 are linearly independent vectors in @xmath1 .
the translation @xmath32 is denoted @xmath33 for each @xmath34 . for each motif vertex @xmath35 and each @xmath36
we denote by @xmath37 the unique vertex for which @xmath38 . for each motif edge
@xmath39 with @xmath40 we let @xmath41 denote the unique edge for which @xmath42 . a velocity vector or flex @xmath19 for @xmath0 is said to be _ strictly periodic _ if @xmath43 for all @xmath44 . in the consideration of infinitesimal rigidity
relative to general periodic flexes , or almost periodic flexes , it is convenient and natural to consider complex velocity vectors @xmath45 . indeed ,
such vectors are infinitesimal flexes if and only if their real and imaginary parts are infinitesimal flexes .
we now define the symbol function @xmath3 of a crystal framework and the rigidity matrix @xmath46 from which it is derived . for @xmath44 the associated monomial function @xmath47
is written simply as @xmath48 .
[ d : phi2 ] let @xmath0 be a crystal framework in @xmath1 with motif @xmath49 and for @xmath50 let @xmath51 .
then @xmath3 is a matrix - valued function on @xmath5 whose rows are labelled by the edges of @xmath29 and whose columns are labelled by the vertex - coordinate pairs in @xmath52 .
the row for an edge @xmath53 with @xmath54 takes the form @xmath55 while if @xmath56 it takes the form @xmath57 [ d : phi2 ] let @xmath0 be a crystal framework in @xmath1 with motif @xmath49 and for @xmath50 let @xmath51 . then @xmath46 is the matrix whose rows are labelled by the edges @xmath41 in @xmath58 and whose columns are labelled by the pairs @xmath37 in @xmath59 .
the row for an edge @xmath60 , with @xmath61 takes the form @xmath62 it follows from this definition that a velocity vector @xmath63 is an infinitesimal flex for @xmath0 if and only if @xmath64 .
a ( real or complex ) velocity vector @xmath19 is said to be _ phase - periodic _ , or , more precisely , @xmath65-phase - periodic for the fixed multi - phase @xmath66 , if @xmath67 for all @xmath68 , @xmath44 . here
@xmath69 is the product @xmath70 .
we also write @xmath71 for this vector , where @xmath72 is the vector @xmath73 in @xmath74 and @xmath75 is the multi - sequence @xmath76 .
the following theorem is given in @xcite and @xcite .
let @xmath0 be a crystal framework in @xmath1 , let @xmath65 be a multi - phase in @xmath5 and let @xmath77 , with @xmath78 , be a @xmath65-phase - periodic velocity field .
then the following conditions are equivalent .
a. @xmath64 .
b. @xmath79 let @xmath0 be a crystal framework in @xmath1 with translation group @xmath7 .
then the following statements are equivalent .
a. the @xmath7-periodic real infinitesimal flexes of @xmath0 are trivial . b. the @xmath7-periodic complex infinitesimal flexes of @xmath0 are trivial
. c. the periodic rigidity matrix @xmath80 has rank equal to @xmath81 .
the _ rum spectrum _ @xmath2 of @xmath0 is defined to be the set of multi - phases @xmath65 for which there exists a nonzero phase - periodic infinitesimal flex for @xmath0 as an infinite - bar - joint framework , or , equivalently , as the set of multi - phases for which the rank of the matrix @xmath82 is less than @xmath83 .
in this section we first outline the proof of the fundamental approximation theorem for uniformly almost periodic functions and its counterpart for almost periodic sequences . these ensure that a function ( or sequence ) which is almost periodic in the sense of bohr , is approximable by trigonometric functions ( or sequences ) that are obtained in an explicit manner from convolution with bochner - fejr kernels .
a convenient self - contained exposition of this fact for function approximation , due to besicovitch @xcite , is given in partington @xcite .
the direct arguments there can be extended to almost periodic vector - valued functions on @xmath84 and this embraces the setting of velocity fields relevant to the almost periodic rigidity of crystal frameworks .
( see definition [ apdef ] . ) the constructive approximation theorem that we require is given in theorem [ t : apzapprox ] .
this theorem together with lemmas [ l1 ] and [ l2 ] lead to the almost periodic rigidity theorem .
first we recall the classical theory for univariable functions on @xmath85 .
the fejr kernel functions are given by @xmath86 the positivity of the @xmath87 and their approximate identity property under convolution with continuous periodic functions feature in a standard proof that a continuous @xmath88-periodic function @xmath89 on the real line is uniformly approximable by the explicit trigonometric functions @xmath90 almost periodic functions on the real line in the sense of bohr are similarly uniformly approximable by an explicit sequence of trigonometric polynomials that are determined by convolution with certain bochner - fejr kernels .
note first that it is elementary that the functions @xmath91 have the form @xmath92 $ ] where @xmath93 $ ] is the mean inner product @xmath94= \lim_{t\to \infty } \frac{1}{2t}\int_{-t}^t f_1(s)\overline{f_2(s)}ds\ ] ] and @xmath95 .
it is classical that for a function @xmath89 in @xmath96 there is a sequence of bochner - fejr kernels @xmath97 which provide , by the same formula , a uniformly approximating sequence of trigonometric polynomials @xmath91 . in this case
the frequencies @xmath98 in the nonzero terms @xmath99 of these approximants appear in a countable set derived ( by rational division ) from the spectrum @xmath100 of @xmath101 defined by @xmath102 \neq 0\}\ ] ] it follows from a parceval inequality for almost periodic functions that this spectrum , which we refer to as the bohr spectrum , is a well - defined finite or countable set .
similar considerations apply to the space @xmath103 of almost periodic sequences .
the approximants are general trigonometric sequences , that is , sequences @xmath104 that have a finite sum form @xmath105
so that , in our earlier notation , @xmath106 where @xmath107 is a finite subset of @xmath85 . the bohr spectrum of a sequence @xmath108 in @xmath103 is defined to be the set @xmath109_{{\mathbb{z}}}\neq 0 , \mbox { for } \omega = e^{i\lambda}\}\ ] ] where @xmath110_{{\mathbb{z}}}= \lim_{n\to \infty}\frac{1}{2n+1}\sum_{|k|\leq n } h_1(k)\overline{h_2(k)}\ ] ] this spectrum of @xmath65 values is now a subset of @xmath111 .
partial counterparts of the fejr kernel functions @xmath112 are given by the fejr sequences @xmath113 associated with a single frequency @xmath114 .
in particular @xmath115 the following fundamental approximation theorem indicates the explicit construction of the bochner - fejr kernels @xmath97 for @xmath108 as coordinate - wise products of appropriate fejr sequences .
[ t : apzapprox ] let @xmath108 be a sequence in @xmath103 , let @xmath116 be a maximal subset of the bohr spectrum @xmath117 which is independent over @xmath118 , and for @xmath119 let @xmath120 then @xmath108 is the uniform limit of the sequence @xmath121 of trigonometric sequences in @xmath103 given by @xmath122_{{\mathbb{z } } } , \quad k\in { { \mathbb{z}}}\ ] ] the main ingredient in the proof of the theorem is that the bohr spectrum is nonempty if @xmath123 and the arguments for this depend on the equivalence of bohr almost periodicity with the bochner condition that the set of translates of @xmath108 is precompact for the uniform norm .
( see @xcite . )
the arguments leading to theorem [ t : apzapprox ] can be generalised to obtain an exact counterpart theorem for @xmath124 .
the approximating trigonometric sequences @xmath125 now have a finite sum form @xmath126 where @xmath127 and @xmath75 , for @xmath128 in @xmath5 , is the pure frequency sequence @xmath129 in @xmath130 with @xmath131 the bohr spectrum @xmath117 is similarly defined and is a countable subset of points @xmath65 in @xmath5 which we freely identify with a countable subset of points @xmath98 in @xmath132 . for notational convenience
we state the general theorem only in the case @xmath133 .
the metric which is appropriate in our context for the approximation of velocity fields is the uniform metric or norm @xmath134 ; for velocity fields @xmath135 we have @xmath136 [ multiapprox ] let @xmath108 be a sequence in @xmath137 , let @xmath138 be a maximal subset of @xmath139 which is independent over @xmath118 , and for @xmath119 let @xmath140 then @xmath108 is the uniform limit of the sequence @xmath121 of trigonometric sequences in @xmath137 given by @xmath141_{{{\mathbb{z}}}^2 } , \quad k\in { { \mathbb{z}}}^2\ ] ] note that here @xmath93_{{{\mathbb{z}}}^2}$ ] is the natural well - defined sesquilinear map from @xmath142 to @xmath143 .
we remark that the theory of uniformly almost periodic multi - variable functions on @xmath1 and on @xmath84 is part of the abstract theory of almost periodic functions on locally compact abelian groups due to bochner and von neumann @xcite .
for further detail see also levitan and zhikov @xcite , loomis @xcite and shubin @xcite .
we now characterise when a crystal framework admits no nontrivial almost periodic infinitesimal flexes and in this case we say that it is almost periodically rigid .
this is evidently a form of rigidity which is independent of any choice of translation group . for a crystal framework
@xmath0 with full rank translation group @xmath7 various linear transformations may be associated with the rigidity matrix @xmath46 .
these transformations are restrictions of the induced linear transformation @xmath144 where @xmath145 is the vector space of all velocity fields on @xmath146 , that is , the space of functions @xmath147 .
the codomain of @xmath46 is the vector space of complex - valued functions on the set of edges .
the right shift operators on the domain and codomain of @xmath46 for the integral vector @xmath148 in @xmath84 are denoted by @xmath149 and @xmath150 respectively . here
the right shift of a sequence @xmath151 by @xmath148 is the sequence @xmath152 .
we note that @xmath153 [ apdef ] let @xmath154 be a velocity field . 1 .
an integral vector @xmath148 in @xmath84 is an @xmath155-translation vector for @xmath108 if @xmath156 .
2 . the velocity field @xmath108 is bohr almost periodic if for every @xmath157 the set of @xmath155-translation vectors @xmath148 is relatively dense in @xmath1 .
[ l1 ] let @xmath125 be the vector - valued trigonometric multi - sequence with finite sum representation @xmath158 with nonzero coefficients @xmath159 in @xmath74 .
if @xmath125 is a nonzero infinitesimal flex for @xmath0 then each component sequence @xmath160 is a nonzero @xmath65-phase - periodic infinitesimal flex . for @xmath161 and @xmath162
let @xmath163 be the linear map on the normed space @xmath164 given in terms of the right shift operators @xmath165 , @xmath44 , by @xmath166 similarly , let @xmath167 be the linear map on @xmath168 given by , @xmath169 the sequence @xmath170 converges uniformly to @xmath171 when @xmath172 and to the zero sequence otherwise since for each @xmath173 , @xmath174 and @xmath175 note also that @xmath46 is a bounded linear transformation from @xmath176 to @xmath177 which commutes with the right shift operators .
thus if @xmath65 is a multi - frequency for @xmath125 then @xmath178 [ l2 ] let @xmath179 be a trigonometric polynomial in @xmath180 and let @xmath108 be an infinitesimal flex for @xmath0 in @xmath181 . then the mean convolution multi - sequence @xmath182 given by @xmath183_{{{\mathbb{z}}}^d}$ ] is an infinitesimal flex for @xmath0 . by linearity it suffices to assume that @xmath179 is the elementary multi - sequence @xmath75 , so that @xmath184 for @xmath44 .
then @xmath125 is the uniform limit of the sequence @xmath185 where @xmath186 to see this note that the convergence is uniform if @xmath108 is a trigonometric sequence .
since the linear maps @xmath187 are contractive for the uniform norm , uniform convergence holds for a general almost periodic velocity sequence .
thus , since the vector space of infinitesimal flexes is invariant under translation it follows that @xmath188 for each @xmath189 and hence that @xmath190 .
the following theorem shows that a crystal framework is almost periodically rigid if and only if it is periodically rigid and the rum spectrum is trivial .
[ apthm ] let @xmath0 be a crystallographic bar - joint framework in @xmath1 .
the following statements are equivalent .
a. every almost periodic infinitesimal flex of @xmath0 is trivial .
b. every strictly periodic infinitesimal flex of @xmath0 is trivial and @xmath191 .
@xmath192 this follows since every phase periodic infinitesimal flex is also an almost periodic infinitesimal flex .
@xmath193 let @xmath19 be an almost periodic infinitesimal flex
. then by theorem [ multiapprox ] , for @xmath4 dimensions , @xmath19 is a uniform limit of the sequence @xmath194 of trigonometric sequences in @xmath195 given by @xmath196_{{{\mathbb{z}}}_d } , \,\,\,\,\,\,\ , k \in { { \mathbb{z}}}^d\ ] ] where @xmath197 , @xmath198 , is the sequence of bochner - fejr kernels for @xmath19 . by lemma [ l2 ] each trigonometric sequence @xmath199 is an infinitesimal flex of @xmath0 and so by lemma [ l1 ] each @xmath199 is a finite linear combination of phase - periodic infinitesimal flexes of @xmath0 . by hypothesis , the rum spectrum is trivial and so each @xmath199 is strictly periodic .
it follows that @xmath19 is strictly periodic as desired .
the bohr spectrum of an almost periodic infinitesimal flex @xmath19 of the crystal framework @xmath0 is the finite or countable set given by @xmath200_{{{\mathbb{z}}}^d } \neq 0 , \mbox { for } \omega = e^{2\pi i\lambda}\in { { \mathbb{t}}}^d\}\ ] ] it follows from theorem [ multiapprox ] , lemma [ l1 ] and lemma [ l2 ] , as in the proof above , that @xmath201 is contained in the rum spectrum of @xmath0 . also , since phase - periodic flexes are almost periodic it follows that the rum spectrum as a subset of @xmath6 is the union of the bohr spectra of all almost periodic infinitesimal flexes .
note that the spectra here depend on the translation group in the following manner .
if @xmath202 has the same underlying bar - joint framework as @xmath0 but full rank translation group @xmath203 then the infinitesimal flex @xmath19 is represented anew as a sequence in @xmath204 where @xmath205 is a vertex motif for @xmath206 .
the bohr spectrum @xmath207 is then the image of @xmath208 under the surjective map @xmath209 induced by the inclusion @xmath203 .
this follows the same relationship as that for the rum spectrum noted in @xcite .
it also follows from this that the dimension of @xmath2 , as a topological space or as an algebraic variety , is independent of the translation group and we refer to this integer , which takes values between @xmath210 and @xmath4 , as the rum dimension of @xmath0 .
we see in the next section that @xmath2 , as a subset of @xmath6 , often decomposes as a union of linear components .
this is the case for example in two dimensions if @xmath3 is a square matrix function whose determinant polynomial @xmath211 either vanishes identically or factorises into simple factors of the form @xmath212 with @xmath213 .
it follows that each almost periodic flex @xmath19 of @xmath0 admits a finite sum decomposition @xmath214 in which each component @xmath215 is an almost periodic flex whose bohr spectrum lies in the @xmath216 linear component .
such component flexes are partially periodic , being periodic in certain directions of translational symmetry .
we now exhibit a number of illustrative examples .
the first two of these show two extreme cases , firstly where the rum spectrum is a singleton , and secondly where the rum spectrum is @xmath5 .
let @xmath217 be the crystallographic bar - joint framework with motif @xmath27 and translation group @xmath7 indicated in table [ cftriangle ] .
simplifying earlier notation , the motif vertex is labelled @xmath218 and the motif edges are labelled @xmath219 , @xmath220 and @xmath221 .
the translation group is @xmath222 where @xmath223 and @xmath224 .
the symbol function for @xmath0 is , @xmath225 note that @xmath226 has rank @xmath227 unless @xmath228 and so the rum spectrum of @xmath0 is the singleton @xmath229 . also , there are no non - trivial strictly periodic infinitesimal flexes of @xmath0 and so , by theorem [ apthm ] , @xmath0 is almost periodically infinitesimally rigid .
in fact @xmath0 is sequentially infinitesimally rigid in the sense that there exists an increasing chain of finite subgraphs @xmath230 of @xmath231 such that every vertex of @xmath231 is contained in some @xmath232 and each subframework @xmath233 is infinitesimally rigid .
for example , for each @xmath234 take @xmath232 to be the vertex - induced subgraph on @xmath235 .
it follows that @xmath0 admits no nontrivial infinitesimal flexes and so is ( absolutely ) infinitesimally rigid as a bar - joint framework .
in @xcite we obtain a general characterisation of countable simple graphs @xmath231 whose locally generic placements are infinitesimally rigid in this sense .
the condition is that @xmath231 should contain a vertex - complete chain of @xmath236-tight subgraphs .
the crystal framework @xmath0 may be viewed as a nongeneric placement of such a graph which remains infinitesimally rigid despite the crystallographic symmetry .
let @xmath237 be the basic one - dimensional grid framework for the lattice @xmath238 in @xmath85 . for any crystal framework @xmath0 in @xmath1 one may construct a _ product framework _ @xmath239 in @xmath240 whose intersection with the hyperplanes @xmath241 are copies of @xmath0 and where these copies are connected by the edges @xmath242 . in the case
that @xmath0 has square matrix symbol function @xmath243 it is straightforward to verify that @xmath244 this leads readily to the identification of the rum spectrum in @xmath245 of such frameworks .
further three - dimensional examples not of this product form may be found in power @xcite and wegner @xcite .
let @xmath237 be the three dimensional framework derived from the regular octagon framework of example [ octagon ] .
then the crystal polynomial admits a three - fold factorisation and from this it follows that the rum spectrum has the topological structure of four two - dimensional tori connected over the common circle of points @xmath246 in @xmath245 .
m. t. dove , a. k. a. pryde , v. heine and k. d. hammonds , exotic distributions of rigid unit modes in the reciprocal spaces of framework aluminosilicates , j. phys . ,
matter 19 ( 2007 ) doi:10.1088/0953 - 8984/19/27/275209 .
a. p. giddy , m. t. dove , g. s. pawley , v. heine , the determination of rigid unit modes as potential soft modes for displacive phase transitions in framework crystal structures .
acta crystallogr . , a49 ( 1993 ) , 697 - 703 .
s. c. power , crystal frameworks , matrix - valued functions and rigidity operators , operator theory : advances and applications .
volume 236 , 2014 : concrete operators , spectral theory , operators in harmonic analysis and approximation , proceedings of iwota 2011 , seville . | a crystallographic bar - joint framework @xmath0 in @xmath1 is shown to be almost periodically infinitesimally rigid if and only if it is strictly periodically infinitesimally rigid and the rigid unit mode ( rum ) spectrum @xmath2 is a singleton .
moreover the almost periodic infinitesimal flexes of @xmath0 are characterised in terms of a matrix - valued function @xmath3 on the @xmath4-torus @xmath5 determined by a full rank translation symmetry group and an associated motif of joints and bars
. [ section ] [ theorem]lemma [ theorem]characterization [ theorem]proposition [ theorem]property [ theorem]problem [ theorem]example [ theorem]remark [ theorem]corollary [ theorem]definition |
let @xmath0 be a simply connected subset of the hexagonal faces in the planar honeycomb lattice .
two faces of @xmath0 are considered adjacent if they share an edge .
suppose further that the boundary faces of @xmath0 are partitioned into a `` left boundary '' component , colored black , and a `` right boundary '' component , colored white , in such a way that the set of interior faces remains simply connected .
( see figure [ hesetuphexagons ] . ) given any black white coloring of the faces of @xmath0 , there will be a unique interface @xmath1 separating the cluster of black hexagons containing the left boundary from the cluster of white hexagons containing the right boundary .
if the colors are chosen via independent bernoulli percolation , we may view @xmath1 as being generated dynamically as follows : simply begin the path @xmath1 at an edge separating the left and right boundary components ; when @xmath1 hits a black hexagon , it turns right , and when it hits a white hexagon , it turns left .
each time it hits a hexagon whose color has yet to be determined , we choose that hexagon s color with a coin toss .
the _ harmonic explorer _ ( he ) is a random interface generated the same way , except that each time @xmath1 hits a hexagon @xmath2 whose color has yet to be determined , we perform a simple random walk on the space of hexagons , beginning at @xmath2 , and let @xmath2 assume the color of the first black or white hexagon hit by that walk .
( see figure [ hesetuphexagons ] . ) in other words , we color @xmath2 black with probability equal to the value at @xmath2 of the function which is equal to @xmath3 on the black faces and @xmath4 on the white faces , and is _ discrete harmonic _ at the undetermined faces ( i.e. , its value at each such face is the mean of the values on the six neighboring faces ) .
denote by @xmath5 the value of this function after @xmath6 steps of the harmonic explorer process ; that is , @xmath7 is @xmath3 if @xmath2 is black , @xmath4 if @xmath2 is white , and discrete harmonic on the faces of undetermined color .
note that @xmath7 is also the probability that a random walk on faces , started at @xmath2 , hits a black face before hitting a white face .
it is easy to see ( and proved below ) that for any fixed @xmath2 , @xmath7 is a martingale and that the harmonic explorer is the only random path with this property .
we will see later that @xmath8 is the only random path with a certain continuous analog of this property .
it was conjectured in @xcite and proved in @xcite that if the interior hexagons are each colored via critical bernoulli percolation ( i.e. , @xmath9 ) , then , in a certain well - defined sense , the random paths @xmath1 tend to the stochastic loewner evolution with parameter @xmath10 ( @xmath11 ) as the hexagonal mesh gets finer .
( see the survey @xcite for background on sle . )
it has been further conjectured @xcite that if colors are instead chosen from a critical fk cluster model ( where one weights configurations according to the total number of clusters and the lengths of their interfaces ) , then @xmath1 will converge to some @xmath12 with @xmath13 , where @xmath14 depends on the weight parameters .
we will prove that , as the mesh gets finer , the harmonic explorer converges to chordal @xmath8 .
there are also natural variants of the harmonic explorer ; for example , we might replace the honeycomb lattice with another three - regular lattice or replace the simple random walk on faces with a different periodic markov chain .
one may even use a non - three - regular lattice provided one fixes an appropriate ordering ( say , left to right ) for determining the color of multiple undetermined faces that are `` hit '' simultaneously by the he path . provided the simple random walk converges to brownian motion as the mesh gets finer , we see no barrier to extending our results to all of these settings .
our proofs are more like the lerw proofs in @xcite ( which hold for general lattices ) than the percolation proof in @xcite ( which uses the invariance of the lattice under @xmath15 rotation in an essential way ) . however , for simplicity , we will focus only on the hexagonal lattice in this paper .
although physicists and mathematicians have conjectured that many models for random self - avoiding lattice walks have conformally invariant scaling limits [ e.g. , the infinite self - avoiding walk , critical percolation cluster boundaries on two - dimensional lattices , critical ising model interfaces , critical fk cluster boundaries and @xmath16 model strands , etc . ] , rigorous proofs are available only in the following cases : percolation interface on the hexagonal lattice ( which converges to chordal @xmath11 ) , harmonic explorer ( chordal @xmath8 ) , loop erased random walk ( lerw ) on a periodic planar graph ( radial @xmath17 @xcite ) , the uniform spanning tree ( ust ) boundary ( chordal @xmath18 @xcite ) and the boundaries of simple random walks ( essentially @xmath19 @xcite here conformal invariance follows easily from the conformal invariance of brownian motion ) .
the harmonic explorer is similar in spirit to the loop erased random walk ( lerw ) and diffusion limited aggregation ( dla ) .
all three models are processes based on simple random walks , and their transition probabilities may all be computed using discrete harmonic functions with appropriate boundary conditions . since simple random walks on two - dimensional lattices have a conformally invariant scaling limit ( brownian motion ) , and since harmonicity ( in the continuous limit ) is a conformally invariant property , one might expect that all three models would have conformally invariant scaling limits .
however , simulations suggest that dla is not conformally invariant .
this paper follows the strategy of @xcite , and uses some of the techniques from that paper .
we will freely quote results from @xcite , and therefore advise the reader to have a copy of @xcite on hand while reading the present paper .
the purpose of this section is to briefly review some background about loewner s equation and sle , and then present the basic strategy of the paper . for more details ,
the reader is encouraged to consult @xcite or @xcite .
let @xmath20 .
suppose that @xmath21\to\overline{{\mathbb{h}}}$ ] is a continuous simple path in the closed upper half plane @xmath22 which satisfies @xmath23\cap{\mathbb{r}}=\{\gamma(0)\}=\{0\}$ ] .
for every @xmath24 $ ] , there is a unique conformal homeomorphism @xmath25 $ ] which satisfies the so - called _ hydrodynamic _ normalization at infinity @xmath26 the limit @xmath27):=\lim_{z\to\infty } z\bigl(g_t(z)-z\bigr)/2\ ] ] is real and monotone increasing in @xmath28 .
it is called the ( half plane ) _
capacity _ of @xmath29 $ ] from @xmath30 , or just capacity , for short .
since @xmath31)$ ] is also continuous in @xmath28 , it is natural to reparameterize @xmath1 so that @xmath31)=t$ ] .
loewner s theorem states that in this case the maps @xmath32 satisfy his differential equation @xmath33 where @xmath34 .
( since @xmath35 is not in the domain of definition of @xmath32 , the expression @xmath36 should be interpreted as a limit of @xmath37 as @xmath38 inside @xmath39 $ ] .
this limit does exist . )
the function @xmath40 is continuous in @xmath28 , and is called the _ driving parameter _ for @xmath1 .
one may also try to reverse the above procedure .
consider the loewner evolution defined by the ode ( [ e.chordal ] ) , where @xmath40 is a continuous , real - valued function .
the path of the evolution is defined as @xmath41 , where @xmath42 tends to @xmath40 from within the upper half plane @xmath43 , provided that the limit exists .
the process ( chordal ) @xmath12 in the upper half plane , beginning at @xmath4 and ending at @xmath30 , is the path @xmath35 when @xmath40 is @xmath44 , where @xmath45 is a standard one - dimensional brownian motion .
( `` standard '' means @xmath46 and @xmath47=t$ ] , @xmath48 .
since @xmath49 has the same distribution as @xmath50 , taking @xmath51 is equivalent . ) in this case a.s .
@xmath35 does exist and is a continuous path .
see @xcite ( @xmath52 ) and @xcite ( @xmath53 ) .
fix @xmath54 , and assume now that @xmath55 and @xmath1 is @xmath56 .
. then @xmath58 gives @xmath59 times the probability that a two - dimensional brownian motion starting at @xmath42 first exits @xmath60 $ ] either in @xmath61 or on the left - hand side of @xmath29 $ ] .
( this follows from conformal invariance of a brownian motion , run until its first exit point , and the fact that the probability that a brownian motion started at @xmath62 first hits @xmath63 at @xmath64 $ ] is @xmath65 .
the latter fact may be seen by conformally mapping the half plane to a strip using the function @xmath66 . ) in other words , for fixed @xmath28 , @xmath67 is the harmonic function that is equal to @xmath59 on one side of @xmath29 $ ] and @xmath4 on the other . for short
, we will sometimes refer to the quantity @xmath58 as simply the _ angle _ of @xmath42 at time @xmath28 .
now , using it s formula , we compute the it derivatives of @xmath68 and @xmath69 : @xmath70 when @xmath71 , we have @xmath72 , and hence @xmath73 . in particular , this implies that for any fixed value of @xmath42,@xmath74 is a martingale which is bounded in the interval @xmath75 $ ] .
the value of this martingale a.s . tends to either zero or @xmath3 as @xmath28 tends to infinity , depending on whether @xmath42 is on the left or the right side of the path ( see @xcite , lemma 3 ) .
hence , at a fixed time @xmath28 , @xmath76 represents the _ probability _ that , conditioned on the sle path up until time @xmath28 , the point @xmath42 will lie to the left of the path .
it is easy to see ( and shown below ) that a discrete version of this property holds for the harmonic explorer .
the strategy of our @xmath8 proof will be , roughly speaking , to show that the fact that this property holds at two distinct values of @xmath42 is enough to force the loewner driving process for the path traced by the harmonic explorer to converge to brownian motion .
this is because the fact that @xmath58 is a martingale at @xmath42 gives a linear constraint on the drift and diffusion terms at that point , and using two values of @xmath42 gives two linear constraints , from which it is possible to calculate the drift and diffusion exactly .
the arguments and error bounds needed to make this reasoning precise are essentially the same as those given in @xcite ( but the martingales considered there are different ) .
the fact that the loewner driving process converges to brownian motion will enable us to conclude that he converges to @xmath8 in the hausdorff topology .
we will then employ additional arguments to show that the convergence holds in a stronger topology .
we remark that we will reuse this strategy in @xcite to prove that a certain zero level set of the discrete gaussian free field ( defined on the vertices of a triangular lattice , with boundary conditions equal to an appropriately chosen constant @xmath77 on the left boundary and @xmath78 on the right boundary ) converges to chordal @xmath8 . to keep notation consistent with @xcite ( which will cite the present paper )
, we will use the dual formulation ( representing hexagons by vertices of the triangular lattice ) in our precise statements and proofs below .
we now introduce the precise combinatorial notation for he that we will use in our proofs .
first , the triangular grid in the plane will be denoted by @xmath79 .
its vertices , denoted by @xmath80 , are the sublattice of @xmath81 spanned by @xmath3 and @xmath82 ; two vertices are adjacent if their difference is a sixth root of unity .
if @xmath83 , and @xmath84 , let @xmath85 denote the _ inradius _ of @xmath0 about @xmath42 ; that is , @xmath86 .
let @xmath87 denote the set of domains @xmath83 whose boundary is a simple closed curve which is a union of edges from the lattice @xmath79 .
if @xmath88 is any set of vertices in @xmath80 , and @xmath89 is a bounded function , then there exists a unique bounded function @xmath90 which agrees with @xmath91 in @xmath88 and is harmonic at every vertex in @xmath92 .
this function is called the _ discrete harmonic extension _ of @xmath91 .
[ in fact , @xmath93 is the expected value of @xmath91 at the point at which a simple random walk started at @xmath94 hits @xmath88 .
uniqueness is easily established using the maximum principle . ]
let @xmath95 .
let @xmath96 denote the set of vertices in @xmath97 .
let @xmath98 and @xmath99 be the centers of two distinct edges of the grid @xmath79 on @xmath97 .
( see figure [ f.setup ] . )
let @xmath100 ( resp .
@xmath101 ) be the positively ( resp . negatively ) oriented arc of @xmath97 from @xmath98 to @xmath99 .
define @xmath102 to be @xmath3 on @xmath103 , and @xmath4 on @xmath104 .
the he [ depending on the triple @xmath105 is a random simple path from @xmath98 to @xmath99 in @xmath106 .
let @xmath107 be i.i.d .
random variables , uniform in the interval @xmath75 $ ] .
( these will be the `` coin flips '' needed to generate the he . )
let @xmath108 be the triangle of @xmath79 whose boundary contains @xmath98 let @xmath109 be the vertex of @xmath110 that is not on the edge containing @xmath98 and let @xmath111 .
let @xmath112 be the middle of the edge of @xmath110 which is on the positively oriented arc from @xmath98 to @xmath109 , and let @xmath113 be the middle of the edge of @xmath110 which is on the positively oriented arc from @xmath109 to @xmath98 .
let @xmath114 be the value at @xmath109 of the discrete harmonic extension of @xmath115 .
if @xmath116 , we let @xmath117 , and otherwise @xmath118 .
the beginning of the he path is chosen as the union of the two line segments from @xmath98 to the center of the triangle @xmath110 and then to @xmath119 .
now define @xmath120 to equal @xmath115 on @xmath88 and set @xmath121 if @xmath122 .
this defines the first step of the he . .
] the process continues inductively . assuming that @xmath123 and @xmath124 , let @xmath125 be the triangle of @xmath79 containing @xmath126 but not @xmath127 .
let @xmath128 be the vertex of @xmath125 which is not on the edge containing @xmath126 , and let @xmath129 .
let @xmath130 be the value at @xmath128 of the discrete harmonic extension of @xmath131 .
let @xmath132 and @xmath133 be the two midpoints of edges of @xmath125 that lie on the positively oriented arcs of @xmath134 from @xmath126 to @xmath128 and from @xmath128 to @xmath126 , respectively . if @xmath135 let @xmath136 , and otherwise @xmath137 .
let the next step of the he consist of the segments from @xmath126 to the center of @xmath125 and from the center of @xmath125 to @xmath138 . also , let @xmath139 agree with @xmath131 , where @xmath131 is defined and set @xmath140 if @xmath141 .
it is easy to verify that this procedure a.s .
terminates when @xmath142 , and that the he so defined is a simple path from @xmath98 to @xmath99 .
let @xmath143 denote the termination time ; that is , the @xmath6 such that @xmath142 .
[ l.hmart ] let @xmath5 denote the discrete harmonic extension of @xmath131 , and let @xmath144 .
then @xmath145 is a martingale and @xmath146 .
given @xmath147 , we have @xmath148 with probability@xmath149 and otherwise @xmath150 .
consequently , @xmath151=h_n(v_{n+1})$ ] .
note that @xmath152 is also the discrete harmonic extension of its restriction to @xmath153 , and similarly for @xmath154 .
since the harmonic extension is a linear operation and @xmath155=h_n(v)$ ] for @xmath156 , the same relation holds for every @xmath94 .
thus @xmath145 is a martingale .
the claim that @xmath146 is clear .
[ r.hemark ] the evolution of the he path may be viewed as a markov chain on the collection of appropriately marked domains . at the @xmath6th step ,
the chain is at @xmath157 , where @xmath158 is the connected component of @xmath159 that has @xmath99 on its boundary .
let @xmath95 .
we assume the setup and notation of section [ s.heconstruction ] .
let @xmath160\to d\cup\{{\bar{v}}_0,{{\bar{v}}_\mathrm{end}}\}$ ] be the he path with the parameterization proportional to arclength , where @xmath161 for @xmath162 .
let @xmath163 be a conformal map onto @xmath43 that takes @xmath98 to @xmath4 and @xmath99 to @xmath30 .
note that @xmath164 is unique up to positive scaling , @xmath165 and @xmath166 .
let @xmath167 . instead of rescaling the grid , we consider larger and larger domains @xmath0 .
the quantity @xmath168 turns out to be the appropriate indicator of the size of @xmath0 , from the perspective of the map @xmath164 . indeed , if @xmath169 is small , then the image under @xmath164 of the grid @xmath79 in @xmath0 is not fine near @xmath170 , and we can not expect @xmath171 to look like @xmath8 . as we will see , @xmath172 does approach @xmath8 when @xmath173 .
let @xmath174 be the path @xmath172 , parameterized by capacity from @xmath30 in @xmath175 , and let @xmath176 be the @xmath8 path in @xmath175 .
let @xmath177 be the metric on @xmath178 given by @xmath179 , where @xmath180 maps @xmath178 onto @xmath181 .
if @xmath182 , then @xmath183 is equivalent to @xmath184 , and @xmath185 is equivalent to @xmath186 . note that although we started by mapping our domain @xmath0 to the half plane ( with boundary points @xmath4 , @xmath30 and inradius measured from the preimage of @xmath170 ) , the above metric corresponds to a mapping to the unit disc ( with boundary points @xmath187 , @xmath3 and inradius measure from the preimage of @xmath4 ) .
the half plane is the most convenient setting for describing loewner evolution and chordal @xmath8 , but the metric derived from the unit disc map is more convenient because it is compact .
[ t.heunifconv ] as @xmath173 , the law of @xmath174 tends to the law of the @xmath8 path @xmath176 , with respect to uniform convergence in the metric @xmath188 . in other words , for every @xmath189 there is some @xmath190 such that if @xmath191 , then there is a coupling of @xmath174 and @xmath176 such that @xmath192<{\varepsilon}.\ ] ]
let @xmath193 denote the loewner driving process for @xmath174 .
let @xmath194 be a standard one - dimensional brownian motion
. a slightly weaker form of theorem [ t.heunifconv ] will follow as a consequence of the fact that for every @xmath20 the restriction of @xmath195 to @xmath196 $ ] converges in law to the restriction of @xmath197 to @xmath196 $ ] .
this , in turn , will be a consequence of the following local statement .
[ p.helocal ] for @xmath198 $ ] let @xmath199)$ ] , @xmath200 $ ] , and let @xmath201 be the conformal map normalized by @xmath202 as @xmath203 in @xmath43 . for every @xmath204
there is an @xmath205 such that the following holds .
fix any @xmath206 . on the event @xmath207 ,
let @xmath208 be the least integer larger than @xmath6 such that @xmath209 .
( note that @xmath210 , since @xmath211 . )
set @xmath212 and let @xmath213 be the event @xmath214
. then @xmath215\bigr ] = w(t_n)+o(\delta^3)\ ] ] and @xmath216\bigr ] = 4{\mathbf{e}}\bigl[t_m - t_n{\vert}\gamma[0,n]\bigr ] + o(\delta^3)\ ] ] both hold on the event @xmath217 . here , and below
, @xmath218 represents any quantity whose absolute value is bounded by @xmath219 , where @xmath220 is any fixed constant . the strategy for proving the proposition is as follows .
we use lemma [ l.hmart ] to conclude that @xmath221=0 $ ] . since @xmath222 is discrete - harmonic , it is approximately equal to the harmonic function on @xmath223 with the corresponding boundary values
. the difference @xmath224 can then be approximated by a function of @xmath225 and @xmath226 .
applying @xmath221=0 $ ] for two distinct choices of @xmath94 then gives the relations ( [ e.hemart ] ) and ( [ e.hevar ] ) .
we start with a lemma describing the approximation of @xmath227 by a ( nondiscrete ) harmonic function .
[ l.harmconv ] given any @xmath189 there is an @xmath228 such that for every vertex @xmath229 and every @xmath230 , if @xmath231 , then @xmath232 where @xmath233 .
note that @xmath234 is harmonic and has the boundary values @xmath4 on @xmath61 and @xmath3 on @xmath235 .
since @xmath236 , @xmath237 is harmonic in @xmath223 , and has boundary values @xmath4 on @xmath101 and on the `` left side '' of @xmath238 $ ] and @xmath3 on @xmath100 and the `` right side '' of @xmath238 $ ] .
since @xmath227 is a discrete harmonic function with similar boundary conditions , the statement of the lemma can be obtained as a consequence of the convergence of random walk on @xmath79 to brownian motion .
we leave the details to the reader .
( also note that more delicate but similar estimates are given in @xcite , section [ sec5 ] . )
proof of proposition [ p.helocal ] assume @xmath239 .
we claim that there is an absolute constant @xmath240 such that @xmath241 if @xmath242 .
let @xmath42 be on the circle @xmath243 .
since @xmath244 , the koebe distortion theorem implies a positive constant lower bound for @xmath245 ( see , e.g. , @xcite , section 1.3 ) .
let @xmath32 be the loewner chain driven by @xmath40
. then @xmath246 . by loewner s equation ( [ e.chordal ] ) ,
@xmath247 , which implies @xmath248 .
thus , @xmath249 .
since @xmath250 , it follows that @xmath251 $ ] if @xmath242 , where @xmath252 is the infimum of all possible values for @xmath253 . in that case ,
@xmath254 , which implies our claim @xmath255 . we will henceforth assume , with no loss of generality , that @xmath242 .
note that the above argument also gives a positive lower bound on @xmath256 .
now fix some vertex @xmath257 satisfying @xmath258 .
let @xmath259 be larger than @xmath260 , in the notation of lemma [ l.harmconv ] .
assume now that @xmath261 holds
. then we may apply ( [ e.harmconv ] ) with @xmath262 , @xmath263 and @xmath264 . since @xmath227 is a martingale
, it satisfies @xmath265 = h_n(w_0)$ ] , and so we get from ( [ e.harmconv ] ) @xmath266\bigr ] = \tilde{h}\bigl(\phi_n(w_0)-w(t_n)\bigr ) + o(\delta^3).\ ] ] below , we need the relations @xmath267
\qquad |w(t)-w(t_n)|= o(\delta),\qquad t_m - t_n = o(\delta^2).\ ] ] by our choice of @xmath208 , we have the first relation when @xmath268 and the second relation when @xmath269 replaces @xmath270 .
the relations ( [ e.aprio ] ) will follow by assuming that @xmath259 is large enough . indeed ,
if @xmath271 and @xmath259 is large , then the harmonic measure from @xmath272 of @xmath273 $ ] in @xmath274 is @xmath275 .
( the beurling projection theorem @xcite tells us that @xmath276 suffices . ) by conformal invariance of harmonic measure , the harmonic measure from @xmath277 of @xmath278 $ ] in @xmath43 is @xmath275 .
we want to use this to conclude that @xmath279)=o(\delta)$ ] .
note that @xmath280 . above
, we have seen that there is a constant positive lower bound for @xmath256 . by ( [ e.chordal ] ) , @xmath281 is monotone decreasing in @xmath28 .
hence , @xmath282 has a constant positive lower bound for @xmath283 . by ( [ e.chordal ] )
, we get @xmath284 for @xmath283 .
integrating then gives @xmath285 for @xmath286 .
as @xmath287 $ ] the distance from @xmath277 to @xmath278 $ ] is @xmath288 .
consequently , the harmonic measure estimate gives the bound @xmath289=o(\delta)$ ] .
the needed estimates ( [ e.aprio ] ) now follow from @xcite , lemma 2.1 , since @xmath290)$ ] is the set of points hitting the real line under loewner s evolution ( [ e.chordal ] ) in the time interval @xmath291 $ ] .
let @xmath292 .
since we have @xmath293 , we may obtain @xmath294 from @xmath295 by flowing according to loewner s equation ( [ e.chordal ] ) between the times @xmath296 and @xmath270 .
as before , we get the bound @xmath297 for @xmath298 $ ] . since @xmath299 , we have @xmath300.\ ] ] by integrating this relation over @xmath301 $ ] , ( [ e.chordal ] ) gives @xmath302 consider now @xmath303 .
we want an estimate for @xmath304 up to @xmath305 terms . for that purpose
, we use a taylor expansion of @xmath306 about @xmath307 .
since @xmath308 and @xmath309 , it suffices to take the terms up to the first derivative of @xmath306 with respect to @xmath42 and the second derivative of @xmath306 with respect to @xmath195 , and no mixed terms .
hence , @xmath310 ( since @xmath42 is complex , @xmath311 is actually a linear map from @xmath81 to @xmath312 . ) by ( [ e.tr ] ) , the conditional expectation of the left - hand side given @xmath313 $ ] is @xmath305 . after calculating the derivatives and applying ( [ e.dz ] ) , we get @xmath314\bigr ] \\ { \noalign { } } & { \displaystyle}{}- { \operatorname{im}}\bigl(\bigl(\phi_n(w_0)-w(t_n)\bigr)^{-1}\bigr ) { \mathbf{e}}\bigl[w(t_m)-w(t_n ) { \vert}\gamma [ 0,n]\bigr ] \\ { \noalign { } } & { \displaystyle}{}- \tfrac12 { \operatorname{im}}\bigl(\bigl(\phi_n(w_0)-w(t_n)\bigr)^{-2}\bigr ) { \mathbf{e}}\bigl [ \bigl(w(t_m)-w(t_n)\bigr)^2{\vert}\gamma [ 0,n]\bigr ] .
\end{array}\ ] ] we now assume that @xmath276 . the koebe distortion theorem ( again ,
see @xcite , section 1.3 ) then implies that a vertex @xmath315 closest to @xmath272 satisfies @xmath316 .
the koebe distortion theorem also shows that a vertex @xmath317 closest to @xmath318 satisfies @xmath319 and @xmath320 .
consequently , we may apply ( [ e.gettingthere ] ) with @xmath321 replaced by each of @xmath322 . with @xmath323 , we get ( [ e.hemart ] ) . now eliminating the term @xmath324 $ ] from ( [ e.gettingthere ] )
[ since it is @xmath305 ] and applying ( [ e.gettingthere ] ) with @xmath323 gives ( [ e.hevar ] ) . [ c.drive ] fix @xmath326 . as @xmath173 ,
the restriction of @xmath327 to @xmath196 $ ] converges in law to the corresponding restriction of standard brownian motion .
let @xmath328 , and let @xmath329\dvtx |w(t/4)|\le{\varepsilon}^{-1}\}$ ] .
let @xmath330 . in order to apply proposition [ p.helocal ] at every @xmath331 , we need to verify @xmath332 for such @xmath6 . from @xcite ,
lemma 2.1 , we get that @xmath211 or @xmath333\}$ ] is unbounded , which implies @xmath334 for @xmath331 .
since @xmath335 and @xmath336 , we have @xmath337 .
we claim that there is a compact subset @xmath338 , which depends only on @xmath339 and @xmath340 , such that @xmath341 holds for each @xmath331 . indeed , @xmath342 flows according to ( [ e.chordal ] ) starting from @xmath343 at @xmath344 to @xmath345 at @xmath346 . for every @xmath347 $ ] , we have , by the monotonicity of @xmath348 with respect to @xmath28 . by ( [ e.chordal ] ) , this shows that @xmath349 .
hence , the bound @xmath350 .
we may therefore take @xmath351 .
since @xmath343 lies in a compact subset of @xmath43 , the koebe distortion theorem implies that @xmath352 .
thus we may assume that @xmath353 holds for every @xmath331 provided we take @xmath354 for some constant @xmath355 .
now the proof that the restriction of @xmath327 to @xmath356 $ ] converges in law to the corresponding stopped brownian motion follows from the proposition and the skorokhod embedding theorem , as in @xcite , section 3.3 .
standard brownian motion is unlikely to hit @xmath357 before time @xmath340 if @xmath339 is small .
thus , we obtain the corollary by taking a limit as @xmath358 .
let @xmath359 denote the hausdorff distance ; that is , for two nonempty sets @xmath360 , @xmath361 [ l.parthaus ] for every @xmath326 and @xmath189 there is some @xmath362 so that if @xmath191 , then there is a coupling of @xmath174 and @xmath176 so that @xmath363,{\tilde{\gamma}}[0,t])\dvtx { 0\le t\le t}\}>{\varepsilon}\bigr]<{\varepsilon}.\ ] ] we know that @xmath176 is a simple path , from @xcite .
let @xmath364 be the @xmath8 loewner chain corresponding to @xmath176 , and let @xmath365 be the brownian motion so that the driving process for @xmath364 is @xmath366 . then @xmath364 is obtained by solving ( [ e.chordal ] ) with @xmath40 replaced by @xmath366 .
let @xmath367 denote the set of points in @xmath368 whose distance from @xmath369 $ ] is @xmath370 .
let @xmath371 .
by continuity of solutions of differential equations , there is some @xmath372 such that if @xmath373\to{\mathbb{r}}$ ] is measurable and satisfies @xmath374\}\le \delta$ ] , then the loewner chain corresponding to @xmath195 satisfies @xmath375 ( this is also easy to verify directly ) .
moreover , the same @xmath376 would apply to every @xmath377 sufficiently close to @xmath378 . by compactness of @xmath367
, there is some @xmath379 , which would work for every @xmath371 .
now , this @xmath376 is random , as it depends on @xmath365 , but it is a.s . positive .
therefore , there is a nonrandom @xmath240 , depending only on @xmath340 and @xmath370 , such that @xmath252 would work for @xmath365 with probability at least @xmath380 . by corollary [ c.drive ] ( and the well - known relation between convergence in law and a.s .
convergence @xcite , theorem 11.7.2 ) , when @xmath169 is sufficiently large , we may couple @xmath1 with standard brownian motion @xmath381 so that @xmath382\}\ge\delta_0\bigr]<{\varepsilon}/2,\ ] ] where now @xmath40 is the driving process for @xmath174 . consequently , when @xmath169 is sufficiently large , @xmath383\cap s(t,{\varepsilon})=\varnothing\bigr ] \ge 1-{\varepsilon}.\ ] ] since @xmath384 $ ] is connected and contains @xmath4 , when @xmath384 $ ] is disjoint from @xmath367 every point in @xmath384 $ ] is within distance @xmath339 from @xmath369 $ ] . now consider a sequence of pairs @xmath385 such that @xmath173 . for each such pair
we take a coupling of the corresponding @xmath384 $ ] and @xmath365 such that @xmath386\}\to0 $ ] in probability .
fix some @xmath24 $ ] .
since the collection of probability measures on the ( compact ) hausdorff space of closed nonempty subsets of @xmath178 is compact under convergence in law , by passing to a subsequence , if necessary , we get a coupling of @xmath365 and a hausdorff limit @xmath387 of @xmath388 $ ] . by the above , @xmath387
is contained in @xmath369 $ ] .
moreover , it is clearly connected .
note that the carathodory kernel theorem ( @xcite , theorem 1.8 ) implies that the maps @xmath389\to{\mathbb{h}}$ ] converge to the normalized conformal map from @xmath390 to @xmath43 .
consequently , the capacity of @xmath387 is @xmath28 a.s .
thus , we conclude that @xmath391 $ ] a.s .
since the limit does not depend on the subsequence , it follows that @xmath392 $ ] is a.s .
the hausdorff limit @xmath387 of @xmath388 $ ] . as @xmath24
$ ] is arbitrary , we conclude that a.s .
@xmath391 $ ] for every rational @xmath28 in @xmath196 $ ] .
the lemma now follows , since @xmath176 is continuous and @xmath393 when @xmath394 .
lemma [ l.parthaus ] gives some form of convergence of @xmath174 to @xmath176 .
our goal now is to improve the quality of the convergence , in two ways .
first , we want to show that the convergence is locally uniform ( i.e. , uniform on compact intervals @xmath196\subset[0,\infty)$ ] ) .
later , it will be shown that the convergence is uniform when we use the metric @xmath188 on @xmath178 . to understand the issues here ,
we describe two examples where one form of convergence holds and another fails .
we start with an example similar to one appearing in @xcite , section 3.4 .
let @xmath189 , @xmath395 and @xmath396 . let @xmath397 be the polygonal path determined by the points @xmath398 , etc .
let @xmath399 be the path @xmath400 , @xmath48 , reparameterized by capacity .
then the path @xmath401 reparameterized by capacity converges to @xmath399 in the sense of lemma [ l.parthaus ] .
moreover , the loewner driving process for @xmath401 converges locally uniformly to the constant @xmath4 , which is the driving process for @xmath402 .
however , one can not reparameterize @xmath401 so that @xmath403 locally uniformly . to illustrate the second issue ,
consider the polygonal path @xmath404 determined by the points @xmath405 , where the last segment can be chosen as any ray from @xmath406 to @xmath30 in @xmath43 .
then @xmath404 , reparameterized by capacity , does converge locally uniformly to @xmath399 .
however , it does not converge uniformly with respect to the metric @xmath188 .
the purpose of this subsection is to develop a tool which will be handy for proving some upper bounds on probabilities of rare events for the he , the discrete excursion measure .
it is a discrete analogue of the ( two - dimensional ) brownian excursion as introduced in @xcite , section 2.4 .
a slightly different variant of the continuous brownian excursion was studied in @xcite .
let @xmath0 be a domain in the plane whose boundary is a subgraph of the triangular lattice @xmath79 .
( we work here with the simple random walk on @xmath79 , but the results apply more generally to other walks on other lattices . )
let @xmath407 denote the set of vertices in @xmath408 .
a directed edge of @xmath79 is just an edge of @xmath79 with a particular choice of orientation ( i.e. , a choice of the initial vertex ) .
if @xmath409 $ ] is a directed edge of @xmath79 , then @xmath410 $ ] will denote the same edge with the reversed orientation .
let @xmath411 denote the set of directed edges of @xmath79 whose interiors intersect @xmath0 and whose initial vertex is in @xmath407 .
let @xmath412 denote the set of directed edges of @xmath79 whose interiors intersect @xmath0 and whose terminal vertex is in @xmath407 ; that is , @xmath413 .
let @xmath414 and @xmath415 .
for every @xmath416 , let @xmath417 be a simple random walk on @xmath79 that starts at @xmath94 and is stopped at the first time @xmath418 such that @xmath419 .
let @xmath420 denote the restriction of the law of @xmath417 to those walks that use an edge of @xmath421 as the first step and use an edge of @xmath422 as the last step .
( this is zero if @xmath94 is not adjacent to an edge in @xmath421 , and generally it is not a probability measure . ) finally , let @xmath423 .
this is a measure on paths starting with an edge in @xmath421 , ending with an edge in @xmath422 and staying in @xmath0 in between .
it will be called the discrete excursion measure from @xmath421 to @xmath422 in @xmath0 . when @xmath424 , we will often abbreviate @xmath425 .
[ l.dxgreen ] let @xmath0 be as above , and let @xmath414 .
fix @xmath144 , and for every path @xmath426 let @xmath427 be the number of times @xmath426 visits @xmath94 .
then @xmath428 where @xmath429 denotes the probability that a simple random walk started from @xmath94 will first exit @xmath0 through an edge in @xmath430 .
in particular , .
let @xmath431 denote the probability space of random walks starting at @xmath94 and stopped when they first exit @xmath0 .
for a pair @xmath432 , let @xmath433 denote the reversal of @xmath434 followed by @xmath435 .
then @xmath2 is a map from @xmath436 to the support of @xmath437 .
clearly , @xmath438 .
if @xmath439 is in @xmath440 , the support of @xmath441 , then the cardinality of the preimage @xmath442 is precisely @xmath443 .
consequently , we have
@xmath444 this proves the claim in the case @xmath445 .
the general case is similarly established .
[ c.hitball ] let @xmath0 and @xmath421 be as above , and let @xmath144 .
assume that @xmath97 is connected .
let @xmath365 be the ball centered at @xmath94 whose radius is @xmath446 , and let @xmath447 be the set of paths that visit @xmath365 .
then @xmath448 for some absolute constant @xmath220 .
it is well known that there is an absolute constant @xmath220 with @xmath449 for every @xmath450 such that @xmath451 is at least half the radius of @xmath365 , where @xmath452 is green s function .
see , for example , @xcite , ( 3.5 ) , where the radius of the ball @xmath365 is different , but the same proof applies . consequently , given that a random walk hits @xmath365 , its expected number of visits to @xmath94 before exiting @xmath0 is between @xmath453 and @xmath220 .
thus , the corollary follows from ( [ e.dxgreen ] ) .
it is also important to note that @xmath454 ; that is , the total mass of @xmath455 is equal to that of @xmath456 .
this is proved by reversing the paths , which gives a measure - preserving bijection between the support of these two measures .
we now return to the setup and notation of section [ s.hesetup ] .
fix some ball @xmath365 that intersects @xmath313 $ ] .
the present goal is to get an upper bound for the conditional probability @xmath457\cap b\ne\varnothing{\vert}\gamma[0,n]]$ ] , under some simple geometric assumptions . to simplify notation , instead of discussing conditional probabilities
, we shall instead obtain a bound on @xmath458\cap b\ne\varnothing]$ ] for a ball @xmath365 intersecting @xmath97 .
the conditional probability estimate then readily follows , because of the markovian property mentioned in remark [ r.hemark ] .
[ p.hitprob ] let @xmath459 .
let @xmath460 be a ball of radius @xmath461 intersecting @xmath97 , and let @xmath462 be the concentric ball with radius @xmath259 .
suppose that there is no component of @xmath463 whose boundary intersects both @xmath101 and @xmath100 .
then @xmath464\cap b(z , r)\ne\varnothing\bigr ] \le o(1 ) ( r / r)^{{\hat{c}}},\ ] ] where @xmath465 is a universal constant
. the reader should think about the case where @xmath97 is rather wild geometrically , as in figure [ f.hitprob ] .
in particular , we need to include the case where @xmath460 intersects both @xmath101 and @xmath100 .
the precise form of the right - hand side of ( [ e.hitprob ] ) will not be important in the following .
what is essential is only that it tends to zero with @xmath466 . if necessary
, one can probably show that @xmath467 works , by using the following proof and a discrete version of the beurling projection theorem .
( some discrete version of the beurling projection theorem is given in @xcite , theorem 2.5.2 , but it is not precisely the same statement that would apply here . ) . ]
proof of proposition [ p.hitprob ] we will assume that @xmath468 .
this involves no loss of generality , because the distance from point on @xmath469 $ ] to @xmath97 is bounded from zero , except near @xmath98 and @xmath99 , and the result clearly holds when @xmath259 is bounded .
let @xmath470 be the event that there is a @xmath471 such that @xmath472 , and on @xmath470 let @xmath473 be the least such @xmath471 .
we assume , with no loss of generality , that @xmath474 , say , since otherwise the statement of the proposition is trivial .
consider the ball @xmath475 . by our assumptions ,
each component of @xmath476 has boundary entirely in @xmath101 or entirely in @xmath100 . on the event @xmath470 ,
let @xmath477 be the connected component of @xmath476 intersecting @xmath478 $ ] .
let @xmath479 be the event that @xmath480 . by symmetry ,
it is enough to prove that @xmath481\le o(1)(r / r)^{{\hat{c}}}$ ] .
let @xmath482 denote the set of directed edges in @xmath411 whose initial vertex is in @xmath483 .
let @xmath484 , where @xmath485 is as in section [ s.heconstruction ] and let @xmath486 denote the set of directed edges connecting vertices in @xmath487 to vertices in @xmath488 .
the reason that the measures @xmath441 are useful here is because the total mass @xmath489 of @xmath490 is a martingale .
one easy way to deduce this is by considering walks that hit the vertex @xmath491 before any other vertex in @xmath492 ( except for the initial vertex of the walk ) . given the part of such a walk up to its first visit to @xmath491 , the probability that it first exits @xmath493 using an edge from @xmath494 is precisely @xmath495 , which is just the probability that @xmath496 .
alternatively , lemma [ l.hmart ] implies that @xmath489 is a martingale , because the total mass of @xmath489 is just a linear combination of the values of @xmath5 on the terminal vertices of @xmath482 .
since @xmath489 is a nonnegative martingale , the optional stopping theorem implies that @xmath497.\ ] ] the bound on @xmath481 $ ] will be proven by estimating @xmath498 and estimating @xmath499 on the event @xmath500 . by our assumption ,
every path starting from an edge in @xmath482 and ending with an edge in @xmath501 which stays in @xmath0 in between must exit @xmath462 .
consequently , @xmath502 , where @xmath503 is the intersection of @xmath0 with @xmath462 ( adjusted to have as its boundary a subgraph of the grid @xmath79 ) , and @xmath504 are the edges connecting vertices in @xmath503 to vertices in @xmath505 . by corollary [ c.hitball ] , under the measure @xmath506 , the expected number of paths that hit @xmath507 is @xmath288 .
for any @xmath508 , a random walk starting near the boundary of @xmath509 has probability bounded away from @xmath4 to exit @xmath0 before hitting @xmath510 .
thus , the probability that a random walk started near the boundary of @xmath507 will hit @xmath511 before exiting @xmath0 is @xmath512 , for some constant @xmath465 .
hence , @xmath513 consider now the case where @xmath500 happens .
as @xmath514 $ ] crosses the annulus @xmath515 , there must be an arc among the connected components of @xmath516 $ ] that connects the right side of @xmath514 $ ] ( where @xmath131 takes the value @xmath3 ) to the boundary of @xmath477 ( where @xmath131 takes the value @xmath4 ) .
by considering vertices along this arc , we can find a vertex @xmath94 close to @xmath517 from which the ratio between @xmath518 and @xmath519 is bounded and bounded away from zero by universal constants ( because these quantities do not vary by more than a constant factor when moving from a vertex to its neighbor ) , or else there is an edge in @xmath520 . in the latter case ,
clearly @xmath521 is bounded away from zero .
consider therefore the case where such a @xmath94 exists . since random walk starting from @xmath94
has probability bounded away from zero to complete a loop going around the annulus @xmath522 before exiting it , we have @xmath523 bounded away from zero .
consequently , each of these summands is bounded away from zero .
let @xmath365 be the ball centered at @xmath94 whose radius is half the distance from @xmath94 to @xmath524 . by corollary [ c.hitball ] ,
the measure under @xmath525 of the set of paths hitting @xmath365 is bounded away from zero .
since @xmath526 is bounded away from zero , a random walk started at any vertex in @xmath365 has probability bounded away from zero to exit @xmath527 in @xmath528 , by the harnack principle ( e.g. , @xmath529 in @xcite , lemma 5.2 ) .
consequently , we see that also in this case @xmath530 is bounded away from zero on the event @xmath500 by an absolute constant .
combining this with ( [ e.optional ] ) and ( [ e.dx0 ] ) establishes @xmath481=o(r / r)^{{\hat{c}}}$ ] .
the proof for the event @xmath531 is entirely symmetric .
[ p.locunif ] in the setting of theorem [ t.heunifconv ] , for every fixed @xmath20 , there is a coupling of @xmath174 and @xmath176 so that @xmath532 in probability as @xmath173 .
consider a sequence of pairs @xmath385 , with @xmath173 .
using lemma [ l.parthaus ] , we couple @xmath176 and the sequence @xmath174 so that for each @xmath48 the set @xmath392 $ ] is a.s .
the hausdorff limit of @xmath388 $ ] .
our strategy will be to prove that the curves @xmath533 $ ] converge to @xmath534 $ ] in the hausdorff sense for all rational pairs @xmath535 ( for notational convenience , we may assume @xmath340 is also rational ) .
we will prove this , in turn , by showing that for all rationals @xmath536 , the hausdorff limit of @xmath537 $ ] is a.s .
disjoint from both @xmath538 $ ] and @xmath539 $ ] .
we will begin by restricting our attention to a large compact set and using proposition [ p.hitprob ] to derive upper bounds for the probability that @xmath540 comes close to @xmath541 , where @xmath542 are chosen below so that @xmath541 is `` well exposed '' and so that the assumptions of proposition [ p.hitprob ] apply .
let @xmath189 , and let @xmath306 be some compact subset of @xmath43 .
for @xmath543 , let @xmath544 be the first @xmath545 such that @xmath546 is in the unbounded connected component of @xmath547)\ge{\varepsilon}\}$ ] .
fix some @xmath548 , and let @xmath549 be the event that @xmath550 and
@xmath551 . we claim that if @xmath169 is sufficiently large , then @xmath552\le c_f ( \delta/{\varepsilon})^{{\hat{c}}},\ ] ] where @xmath553 is the same constant as in proposition [ p.hitprob ] , and @xmath554 is a constant depending only on @xmath306 . by the koebe distortion theorem
, there is a constant @xmath555 such that for every @xmath556 with @xmath557 we have @xmath558 [ in other words , the metric in @xmath306 is comparable to the metric in @xmath559 scaled by @xmath560 .
] let @xmath6 be the least integer such that @xmath561)\supset{\gamma^\phi}[0,s(s(t))]$ ] .
( recall that the parameterization of @xmath1 is not by capacity , but is proportional to arclength . )
now condition on @xmath313 $ ] , and assume that @xmath562 .
( if that has zero probability , then @xmath563=0 $ ] . )
let @xmath564 .
assume that @xmath169 is sufficiently large so that @xmath565 ; indeed , how large @xmath169 is required to be can be determined from the constant @xmath220 in ( [ e.metriccompare ] ) .
the metric comparison ( [ e.metriccompare ] ) implies that @xmath566 .
this and the of @xmath567 imply that there is a path in @xmath568\cup b(z_0,{\rho}{\varepsilon}/(2c)))$ ] connecting @xmath569 and @xmath99
. we may then apply proposition [ p.hitprob ] with @xmath0 replaced by the @xmath158 of remark [ r.hemark ] to conclude that conditioned on @xmath313 $ ] the probability that @xmath570 $ ] intersects @xmath571 is @xmath572 .
now ( [ e.aprob ] ) follows by another application of ( [ e.metriccompare ] ) .
let @xmath573 denote the hausdorff limit of @xmath574 $ ] , when it exists .
note that we may pass to a subsequence of pairs @xmath385 so that @xmath574 $ ] converges in law .
consequently , by using @xcite , theorem 11.7.2 again , we may assume that almost surely the limits @xmath573 , @xmath575 , @xmath576 and @xmath577 exist for every pair of rationals @xmath578 .
let @xmath579 be rational , and let @xmath580 be the set of @xmath581 such that @xmath582 is in the unbounded component of @xmath583)\ge b\}$ ] . by construction , it is clear that @xmath584)={\varepsilon}$ ] , @xmath585 $ ] , @xmath586 and @xmath587 .
let @xmath588 be a rational satisfying @xmath589 .
now , the following a.s .
statements will hold on the event @xmath590 and @xmath591 .
first , by ( [ e.aprob ] ) , we have a.s .
@xmath592 . note that @xmath593\setminus{\tilde{\gamma}}[0,t]$ ] . since @xmath176 is a simple path , @xmath369\setminus p_t$ ] has two disjoint components , one containing @xmath392 $ ] , and the other containing @xmath594 .
since @xmath595 is connected and contains @xmath594 , we conclude that a.s .
@xmath596=\varnothing$ ] [ on the event @xmath591 , @xmath597 .
now let @xmath598 be any rational in @xmath599 .
since @xmath176 is a simple path , there is some small @xmath189 such that @xmath600 . because @xmath306 is an arbitrary compact subset of @xmath43 , @xmath601 a.s . and
in the above discussion @xmath189 was arbitrary , it follows that @xmath602=\varnothing$ ] a.s . since @xmath603 $ ] and is a.s .
disjoint from @xmath604\cup{\tilde{\gamma}}[t_1,t]$ ] whenever @xmath605 ( by the above ) , it follows that @xmath606 $ ] a.s . for every pair of rationals @xmath607
consequently , @xmath606 $ ] a.s . for every pair @xmath608 ( the cases @xmath344 and @xmath609 are similarly treated ) .
thus , @xmath610 } |{\gamma^\phi}(t)-{\tilde{\gamma}}(t)|=0 $ ] .
since every sequence of @xmath385 with @xmath173 has a subsequence such that this holds , this also holds without passing to a subsequence .
the proposition is thus established . since @xmath611
, @xmath174 is transient ; that is , @xmath612 .
the following is a uniform version of this statement .
[ p.uniftrans ] for every @xmath189 and @xmath613 there is a @xmath614 such that @xmath615<{\varepsilon}\ ] ] if @xmath169 is sufficiently large .
the reader may note the clear similarity with proposition [ p.hitprob ] .
the main difference is that there the path considered was in the domain @xmath0 , whereas here the path @xmath174 is in the image under the conformal map , that is , in the upper half plane @xmath43 . indeed
, the proof is quite similar .
the following lemma about the excursion measures will be needed .
[ l.phidx ] let @xmath613 .
suppose that at each vertex @xmath616 a simple random walk @xmath417 is started , and the walk is stopped at the first time @xmath418 such that @xmath617 exits @xmath0 or @xmath618 .
then the expected number of walks which stop when @xmath619 is @xmath288 .
let @xmath620 .
for @xmath621 , let @xmath622 denote a simple random walk on @xmath623 , where at each step the walk jumps with equal probability along each of the edges @xmath624 with @xmath625 and @xmath626 .
let @xmath627 .
let @xmath628 be the expected number of walks @xmath622 with @xmath629 such that @xmath630 , where @xmath631 .
it clearly suffices to show that @xmath632 .
( the difference from the @xmath417 is that the @xmath622 are reflected off of @xmath97 , rather than killed there . ) for a function @xmath2 on @xmath623 and @xmath621 , let @xmath633 } ( f(u)-f(v))$ ] , where the sum extends over edges @xmath634 $ ] containing @xmath94 such that @xmath625 . now
let @xmath635 $ ] be the unique function such that @xmath636 for @xmath637 , @xmath638 on @xmath88 and @xmath639 on @xmath640 ; that is , @xmath641 is harmonic in @xmath642 with the appropriate boundary values .
it is immediate to verify that @xmath643 where @xmath644 is the number of edges containing @xmath94 in @xmath106 , since @xmath645 is the probability that @xmath622 hits @xmath640 before @xmath88 . because @xmath646 , we have @xmath647 let @xmath648 } ( f(v)-f(u))^2 $ ] , where the sum runs over all edges in @xmath106 . it is well known ( and simple to show ) that @xmath641 minimizes @xmath430 among functions mapping @xmath640 to @xmath3 and @xmath88 to @xmath4 . for each edge
@xmath649 $ ] , we may distribute the quantity @xmath650 by giving @xmath651 to the vertex @xmath94 and @xmath652 to the vertex @xmath653 .
consequently , by summing over the contributions to each vertex , we find @xmath654 hence , @xmath655 , which gives @xmath656 .
there are several different ways to estimate @xmath657 and complete the proof .
we opt for an easy and short argument , which unfortunately does require terminology and results from the literature .
( similar arguments appear , e.g. , in @xcite . )
let @xmath658 be the annulus @xmath659 .
let @xmath660 when @xmath661 and @xmath662 when @xmath663 , @xmath664 .
then @xmath665 , by the characterization of @xmath641 as a minimizer .
following @xcite , we say that a set @xmath666 is @xmath378-fat if for every disk @xmath667 with @xmath668 and @xmath669 , we have @xmath670 . consider a triangle @xmath671 of the grid @xmath79 with @xmath672 .
then @xmath673 is a @xmath674-quasidisk , for some constant @xmath674 , by @xcite .
[ there , it is required that @xmath675 , but @xmath676 works too , by standard compactness properties . besides , it suffices for the argument given below that @xmath677 is a @xmath674-quasidisk when @xmath678 is a slightly rescaled copy of @xmath671 contained in the interior of @xmath671 , provided that @xmath674 does not depend on the scaling factor . ) consequently , by @xcite , corollary 2.3 , @xmath673 is @xmath378-fat for some constant @xmath679 .
thus , @xmath680 .
hence , @xmath681 , where the sum extends over all the triangles of the grid @xmath79 that are contained in @xmath0 .
now , if @xmath682 $ ] is an edge in @xmath106 , then there is a grid triangle @xmath672 with @xmath682\subset{\partial}\triangle$ ] .
we have then @xmath683 .
since each triangle has three edges , we get @xmath684 this completes the proof . proof of proposition [ p.uniftrans ] we choose @xmath685 much larger than @xmath259 .
let @xmath686 , and let @xmath340 be a constant such that @xmath687<{\varepsilon}/3.\ ] ] then , by lemma [ l.parthaus ] , when @xmath169 is sufficiently large , we have @xmath688<{\varepsilon}/3 $ ] .
consequently , it suffices to show that @xmath689<{\varepsilon}/2.\ ] ] let @xmath208 be the least integer such that @xmath690\supset{\gamma^\phi}[0,t_1]$ ] .
the proof now proceeds as in proposition [ p.hitprob ] , with only minor changes , which will be henceforth described .
let @xmath470 be the event that there is a @xmath691 such that @xmath692 , and on @xmath470 let @xmath473 be the least such @xmath471 .
as in the proof of proposition [ p.hitprob ] , the event @xmath479 is defined . for integer @xmath693 $ ] , we consider the excursion measure in @xmath694 with excursions started at the vertices in @xmath695 and terminating at vertices in @xmath696 .
the total mass of this measure is a martingale .
it suffices to show that this is very small at @xmath697 , but is bounded away from zero at @xmath698 on the event @xmath500 .
we first do the estimate for @xmath697 .
the expected number of excursions in @xmath0 from @xmath699 that hit @xmath700 is the same as the number of excursions in the domain which consists of the grid triangles intersecting @xmath701 starting at vertices in @xmath702 that hit @xmath699 , by symmetry , and this quantity is bounded by the number of excursions in the domain which is essentially @xmath703 starting at vertices in @xmath702 that hit @xmath704 . by symmetry again , this is the same as the expected number of excursions in the reverse direction .
this quantity is @xmath288 , by lemma [ l.phidx ] .
consequently , the expected number of excursions from @xmath699 in @xmath0 that cross @xmath700 is @xmath288 .
it is not hard to see that when @xmath169 is sufficiently large , there will not be any grid edge crossing both @xmath700 and @xmath705 , for example , by considering the harmonic measure from @xmath706 of such an edge .
now , @xcite , lemma 5.4 tells us that a random walk started in @xmath707 has probability @xmath708 to exit @xmath709 before exiting @xmath0 , uniformly as @xmath710 .
( that lemma refers to the square grid , but the proof applies here as well .
also , in that lemma the image conformal map is onto the unit disk @xmath711 , but this is simply handled by choosing an appropriate conformal homeomorphism from @xmath711 to @xmath43 . ) it remains to prove a bound from below for measure of excursions in @xmath712 from @xmath699 to @xmath713 , on the event @xmath500 .
as in the proof of proposition [ p.hitprob ] , it suffices to find a vertex @xmath94 such that the discrete harmonic measure in @xmath714 $ ] from @xmath94 of each of the sets @xmath699 and the left side of @xmath514 $ ] is bounded from below .
consider any vertex @xmath715 near @xmath716 .
the continuous harmonic measure from @xmath717 of @xmath312 in the domain @xmath718 is bounded from below . by the convergence of discrete harmonic measure to continuous harmonic measure
, when @xmath169 is large , a random walk started at @xmath715 will have probability bounded from below to hit @xmath97 before exiting @xmath719 .
( specifically , while not close to the boundary , the random walk behaves like brownian motion , which is conformally invariant . once it does get close to the boundary , we may apply @xcite , lemma 5.4 , say . ) as in the proof of proposition [ p.hitprob ] , on @xmath500 we can find a vertex @xmath94 near @xmath716 where the discrete harmonic measure of @xmath699 is comparable to that of @xmath713 .
hence , both are bounded away from zero .
this completes the proof .
proof of theorem [ t.heunifconv ] the theorem follows immediately from propositions [ p.locunif ] and [ p.uniftrans ] .
we wish to thank richard kenyon and david wilson for inspiring and useful conversations . | the harmonic explorer is a random grid path .
very roughly , at each step the harmonic explorer takes a turn to the right with probability equal to the discrete harmonic measure of the left - hand side of the path from a point near the end of the current path .
we prove that the harmonic explorer converges to sle(4 ) as the grid gets finer . and . |
quantum computation and information theory are research areas which have developed amazingly fast .
this rapid and accelerated development is due to the promises of qualitatively new modes of computation and communication based on quantum technologies , that in some cases are much more powerful than their classical counterparts , and also due to the impact that these new technologies can provide to society .
advances occur in different directions and the theoretical proposals as well as the experimental achievements involves fundamental concepts that live at the heart of quantum mechanics .
one of the most remarkable examples is the measurement - based quantum computation @xcite .
while the standard quantum computation is based on sequences of unitary quantum logic gates , the measurement - based quantum computation is realized using only local conditioned projective measurements applied to a highly entangled state called the cluster state . as it is entirely based on local measurements instead of unitary evolution ,
the computation is inherently irreversible in time .
recently , this one - way quantum computation has attracted a lot of attention of the scientific community @xcite and has been studied considering : ( i ) various important aspects which might influence it , such as decoherence during the computation @xcite and ( ii ) novel schemes of implementation @xcite .
in addition , since the technical requirements for the latter can be much simpler than those for the standard circuit model , it has been realized in several experiments @xcite . the noise process , that emerges from the inevitable interaction of the qubits with their environment , remains one of the major problems to be overcome before we shall be able to manufacture a functional quantum computer .
the interaction of the qubits with the environment , which depends on the quantum computer architecture , is a crucial problem that is far from being well understood .
however , it is known that this interaction can result in non - monotonical dynamics of the density matrix coherences @xcite and a complete understanding of this behavior is a matter of great importance . moreover , this non - monotonical behavior , is a subject of broad interest @xcite , since this peculiar property brings up unexpected dynamics for quantum fidelity @xcite as well as for quantum entanglement @xcite and quantum discord @xcite .
when an open quantum system interacts with the outside world there are two main effects that have to be considered : the relaxation and the decoherence processes .
the relaxation process is associated with an expected loss of energy of the initial state of the system which happens at the rate @xmath1 , where the time scale @xmath2 is known as the relaxation time scale of the system . on the other hand ,
the decoherence process is associated with the reduction of purity of this physical state and takes place within a time scale @xmath3 .
depending on the physical system one considers , the time scale @xmath3 can be much shorter than @xmath2 , making quantum computers more sensitive to decoherence than to relaxation process .
this is exactly the situation we are going to address in this paper .
based on the foregoing reasoning , we use an exact solvable model @xcite to calculate the dissipative fidelity dynamics in a linear measurement - based quantum computer ( mbqc ) @xcite composed by few qubits which interact collectively with a dephasing environment .
although this kind of model does not describe relaxation processes , it does indeed adequately describe decoherence effects .
the interaction hamiltonian is given by @xmath4 , where @xmath5 is the total azimuthal angular momentum in a system of @xmath0 qubits , and @xmath6 is the operator that acts only on the environmental degrees of freedom . we show that for any initial state given by an eigenstate superposition of the @xmath7 operator , whose eigenvalues are different in modulus , the fidelity exhibits a non - monotonical character . to be more precise , suppose we have a system whose initial state can be written in a suitable basis as @xmath8 where @xmath9 , with @xmath10 ; if the vectors that characterize the state @xmath11 are such that their @xmath7 eigenvalues are all equal in modulus , i.e. , if their @xmath12 s are all equal , the fidelity dynamics does not exhibit a non - monotonical shape , but if at least one vector that composes the state @xmath11 has a @xmath12 different from the other vectors we can observe a non - monotonical behavior of the fidelity dynamics .
this condition reveals itself to be necessary for the fidelity to present an oscillatory shape , and , remarkably , it does not depend on the initial entanglement ( depending only on the initial configuration of the state of the system ) .
furthermore , we study the implications of this oscillatory behavior to the mbqc where a sequence of local projective measurements are applied on the qubits to implement a quantum gate . as we shall see , we can take advantage of the revival times of the coherence of the cluster state to apply the projective measurements at times such that we get the best gate fidelity values .
therefore , we can choose the best possible instants that produce the best possible results . in this sense , fast measurements can result worse than slow , but conveniently applied measurements .
that is , under the action of a common dephasing environment , the mbqc can provide better computational fidelities even for slow measurement sequences .
this result , as we will see bellow , is a natural consequence of the oscillatory behavior of the density - matrix coherences . to illustrate our finding we examine the fidelity of some single qubit quantum gates that are frequently found in the literature @xcite developed using the mbqc scheme .
this manuscript is organized as follows . in section
ii we describe the exact dissipative dynamics of the @xmath0-qubit system interacting with a common dephasing environment @xcite . in section
iii we show the necessary condition for the non - monotonical behavior of the fidelity to take place and , in section iv , we present the implications of our results to the mbqc .
we conclude in section v.
consider the following hamiltonian of a system composed of @xmath0 qubits interacting with a common dephasing environment : @xmath13 where the first two terms account for the free evolution of the qubits and the environment , and the third term describes the interaction between them .
the environment operators , @xmath14 and @xmath15 , are the customary creation and annihilation operators which follow the heisenberg s algebra @xmath16 = \delta_{k , k'}$ ] .
the qubit operator @xmath17 is the pauli @xmath18 operator of the @xmath19-th qubit .
the complex constant @xmath20 has dimension of frequency and indicates the coupling between the qubits and the field modes .
in addition , @xmath21 is the difference of energy between the ground and excited levels of the @xmath19-th qubit and @xmath22 is the energy associated with the @xmath23-th mode of the field , with @xmath24 and @xmath25 being , respectively , the transition frequency and the field frequency of the @xmath23-th mode .
we assume that initially the state of the qubits and the state of the environment are decoupled , i.e. , the total initial state of the system and the environment ( which we call the combined system from now on ) can be represented by a density operator given by the following tensor product @xmath26 , where @xmath27 and @xmath28 stand for the initial state of the qubits and the initial state of the environment , respectively .
we also assume that the environment is initially in thermal equilibrium : @xmath29 where @xmath30 is the environment hamiltonian , @xmath31 $ ] is the partition function , and @xmath32 , with @xmath33 representing the boltzmann constant and @xmath34 being the environment temperature .
now , let us rewrite the hamiltonian of eq .
( [ edd.1 ] ) as a sum of two terms , @xmath35 , where @xmath36 and @xmath37 . in the interaction picture ,
@xmath38 , that is , @xmath39 where @xmath40 .
moreover , the time evolution operator is given by @xmath41 where @xmath42 is the dyson time - ordering operator .
if we substitute eq .
( [ edd.3 ] ) into eq .
( [ edd.4 ] ) , we obtain , after some algebra , the following expression for the time evolution operator @xcite : @xmath43\right\}}\nonumber \\ & \times & \exp\left\{i\sum_{k}\sum_{m , n } \vert g_{k}\vert^{2}\sigma^{\left(m\right)}_{z}\sigma^{\left(n\right)}_{z}s\left(\omega_{k},t\right)\right\ } , \label{edd.5}\end{aligned}\ ] ] where @xmath44 and @xmath45 . the dynamics of the @xmath0-qubit system can be written in terms of the density operator of the combined system as follows @xmath46 . \label{edd.6}\ ] ] hence , the matrix elements of the reduced density matrix can be expressed as @xmath47 where @xmath48 refers to the @xmath0-qubit state .
here we have @xmath49 , i.e. , the @xmath50 are the eigenvalues of the @xmath17 pauli operator associated with the two - level qubit states @xmath51 and @xmath52 , and @xmath53 are the eigenvalues of the @xmath17 pauli operator associated with the respective dual space @xmath54 and @xmath55 .
after some manipulations , we finally obtain the explicit dynamics of the elements of the density matrix of the @xmath0-qubit system @xcite : @xmath56^{2}\right\ } \nonumber \\ & \times & \exp\left\{i\theta\left(t\right)\left[\left(\sum^{n}_{n=1}i_{n}\right)^{2 } - \left(\sum^{n}_{n=1}j_{n}\right)^{2}\right]\right\}\rho^{q}_{\left\{i_{n},j_{n}\right\}}\left(0\right ) , \nonumber \\ & & \label{edd.12}\end{aligned}\ ] ] where @xmath57 and @xmath58 with @xmath59 . in the continuum limit ,
( [ edd.12 ] ) reads @xmath60^{2}\right\ } \nonumber \\ & \times & \exp\left\{i\theta\left(t\right)\left[\left(\sum^{n}_{n=1}i_{n}\right)^{2 } - \left(\sum^{n}_{n=1}j_{n}\right)^{2}\right]\right\}\rho^{q}_{\left\{i_{n},j_{n}\right\}}\left(0\right ) , \nonumber \\ & & \label{edd.15}\end{aligned}\ ] ] where @xmath61 @xmath62 and @xmath63 is the environment spectral density .
this function has a cutoff frequency @xmath64 , whose value depends on the environment and @xmath65 for @xmath66 .
here we should stress that since we are going to analyze average values involving the reduced density operator - the fidelity in our particular case - we can safely use this reduced state of the system in the interaction picture . in the schrdinger picture
there are additional terms oscillating with frequency @xmath67 in the off - diagonal elements of that operator .
in our model we assume an ohmic spectral density , @xmath68 where @xmath69 is a dimensionless proportionality constant that characterizes the coupling strength between the system and the environment . substituting eq .
( [ edd.17a ] ) into eq .
( [ edd.16 ] ) and eq .
( [ edd.17 ] ) we obtain @xmath70 and @xmath71 the result of the integration in eq .
( [ edd.17b ] ) is also well - known @xcite and reads : @xmath72 where @xmath73 .
it is easy to note , from eqs .
( [ edd.15 ] ) and ( [ edd.17d ] ) , that the decoherence effects arising from thermal noise can be separated from those due to the vacuum fluctuations .
this separation allows for an exam of different time scales present in the dynamics .
the shortest time scale is determined by the cutoff frequency ( see eq .
( [ edd.17c ] ) above ) , @xmath74 .
the other natural time scale , @xmath75 , is determined by the thermal frequency @xmath76 ( see eq .
( [ edd.17d ] ) above ) which is related to the relaxation of the off - diagonal elements of the density operator .
with these two time scales we can define two different regimes of time @xcite : the _ thermal _ regime and the _ quantum _ regime .
thermal effects will affect the @xmath0-qubit system predominantly only for times @xmath77 whereas the quantum regime dominates over any time interval @xmath78 such that @xmath79 , when the quantum vacuum fluctuations contribute predominantly .
besides , we can see from eqs . ( [ edd.17c ] ) and ( [ edd.17d ] ) that for a sufficiently high - temperature environment , i.e. , @xmath80 , the phase damping factor , which is the main agent responsible for the decoherence , behaves as @xmath81 causing an exponential decay of the off - diagonal elements of the density operator .
moreover , as the phase factor @xmath82 implies an oscillation with frequency @xmath83 , this time evolution is always slightly underdamped .
notice that one should never reach the overdamped regime since as @xmath84 is the shortest time scale in the problem it does not make any sense to make @xmath85 . in the low temperature limit ,
when @xmath86 , the relaxation factor behaves as @xmath87 which leads the off - diagonal matrix elements to an algebraic decay of the form @xmath88 . in this case
, what we have called above the thermal regime is only reached for very long times , @xmath89 .
as we will show below , the quantum regime implies a very different dynamics of the fidelity and shows up peculiar results to the mbqc , where delayed measurements can result in better computation fidelities .
in this section we are interested to know when the fidelity dynamics of an @xmath0-qubit system , interacting collectively with a dephasing environment , will oscillate in time .
we introduce , for the quantum regime , a necessary condition for the non - monotonical behavior to be present . here , since we always consider a pure state as our initial condition , the fidelity as a function of time , in the interaction picture , is given by @xmath90 the dynamics of qubits interacting with a common environment is strongly dependent upon the initial condition , and we will show that , for the quantum regime , the fidelity will always present a non - monotonical behavior when the @xmath0-qubit system is written as a coherent superposition of @xmath7 eigenstates whose eigenvalues are different in modulus . as we can see from eq .
( [ edd.15 ] ) , the second exponential term is responsible for the oscillations and it is identical to the unity when @xmath91 , i.e. , when @xmath92 . note that @xmath93 are the eigenvalues of the total pauli operator associated with the eigenstates @xmath94 .
thus , if the initial state of the @xmath0-qubit system is a coherent superposition of eigenstates of the @xmath7 operator , whose eigenvalues are equal in modulus , the condition @xmath92 is automatically satisfied and the fidelity dynamics does not oscillate at all . on the other hand , a state of @xmath0 qubits that is not written in this way , i.e. , a state that is written as a coherent superposition of the @xmath7 eigenstates whose eigenvalues are not all equal in modulus ( e.g. , if exist at least one eigenvalue different from the others in modulus ) , has a fidelity which indeed oscillates in time . consequently , in the quantum regime , the condition @xmath95 is a necessary condition for the non - monotonical behavior of the fidelity to take place .
it is important to emphasize that this behavior is intrinsic to the geometry of the initial condition , that is , it depends only on the basis vectors spanning the initial state , and this property is not correlated with the initial entanglement .
a simple example is the two - qubit state given by @xmath96 although disentangled , the state is written as a coherent superposition of eigenstates of the @xmath7 operator whose eigenvalues have different moduli and , therefore , its fidelity oscillates in time following the equation below @xmath97 .
\label{fd.16}\ ] ]
from now on we will be concerned with the mbqc fidelity dynamics where the cluster states are subject to a dephasing channel .
we will show how our previous result brings crucial consequences to the computational outcomes we can obtain , depending on the moment we decide to apply our set of projective measurements . to elucidate these aspects we analyze some common single qubit gates @xcite under the mbqc scheme @xcite . following reference @xcite ,
an arbitrary rotation can be achieved in a chain of five disentangled qubits , @xmath98 where the qubits 2 to 5 are initially prepared in the state @xmath99 while the qubit 1 is prepared in some input state which is to be rotated .
we adopt the most general form @xmath100 for the input state , where @xmath101 and @xmath102 are complex numbers that satisfy the relation @xmath103 . to obtain the desired cluster state , the state ( [ fd.17 ] )
becomes entangled by an unitary operation @xmath104 ( see references @xcite and @xcite for further details ) @xmath105 where @xmath106 and @xmath107 .
the state @xmath108 is rotated by measuring the qubits 1 to 4 one by one , and , at the same time that these measurements disentangle completely the cluster state ( [ ap2.5 ] ) , they also implement a single qubit gate , printing " the outcome on qubit 5 .
an arbitrary rotation usually requires three angles in the euler representation , the euler angles , and it can be seen as a composition of three other rotations of the form : @xmath109 , where @xmath110 . in the mbqc scheme , the qubits @xmath111
are measured in appropriately chosen bases @xmath112 .
of course each measurement can possibly furnish two distinct results , up " ( if the qubit @xmath113 is projected onto the first state of @xmath114 ) or down " ( if the qubit @xmath113 is projected onto the other state of @xmath114 ) , and the choice of the basis to measure the subsequent qubit depends on the previous results . in our examples we suppose ( without incurring the risk of weakening our scheme ) that all measurements give the result _ up _ , which is not an event too rare , with the probability @xmath115 of occurrence . in this case , the first projector acting on the first qubit will be necessarily @xmath116 , irrespectively of the single qubit gate that we want to execute . however , the other three projectors @xmath117 , @xmath118 and @xmath119 that act on qubits 2 , 3 and 4 are dependent on the specific choice of the single qubit gate , and they can be specified by the three euler angles in terms of the first state of each one of the bases @xmath120 , @xmath121 and @xmath122 , respectively . after the first measurement , the resulting four qubit state that evolves under the influence of the environment is given by @xmath123 , where @xmath124 since the state ( [ owqc.2 ] ) is written as a combination of eigenstates of the @xmath7 with eigenvalues which are different in modulus , the fidelity dynamics associated with this state is an oscillatory function of time and is written as @xmath125+\frac{3}{8}{e^{-4\gamma\left(t , t\right)}}\cos\left(4\theta\left(t\right)\right ) + \left[\frac{1}{16}-\frac{1}{32}\left(\alpha^{*}\beta+\alpha\beta^{*}\right)^{2}\right]{e^{-36\gamma\left(t , t\right)}}\cos \left(12\theta\left(t\right)\right)\nonumber\\ & + & \left[\frac{1}{16}+\frac{1}{32}\left(\alpha^{*}\beta+\alpha\beta^{*}\right)^{2}\right]{e^{-4\gamma\left(t , t\right)}}\cos\left(12\theta\left(t\right)\right ) + \left[\frac{1}{128}-\frac{1}{128}\left(\alpha^{*}\beta+\alpha\beta^{*}\right)^{2}\right]{e^{-64\gamma\left(t , t\right)}}\nonumber\\ & + & \left[\frac{1}{8}-\frac{1}{32}\left(\alpha^{*}\beta + \alpha\beta^{*}\right)^{2}\right]{e^{-16\gamma{\left(t , t\right ) } } } + \frac{5}{128}\left(\alpha^{*}\beta+\alpha\beta^{*}\right)^{2}+\frac{35}{128}.\nonumber\\ \label{qwqc.3}\end{aligned}\ ] ] the state fidelity ( [ qwqc.3 ] ) will present an oscillatory behavior in the quantum regime ( @xmath126 ) where the exponential decay factor ( [ edd.17d ] ) plays a minor role compared with the oscillatory factor ( [ edd.17c ] ) and the quantum vacuum fluctuations contributes predominantly .
although eq .
( [ qwqc.3 ] ) is written in terms of an input state that is a function of complex coefficients , there is no loss of generality if we regard them as real numbers . in fig .
[ statefidelity ] we show the state fidelity dynamics ( [ qwqc.3 ] ) in the quantum regime assuming @xmath127 and @xmath128 , i.e. , we choose the state @xmath129 as our input state . in the quantum regime for @xmath127 ,
[ statefidelity ] shows that the state fidelity oscillates between maximum values , such as @xmath130 and @xmath131 , at times @xmath132 and @xmath133 ( where we have peaks ) and minimum values , such as @xmath134 and @xmath135 , at times like @xmath136 and @xmath137 ( where we have valleys ) . ) for @xmath127 in the quantum regime with @xmath82 and @xmath138 given by equations ( [ edd.17c ] ) and ( [ edd.17d ] ) respectively . as we can see
the oscillations is a characteristic feature in the quantum regime when the state is written as a coherent superposition of eigenstates of @xmath7 , whose eigenvalues are different in modulus .
here we consider @xmath139 , @xmath140 , and @xmath141 . ] but this oscillatory behavior is not a privilege of the input states @xmath142 or @xmath143 , and more general input states will also present a qualitatively similar oscillatory behavior of the state fidelity dynamics , as we can see directly from eq .
( [ qwqc.3 ] ) . with this in mind , what can we say about the fidelity of quantum computation in this peculiar regime ?
the question above is relevant in the sense that in any realistic experimental realization , the construction of the state ( [ fd.17 ] ) , the unitary operation of entanglement @xmath104 , as well as the four subsequent projective measurements , are made in a finite time interval rather than instantaneously .
hence , it is worth analyzing the computation when the system is subject to the deleterious effects caused by the environment . here , we assume for simplicity that our initial state is given by ( [ owqc.2 ] ) ,
that is , our state at @xmath144 is the cluster state ( [ ap2.5 ] ) prepared to perform the one - way quantum computation with the first measurement @xmath145 already applied on qubit @xmath146 .
the subsequent measurements @xmath117 , @xmath118 and @xmath119 are supposed to be applied on qubits @xmath147 , @xmath148 , and @xmath149 , respectively , in two different scenarios : in one of them the subsequent measurements are applied in sequence and at different instants of time , i.e. , at time @xmath150 we apply the measurement @xmath117 , at time @xmath151 we apply the measurement @xmath118 and finally at time @xmath152 we apply the measurement @xmath119 ( between the times @xmath153 , @xmath154 and @xmath155 the system evolves according to eq .
( [ edd.15 ] ) ) ; in the other scenario the measurements are applied in sequence ( @xmath156 ) but practically at the same time a time that we call @xmath157 ( see fig .
[ illustrative_scheme ] ) .
[ h ] we consider that the five - qubit state is already entangled and each qubit is ready to be measured . besides , the first qubit is also projected at @xmath144 . in ( a )
we suppose that the three subsequent measurements are applied at different instants of time and the qubits evolve non - unitarilly between the measurements . in ( b ) , after wait a time gap , the other three subsequent measurements are made instantaneously at @xmath158 .
the result of the computation is printed " in the fifth qubit state ( in green).,title="fig : " ] with these two different scenarios in mind we are able to show our main result . analyzing the implications of the state fidelity oscillations for the mbqc , we could verify that delayed measurements can in fact give better computational fidelity outcomes . as a matter of fact , there are time slots where we obtain better or worse computational fidelities defining periodic optimum waiting times . to clarify this assumption we analyze three different one qubit gate fidelities : the not gate , the hadamard gate and the phase gate .
primarily , let us consider the first scenario where the measurements that characterize the specific one qubit gate are performed at different times and the state evolves non - unitarilly between measurements . to begin with , we consider a not gate acting on an input state given by @xmath159 .
for this particular example the other three projectors are given by @xmath160 , @xmath161 and @xmath162 , and the result of these projections would be represented by the output state @xmath163 if all measurements had been made before the environment starts its deleterious effect .
suppose , on the other hand , that these measurements are performed at later instants of time .
let us assume , for instance , that the projections are performed around the first valley of the state fidelity ( see fig .
[ statefidelity ] ) , within intervals @xmath164 starting at @xmath165 , that is , @xmath166 , @xmath167 and @xmath168 . obviously we will not obtain a good fidelity for this computation since the state fidelity is very small within this time interval and , in fact , the not gate fidelity is @xmath169 showing that the probability of the output state be the desired state is approximately @xmath170 . now ,
if we consider that @xmath117 , @xmath118 and @xmath119 are performed at @xmath171 , @xmath172 and @xmath173 , i.e. , are performed around the first peak of the state fidelity , the not gate fidelity is @xmath174 .
but we can obtain better results than these simply choosing another set of instants of time .
if , for example , we set the controls to perform our measurements at slightly different times , choosing to apply the measurements in a smaller neighborhood around the first peak , we can get better results such as @xmath175 or @xmath176 for the set @xmath177 respectively at @xmath178 or @xmath179 .
therefore , if we perform the measurements around the first valley we obtain a gate fidelity of @xmath169 whereas if we perform them at a later time , waiting to reach the surroundings of the fist peak , we obtain a much better gate fidelity .
another possibility can be imagined if we consider that the set of measurements is performed separately at each of the first three consecutive minima of the state fidelity , that is , we apply the first projector at the first valley , the second projector at the second valley and the third projector at the third valley of the state fidelity ; in this case @xmath180 and the not gate fidelity assumes the value of @xmath181 . on the other hand , if we do exactly the contrary , choosing the instants of time of the first three consecutive peaks , we obtain a not gate fidelity of @xmath182 at @xmath183 . considering another example of one qubit gate acting on the same input state @xmath159
, we can analyze the effect of the cluster state s oscillatory behavior on the mbqc fidelity in another situation of interest .
as is well known , the hadamard gate transforms the state @xmath184 into the state @xmath185 , so , in an idealized situation , we would expect the output state @xmath186 to assume the desired outcome @xmath187 . on the other hand ,
since our cluster state is interacting with the dephasing channel , the outcome of the mbqc can be very different from the expected one , mainly if we choose the wrong instants of time to apply the projective measurements , as we will see .
for this particular example , we have @xmath188 , @xmath189 , and @xmath162 , where @xmath190 . admitting that these measurements are performed around the first valley of the state fidelity within intervals @xmath164
starting at @xmath166 , as before , the hadamard gate fidelity is @xmath191 , while if we consider that @xmath171 , @xmath172 and @xmath173 , and the measurements are performed around the first peak , we have a hadamard gate fidelity of @xmath192 .
nevertheless , a much better result can be obtained if @xmath193 ; in which case , the probability of the output state be the desired one is @xmath194 .
now , consider again that the set of measurements is performed at the first three consecutive minima of the state fidelity ; again @xmath180 and the gate fidelity assumes the value @xmath181 . on the other hand , choosing the times of the first three consecutive peaks we get a hadamard gate fidelity of @xmath130 at @xmath183 .
finally , we examine another one qubit gate example that is often found in the literature .
the phase gate under the mbqc can be accomplished with @xmath195 , @xmath196 , and @xmath162 , and the input state @xmath197 is rotated , teleporting and printing the outcome in qubit number 5 whose output state @xmath186 acquires a relative phase @xmath198 assuming the idealized value @xmath199 .
once again , taking into account the interaction of our quantum computer " with the dephasing environment , and applying the projectors at @xmath166 , @xmath167 and @xmath168 , we obtain a gate fidelity of @xmath200 while if we wait to apply the projectors at @xmath171 , @xmath172 and @xmath173 we get a gate fidelity of @xmath201 .
however , applying the measurements in a smaller neighborhood of the first peak we can get a gate fidelity of @xmath202 at @xmath203 , @xmath204 and @xmath205 . considering that the set of measurements is performed at the first three consecutive minima of the state fidelity , at @xmath206 , @xmath207 and @xmath208 , the value of the gate fidelity is @xmath209 , while if the set of measurements is applied on the first three consecutive maxima we get @xmath194 for the gate fidelity . therefore , depending on the times we choose to perform our set of projective measurements we will obtain better or worse computational fidelity results .
now , let us consider the other scenario where the subsequent measurements are performed in sequence but practically at the same time .
again we consider the same three examples of gate fidelities .
we start with the not gate fidelity dynamics and in fig .
[ fig1 ] we show this gate fidelity for @xmath127 and @xmath210 as a function of time in the quantum regime . as in our previous examples
, we observe that there are times that maximize the value of the gate fidelity and times that minimize it .
it is clear that if all measurements are performed at @xmath211 the computation fidelity is @xmath212 , but if there is a gap between the first and the three subsequent measurements , then there is an optimum value of the time gap . in the example illustrated in fig .
[ fig1 ] , if @xmath157 is greater than @xmath213 ( where the gate fidelity is @xmath214 ) the best gate fidelity is obtained for @xmath215 , when it reaches @xmath214 again .
if we apply the same gate operation at later times like @xmath216 or @xmath217 , we still get a gate fidelity better than @xmath218 . however ,
if we apply this operation at @xmath219 we obtain a gate fidelity of @xmath220 , showing that fast measurements is not a warranty of good mbqc results .
when the three subsequent measurements are performed almost simultaneously .
we consider @xmath139 , @xmath140 , and @xmath141 . ]
considering a hadamard gate fidelity dynamics , we can see from fig .
[ fig2 ] that a input state @xmath159 is rotated to the output state @xmath221 with probability greater than @xmath218 at times like @xmath215 , @xmath216 or @xmath217 while it is rotated to the same output state with probability of less than @xmath222 at times like @xmath223 , @xmath224 or @xmath225 . when the three subsequent measurements are performed almost simultaneously .
we choose the state @xmath226 to be rotated to @xmath227 .
we consider @xmath139 , @xmath140 , and @xmath141 . ]
[ h ] for the quantum regime .
we choose the state @xmath228 to be rotated to @xmath229 .
we consider @xmath139 , @xmath140 , and @xmath141 .
, title="fig : " ] in fig .
[ fig3 ] we can see the phase gate fidelity dynamics and observe once again that the oscillatory behavior presented by the cluster state interacting with the specific kind of quantum channel considered in this paper produces instants of time that optimize the value of the computation compared with times that do exact the opposite . again we see that at times like @xmath230 , @xmath231 or @xmath232 we have a gate fidelity of @xmath233 , @xmath202 and @xmath214 , respectively , while at times such as @xmath234 , @xmath235 or @xmath236 we have a gate fidelity of @xmath237 , @xmath238 and @xmath239 .
it is important to emphasize that ultra fast measurements , which have to be performed in the very short bath correlation time scale , can be produced with current technology @xcite .
these are the basis of the dynamical decoupling techniques that are applied to beat the decoherence process @xcite .
furthermore , even in this very short time scale , the time that each measurement is applied can be very precise , as we can see , for example , in the experimental realization of the uhrig dynamical decoupling , the carr - purcell - meiboom - gill - style multi - pulse spin echo @xcite , and others .
this implies that the scenario studied in the manuscript is very realistic and that any mbqc realized with ultra fast measurements needs to account for the oscillatory behavior of the dynamics .
we study the exact dynamics of an @xmath0-qubit system interacting with a common dephasing environment and we introduce a necessary condition for the system fidelity to present a non - monotonical behavior .
our approach reveals that this characteristic does not depend on the initial quantum entanglement and , in fact , is a property connected with the geometry of the state .
actually , for any initial state given by a superposition of eigenstates of the total pauli @xmath7 operator , the fidelity exhibits a non - monotonical character if at least one of the eigenvalues of the components differs from the others .
we show that this behavior of the fidelity brings crucial implications to the mbqc , that is , we show that , under the action of a common dephasing environment , this non - monotonical time dependence can provide us with appropriate time intervals for the preservation of better computational fidelities .
we have illustrated our findings by examining the fidelity of a not , a hadamard and a phase quantum gates realized via mbqc .
r. raussendorf and h. j. briegel , phys .
lett . * 86 * , 5188 ( 2001 ) . h. j. briegel , d. e. browne , w. dr , r. raussendorf and m. van den nest , nature phys . * 05 * , 19 - 26 ( 2009 ) .
h. j. briegel and r. raussendorf , phys .
lett . * 86 * , 910 ( 2001 ) .
d . gross and j. eisert , phys .
lett . * 98 * , 220503 ( 2007 ) ; d. gross , j. eisert , n. schuch , and d. perez - garcia , phys .
a * 76 * , 052315 ( 2007 ) ; d .
gross and j. eisert , phys .
a * 82 * , 040303 ( 2010 ) .
cai , w. dr , m. van den nest , a. miyake , and h. j. briegel , phys .
lett . * 103 * , 050503 ( 2009 ) .
t. morimae , phys .
a * 81 * , 060307(r ) ( 2010 ) ; d. bacon , s. t. flammia , phys .
a * 82 * , 030303(r ) ( 2010 ) ; y. s. weinstein , phys . rev .
a * 79 * , 052325 ( 2009 ) ; m. s. tame , m. paternostro , m. s. kim , and v. vedral , phys .
a * 72 * , 012319 ( 2005 ) . c. m. dawson , h. l. haselgrove , and m. a. nielsen , phys .
a * 73 * , 052306 ( 2006 ) ; o. guhne , f. bodoky , and m. blaauboer , phys . rev .
a * 78 * , 060301 ( 2008 ) ; d. cavalcanti , r. chaves , l. aolita , l. davidovich , a. acin , phys .
103 * , 030502 ( 2009 ) ; g. gilbert and y. s. weinstein , quantum information and computation viii .
edited by e. j. donkor , a. r. pirich , and h. e. brandt , proceedings of the spie , * 7702 * , pp .
77020j-77020j-9 ( 2010 ) ; y. s. weinstein and g. gilbert , j. of mod . opt .
* 19 * , 1961 ( 2010 ) ; l. aolita , d. cavalcanti , r. chaves , et al . ,
phys . rev .
a * 82 * , 032317 ( 2010 ) ; r. chaves and f. de melo , phys .
a * 84 * , 022324 ( 2011 ) .
m. a. nielsen , phys .
* 93 * , 040503 ( 2004 ) ; n. c. menicucci , p. van loock , m. gu , _
et al_. , phys .
* 97 * , 110501 ( 2006 ) ; j .-
cai , a. miyake , w. dr , h. j. briegel , phys .
a * 82 * , 052309 ( 2010 ) .
f. meier , j. levy , and d. loss , phys .
lett . * 90 * , 047901 ( 2003 ) ; p. walther , et .
al , nature * 434 * , 169 ( 2005 ) ; n. kiesel , _ et al_. , phys .
* 95 * , 210502 ( 2005 ) ; y. tokunaga , t. yamamoto , m. koashi , and n. imoto , phys .
a * 74 * , 020301 ( 2006 ) ; r. prevedel , m. s. tame , a. stefanov , _
et al_. , phys .
lett . * 99 * , 250503 ( 2007 ) ; c. y. lu , et .
al , nat . phys . * 3 * , 91 ( 2007 ) ; r. ceccarelli , g. vallone , f. demartini , p. mataloni , a. cabello , phys .
* 103 * , 160401 ( 2009 ) ; r. kaltenbaek ,
_ et al_. , nat . phys . * 6 * , 850 ( 2010 ) .
b . gao , _
et al_. nature photonics * 5 * , 117 ( 2011 )
. j. h. reina , l. quiroga and n. f. johnson , phys .
a * 65 * , 032326 ( 2002 ) .
h. -p . breuer and f. petruccione ,
_ the theory of open quantum systems _ ( oxford university press , oxford , new york , 2002 ) .
b. bellomo , r. lofranco , and g. compagno , phys .
lett . * 99 * , 160502 ( 2007 ) . m. m. wolf , j. eisert , t. s. cubitt , and j. i. cirac , phys . rev
. lett . * 101 * , 150402 ( 2008 ) .
breuer , e. -m .
laine , and j. piilo , phys .
lett . * 103 * , 210401 ( 2009 ) .
laine , j. piilo , and h. -p .
breuer , phys .
a * 81 * , 062115 ( 2010 ) .
b. vacchini and h. -p .
breuer , phys .
a * 81 * , 042103 ( 2010 ) .
a. k. rajagopal , a. r. usha devi , and r. w. rendell , phys .
a * 82 * , 042107 ( 2010 ) .
z. he , j. zou , b. shao , and s. kong , j. phys .
b : at . mol .
phys * 43 * , 115503 ( 2010 ) .
x. xiao , m. fang , and y. li , j. phys .
b : at . mol . opt .
phys * 43 * , 185505 ( 2010 ) .
a. rivas , s. f. huelga , and m. b. plenio , phys .
* 105 * , 050403 ( 2010 ) .
f. f. fanchini , t. werlang , c. a. brasil , l. g. e. arruda , and a. o. caldeira , phys .
a * 81 * , 052107 ( 2010 ) .
m. a. nielsen and i. l. chuang , _ quantum computation and quantum information _
( cambridge university press , cambridge , england , 2000 ) .
a. j. legget , s. chakravarty , a. t. dorsey , m. p. a. fisher , a. garg and w. zwerger , rev .
* 59 * , 1 ( 1987 ) .
s. m. clark , et al .
lett . * 102 * , 247601 ( 2009 ) ; j. f. du , nature * 461 * , 1265 ( 2009 ) ; g. de lange , et al .
science * 330 * , 60 ( 2010 ) .
m. j. biercuk , et al .
a * 79 * , 062324 ( 2009 ) .
l. viola and s. lloyd , phys .
a * 58 * , 2733 ( 1998 ) ; l. viola , e. knill , and s. lloyd , phys .
* 82 * , 2417 ( 1999 ) ; l. viola , s. lloyd , and e. knill , phys .
lett . * 83 * , 4888 ( 1999 ) . | measurement - based quantum computation is an efficient model to perform universal computation .
nevertheless , theoretical questions have been raised , mainly with respect to realistic noise conditions . in order to shed some light on this issue
, we evaluate the exact dynamics of some single qubit gate fidelities using the measurement - based quantum computation scheme when the qubits which are used as resource interact with a common dephasing environment .
we report a necessary condition for the fidelity dynamics of a general pure @xmath0-qubit state , interacting with this type of error channel , to present an oscillatory behavior and we show that for the initial canonical cluster state the fidelity oscillates as a function of time .
this state fidelity oscillatory behavior brings significant variations to the values of the computational results of a generic gate acting on that state depending on the instants we choose to apply our set of projective measurements .
as we shall see , considering some specific gates that are frequently found in the literature , neither fast application of the set of projective measurements necessarily implies high gate fidelity , nor slow application thereof necessarily implies low gate fidelity .
our condition for the occurrence of the fidelity oscillatory behavior shows that the oscillation presented by the cluster state is due exclusively to its initial geometry .
other states that can be used as resources for measurement - based quantum computation can present the same initial geometrical condition
. therefore , it is very important for the present scheme to know when the fidelity of a particular resource state will oscillate in time and , if this is the case , what are the best times to perform the measurements . |
the study of self - assembled quantum dots , as realized by the stranski - krastanow growth mode , has been a fascinating research area during the last decade .
the possibility to discover new physics in these zero - dimensional ( 0d ) structures together with possible applications in opto - electronics has led to many experimental and theoretical results .
most studies were devoted to type i quantum dots@xcite , where both electron and hole are spatially located inside the quantum dots .
relatively few works however were done for type ii quantum dots , where electrons and holes are spatially separated .
one expects from this type of dots interesting properties , such as large tunability of the emission energy and radiative lifetimes which are considerably longer than for their type i counterparts@xcite . within the group of type
ii quantum dots , one can still distinguish between two systems .
namely , the system which confines the electrons inside the quantum dot , and forms an antidot for the hole , like e.g. the @xmath9 dots ( if strain is neglected ) . besides this , there are the systems , such as @xmath10 @xcite and @xmath11_si _ @xcite , where the holes are confined within the dot , but where the electrons are located outside .
experimental studies on this type of quantum dots have been performed in refs .
@xcite for the @xmath9 system and in refs .
@xcite for the @xmath10 system .
theoretical investigations on @xmath9 type ii dots were performed by pryor _
et al_. @xcite and tadic _ et al . _
@xcite , who used a strain - dependent @xmath12 hamiltonian to calculate the electronic structure .
et al_. @xcite calculated the landau levels in a high magnetic field , solving the hartree equations self - consistently , while the present authors studied an exciton in a planar quantum dot using a hartree - fock approximation @xcite .
et al_. @xcite calculated the binding energy of excitons , charged excitons and biexcitons in @xmath10 dots using the hartree - fock approximation .
the magneto - exciton in a @xmath13 dot was investigated by kalameitsev _
et al_. @xcite . in the present paper
we study a single electron - hole pair bound by the coulomb interaction , i.e. an exciton , in a _ model
_ type ii quantum dot .
the quantum dot is modelled by a quantum disk of finite height , which is an extension of our previous work @xcite where we did not take into account the @xmath14 -direction . as an example
, we consider the electron in the dot and the hole in the barrier material .
the conclusions of our paper are also valid for the reverse situation where the hole is inside the dot and the electron is outside the dot . in the present work , we neglect strain effects and for the confinement potential we use a hard wall of finite height , confining the electron inside the dot and repelling the hole to the barrier region .
furthermore , a magnetic field was applied in the growth direction , i.e. @xmath15 . the first part of this paper deals with the study of the exciton in a single dot .
investigation of the influence of the disk parameters , namely the radius @xmath0 and thickness @xmath16 showed that one can distinguish between two regimes : the disk - like regime and the pillar - like regime .
for the first one , the disk thickness @xmath1 is much smaller than its diameter , i.e. @xmath17 whereas the latter one describes the system with @xmath3 .
although the magnetic field couples only with the in - plane electron and hole coordinates , we will show that these systems , with different @xmath14-extension , behave very differently under an applied magnetic field . in the second part of the paper , we study the properties of vertically coupled quantum dots .
the interest of this study lies in the following .
it is generally known that in reality self - assembled quantum dots resemble more our disk - like system than our pillar - like system .
however , it is also known that it is not too difficult to form vertical stacks of these disk - like dots and thereby create a system which could behave like a pillar - like system .
the paper is organized as follows . in sec .
ii , we give a brief discussion of our theoretical model .
the numerical results in the absence of a magnetic field are presented in sec .
section iv deals with the result for a perpendicular applied magnetic field .
part a of this section discusses the results for the single dot disk - like system , whereas part b is dedicated to the pillar - like system .
parts c and d deal with two vertically coupled dots , respectively disks and truncated cones .
parts e and f deal with three vertically coupled dots , respectively for small and large interdot distances .
finally , we make a small revision of the results for the single disk in part g. our results are summarized in sec .
v. in the appendix , we discuss in more detail the numerical approach we used for the calculation of the hartree integral .
we extended our previous approach @xcite , which was valid for planar dots , and include the @xmath14-direction .
the energies and wavefunctions are solved within the effective mass approximation ( with @xmath18 and @xmath19 the effective electron and hole masses , respectively , @xmath20 @xmath21 and @xmath22 ) and the hartree - fock ( hf ) single particle equations can be written as @xmath23 \psi _ { e}(r_{e},z_{e})=\epsilon _ { e}\psi _ { e}(r_{e},z_{e } ) , & & \eqnum{1a } \\ \left [ -\frac{\hbar ^{2}}{2m_{h}}\frac{1}{r_{h}}\frac{\partial } { \partial r_{h}}\left ( r_{h}\frac{\partial } { \partial r_{h}}\right ) -\frac{\hbar ^{2}}{% 2m_{h}}\frac{\partial ^{2}}{\partial ^{2}z_{h}}+\frac{\hbar ^{2}}{2m_{h}}% \frac{l_{h}^{2}}{r_{h}^{2}}-\frac{l_{h}}{2}\hbar \omega _ { c , h}+\frac{1}{8}% m_{h}\omega _ { c , h}^{2}r_{h}^{2}\right . & & \nonumber \\ \left .
+ v_{h}(r_{h},z_{h})-\frac{e^{2}}{4\pi \epsilon } \int \frac{\rho _ { e}(r^{\prime } , z^{\prime } ) } { |{\bf r}-{\bf r}^{\prime } |}d{\bf r}^{\prime } % \right ] \psi _ { h}(r_{h},z_{h})=\epsilon _ { h}\psi _ { h}(r_{h},z_{h } ) , & & \eqnum{1b}\end{aligned}\ ] ] where we made use of the axial symmetry by taking @xmath24 and @xmath25 and where the densities @xmath26 and @xmath27 are given by respectively @xmath28 and @xmath29 .
we made use of a finite difference scheme to solve the hartree - fock equations .
more details about the implementation of this finite difference scheme can be found in refs .
note that we consider only a single electron and single hole and therefore there are no exchange terms .
the reason that these equations can still be called hf lies in the fact that the self - interaction is excluded . as confinement potentials we take hard walls of finite height : @xmath30 with @xmath0 the radius of the disk , and where we took @xmath31 positive and @xmath32 negative . note
that the only good quantum number is the total angular momentum in the @xmath14-direction , defined by @xmath33 we solved the equations self - consistently using an iterative procedure .
since only the electron is confined in the absence of any coulomb interaction , we start with the free electron solution . the hartree integrals
are then integrated numerically as @xmath34 where @xmath35 is the complete elliptic integral of the first kind .
more details about the calculation and numerical implementation of this integral is given in the appendix .
after convergence of the iteration procedure , the total energy is given by @xmath36 the contribution of the correlation to the total energy is neglected in hf , but for the self - assembled quantum dots , it is expected to be less than 2% @xcite and for type ii dots this will be even less .
in our model system , there are two main parameters which can be varied : the radius of the disk @xmath37 and the thickness of the disk @xmath38 in the first part of our numerical study we will investigate the influence of these parameters on the exciton energy and wavefunction in the absence of a magnetic field . following material parameters ,
typical for the @xmath39 dot system , were taken : @xmath40 @xmath41 @xmath42 @xmath43 and @xmath44 the variation of the disk radius @xmath0 and thickness @xmath1 has large consequences for the hole wavefunction .
namely , the hole always tends to sit as closely as possible to the electron . in this way ,
when the disk thickness is larger than its diameter @xmath45 , the hole will prefer to sit at the radial boundary of the disk .
when we consider very thin disks however @xmath46 , it will be much more favourable for the hole to be located above and below the disk .
this state was not possible in our previous study @xcite of the two - dimensional flat quantum dot , where the hole was forced to sit at the radial boundary of the dot . to illustrate this , we calculated the probability to find the hole at the radial boundary of the disk as @xmath47 when calculating this value for varying @xmath0 and @xmath16 we obtained a phase diagram for the position of the hole wavefunction , which is shown in fig .
the hole confinement potential in all these calculations was fixed to @xmath44 the solid curve indicates where 50% of the hole is located at the radial boundary , for the dashed curves it is respectively 25% and 75% . in order to have a more visual picture of the hole state , we present contourplots of the hole wavefunction @xmath48 for three characteristic situations in the phase diagram , as indicated by the numbered stars .
1(b ) shows the hole density for @xmath49 @xmath50 where the dotted lines indicate the boundary of the quantum disk .
we clearly see that the hole is mainly situated at the radial side of the disk .
1(c ) was made for @xmath51 @xmath52 and at this position , exactly 50% of the wavefunction is located at @xmath53 .
the third plot , fig .
1(d ) , depicts the result for a thin disk , with @xmath54 and @xmath55 where the hole wavefunction is mostly located above and below the disk , where it is nearer to the electron .
thus we can distinguish between two main systems : a ) a disk - like system with @xmath17 where the hole will be located above or below the quantum disk , and b ) a pillar - like system with @xmath56 where the hole sits mainly at the radial boundary of the disk .
the solid line in fig .
1(a ) can be approximated by the curve @xmath57 we will see in the following section that a magnetic field has a very different effect on these two classes of systems .
after having discussed the influence of the main parameters of our system , we are now ready to consider an extra feature , i.e. a magnetic field applied along the @xmath14-direction , parallel to the growth direction
. the magnetic field will squeeze the wavefunctions in the radial direction .
when @xmath17 the hole is above and below the disk , and will be pressed stronger to the center of the disk .
2 depicts the exciton energy as a function of the magnetic field for a disk with @xmath58 and @xmath59 for two different values of the hole angular momentum @xmath60 note that this exciton energy includes both the one particle energies and the coulomb energy , i.e. @xmath61 in the present case the single hole energy @xmath62 will be 0 , as the single hole is not confined .
the single electron energies at @xmath63 and @xmath64 are given by respectively @xmath65 and @xmath66 we find a strong enhancement of the exciton energy of almost @xmath67 for @xmath68 due to the magnetic field , the electron and hole are both pushed closer to the center , which enhances the coulomb interaction .
however , also the one particle electron energy increases drastically with increasing magnetic field , thus cancelling the effect of the enhanced ( negative ) coulomb energy .
the two insets show what happens with the electron ( dashed curves ) and hole ( full curves ) wavefunctions when a magnetic field is applied . for @xmath63 ( fig .
2(a ) ) the electron is strongly confined in the quantum disk , and the hole is localized above and below the disk . from the figure , it is obvious that this position is energetically most favourable for the hole , since the hole will be very close to the electron . because of the very small thickness
, there is some smearing of the electron wavefunction out of the disk in the @xmath14-direction , which attracts the hole to this position .
the corresponding wavefunctions for @xmath64 are plotted in fig .
2(b ) , where again the dashed curve indicates the electron wavefunction , and the solid curve the hole .
we see that both particles are compressed in the radial direction . for the electron ,
this leads to a further penetration of the wavefunction into the barrier .
because of this , the hole will be even stronger attracted to the top and bottom of the disk , which leads to a larger penetration of the hole in the disk . as a result
, the hole wavefunction will also be slightly squeezed in the @xmath14-direction . the exciton energy as a function of the magnetic field is plotted in fig .
3 for the case of @xmath49 @xmath69 and @xmath44 an interesting feature which appears is the occurrence of ( hole ) angular momentum transitions with increasing magnetic field .
the origin of the angular momentum transitions is the following .
as discussed in sec .
iii , the preferred position of the hole for a disk with a thickness larger than its diameter will be at the radial boundary of the disk .
when a magnetic field is applied along the @xmath14-direction , the hole will be pushed against the border of the disk . since the disk forms a barrier for the hole
, it will be energetically more favourable to jump to a higher angular momentum @xmath6 state .
notice that @xmath70 is an approximate quantum number .
this feature was also apparent in our previous study of planar type ii dots @xcite , where we forced the hole to sit at the radial boundary .
this is similar to an exciton in a quantum - ring structure @xcite . in the present study however , we allow for the hole to relax also in the @xmath14-direction
. therefore , for a small thickness of the disk ( disk - like system ) , it will be much more favourable for the hole to move in the @xmath14-direction and only for thick enough disks , such @xmath6 transitions occur .
contourplots of the wavefunctions are shown , respectively for the electron at @xmath63 ( inset of fig . 3 ) and for the hole at @xmath71 @xmath72 @xmath64 and @xmath73 ( figs .
4(a)-(d ) ) .
we find that the electron is located in the center of the disk , with almost no penetration into the barrier . at @xmath71
the hole is mainly sitting at the radial boundary of the disk , although there is some extent of the wavefunction towards the top and bottom of the disk , and there is also some penetration into the dot .
when increasing the magnetic field , the hole jumps to the @xmath74 state around @xmath75 we plotted the hole wavefunction at @xmath76 ( fig .
4(b ) ) , and we see that the hole is pushed at the side , where it has now more space to extend in the radial direction .
further increasing the magnetic field , leads to a jump to @xmath77 at @xmath78 and to @xmath79 at @xmath80 the contourplots were made for respectively @xmath64 ( fig .
4(c ) ) and @xmath73 ( fig .
4(d ) ) , and we see again that the hole is sitting at the side of the disk .
notice that , for all four plots , the probability to find the hole at @xmath81 is almost the same , i.e. @xmath82 in the absence of those @xmath6 transitions , the magnetic field would have compressed the hole much more strongly to the radial boundary of the disk which leads to a larger energy as is shown clearly by the solid curve in fig . 3 . in order to investigate the influence of the hole confinement potential @xmath32 on the angular momentum transitions , we made a phase diagram of the angular momentum state of the hole as a function of both the confinement potential and the magnetic field .
this is shown in fig . 5 for a disk with radius @xmath83 and thickness
@xmath84 the figure shows that up to @xmath85 no angular momentum transitions occur . in this region @xmath32
is too small to form a barrier for the hole , and the hole jumps inside the disk due to the coulomb interaction , forming a type i system .
we also see that the transition between type i and type ii is very sharp . in our previous study of the planar dots @xcite
, we found from a similar figure a re - entrant behaviour to the @xmath7 state for sufficient large magnetic fields .
this is not so evident in the present case , as we find that the line which separates the type i from the type ii behaviour is almost vertical . from an experimental point of view , it is interesting to look at the probability for recombination of the exciton , which is proportional to the square of the overlap integral @xmath86 the integral over the angle gives @xmath87 this means that the probability for de - excitation is only non - zero for @xmath88 this implies that , after an angular momentum transition , the probability for recombination decreases strongly , leading to a vanishing of the photoluminescence ( pl ) spectrum after a certain value of the magnetic field .
6 depicts the overlap integral @xmath89 as a function of the confinement potential of the hole , for both @xmath63 ( solid curve ) and @xmath76 ( dashed curve ) .
for @xmath63 the ground state will always be the @xmath7 state , and the overlap integral will always be non - zero .
however , as the system goes from type i to type ii we see a strong decrease of the overlap , because the exciton becomes spatially indirect . for @xmath76
we find a different picture : after the system has become type ii , the overlap integral becomes zero .
this is due to the fact that at @xmath76 an angular momentum transition has occurred and the condition @xmath90 is no longer fulfilled . the inclusion of the @xmath14-direction in our calculation enables us to investigate systems of vertically coupled quantum dots .
we have now an extra parameter to vary , namely the interdot distance @xmath5 .
in this first part , we focus on two stacked dots of the same size , namely radius @xmath83 and thickness @xmath91 , and with @xmath92 .
the total stack height of @xmath93 is thus larger than the diameter of the disks , being @xmath94 .
therefore we expect to find a magnetic field behaviour which resembles that of the single disk pillar - like system .
the result for the exciton energy as function of the magnetic field is shown in fig .
indeed we find angular momentum transitions with increasing magnetic field .
the inset , fig .
7(a ) , shows a contourplot of the electron wavefunction at @xmath63 , which is symmetrically distributed over the two disks .
the coulomb interaction attracts the hole as close as possible to the electron , in this case the radial boundary of the disks .
as one can see in fig .
7(b ) , the hole tends also to sit between the disks , to be even closer to the electron . a magnetic field acts on the wavefunctions in the radial direction , and thereby pushes the hole stronger between the disks .
there is however not enough space , and therefore it will be energetically more favourable for the hole to jump to a higher @xmath6-state . when giving a closer look at the behaviour of the different @xmath6-states in a magnetic field ( fig . 7 )
, a remarkable feature appears : the @xmath7 curve exhibits a kind of kink around @xmath95 , and we find a similar feature for @xmath74 around @xmath96 .
investigation of the wavefunctions ( figs .
8(a - c ) ) learns us that we are dealing with a _
_ spontaneous symmetry breaking__@xcite . the corresponding hole wavefunction at @xmath97 and @xmath98 , for @xmath7 is shown in fig . 8(a , b , c ) respectively , where one can see clearly the increase of asymmetry with magnetic field .
but for such magnetic fields the @xmath7 state is not the ground state .
this means that the jump to a higher angular momentum state is still preferred above the asymmetric state ( see fig .
8(d ) for @xmath79 and @xmath64 ) .
we also found that with increasing @xmath6 the symmetry breaking will occur at larger magnetic fields . in order to investigate whether or not the ground state configuration is asymmetric , we calculated @xmath99 which expresses the degree of asymmetry of the hole wavefunction in the @xmath14-direction .
this quantity is plotted in fig .
9 as function of the magnetic field for @xmath100 .
this plot confirms that after @xmath101 the @xmath7 state becomes highly asymmetric , and for higher @xmath8 also @xmath74 becomes asymmetric .
the asymmetric states for the subsequent @xmath6 states occur for @xmath102 .
the contourplot of the hole wavefunction at @xmath64 is shown in fig .
8(d ) for the ground state , which is @xmath79 . the inset of fig .
9 shows the dipole moment for the different @xmath6 states with increasing magnetic field .
this is an interesting quantity , as it can be measured experimentally .
note that for our symmetric system , we have the unique feature that a magnetic field is able to induce a dipole moment !
we can understand the spontaneous symmetry breaking as follows .
in the asymmetric case the electron wavefunction is of course attracted to the hole ( see fig .
8(b ) , dashed lines ) and thus stronger confined , in this way increasing its single particle energy . therefore , the broken symmetry state can only occur if it can overcome this increase in energy due to a larger coulomb attraction between electron and hole .
we mentioned that an increasing magnetic field favoured the broken symmetry state .
the reason is that with increasing magnetic field the electron becomes stronger confined in the dot , and consequently expells more in the @xmath14-direction . in the broken symmetry case it even expells more , resulting in a larger overlap with the hole , which can eventually make the asymmetric state lower in energy than the symmetric state . if this is true , than the broken symmetry would occur even sooner with increasing magnetic field for thinner disks , where the electron is expelled more in the barrier material in the @xmath14-direction . that this is indeed the case
is shown in fig .
10 where for two thin vertically coupled disks ( @xmath103 nm , @xmath104 nm and @xmath1053 nm , @xmath106 ) we find already a broken symmetry state for zero magnetic field . in this case
, this broken symmetry state is also the ground state .
it is now interesting to check whether the above angular momentum transitions survive in a system which is completely asymmetric from the beginning , such as a system of two vertically coupled truncated cones .
the two cones are equal , and defined by a base radius @xmath107 of @xmath108 , a thickness @xmath91 , and an angle of @xmath109 ( dashed lines in the insets of fig .
the hole and electron wavefunctions ( figs . 11 ( a ) and ( b ) ) show indeed the asymmetric behaviour , at @xmath63 and for @xmath7 .
the result for the exciton energy as a function of the magnetic field is depicted in fig .
the interesting feature is that , even for this system , angular momentum transitions occur .
since there is still some part of the hole wavefunction located at the radial border and also some part between the disks , it appears to be still more favourable for the hole to jump to the higher @xmath6-state than to move more and more below the disk . figs .
12(a - c ) show contourplots of the hole wavefunction in the ground state for subsequent @xmath6-states , respectively @xmath74 at @xmath110 , @xmath77 at @xmath111 , and @xmath79 at @xmath64 . for simplicity
we consider only the case of three identical dots .
the first system under investigation contains three vertically coupled quantum disks , each with disk radius @xmath51 disk thickness @xmath112 and with interdot distance @xmath113 in this case the confinement of the hole @xmath32 was taken to be @xmath114 fig .
13 shows the result for the exciton energy as a function of the magnetic field .
similar to the case of the pillar - like single dot , we find angular momentum transitions with increasing magnetic field .
the origin of these angular momentum transitions can be understood by looking at the wavefunctions .
the insets of fig .
13 show the electron wavefunctions at ( a ) @xmath63 and ( b ) @xmath64 , and it appears that the main part is located in the middle dot , whereas there is some small extent of the wavefunction into both the upper and lower dots . at @xmath115
the wavefunction is more squeezed in the radial direction .
the evolution of the hole wavefunction with increasing magnetic field is depicted in fig .
14 . at @xmath63 ( fig .
14(a ) ) , the main part of the hole wavefunction is situated at the radial boundary of the stacked system and its probability to sit at the radial side is @xmath116 there is some extent of the wavefunction towards the top and bottom of the stack , and also between the dots .
when the magnetic field increases ( fig .
14(b ) , @xmath117 ) , the wavefunction is pushed further between the disks , because it prefers to sit as closely as possible to the electron .
however , there is not enough space between the disks to confine the whole wavefunction , and therefore it is energetically more favourable for the hole to jump to a higher angular momentum state . further increasing the magnetic field leads to more angular momentum transitions ( figs .
14(c - f ) ) .
as can be seen from the wavefunctions , the ground state for this system is always symmetric .
the asymmetry , as found in the two - disks system , is however still present , but it turns out not to be the ground state . for the @xmath7 state ,
the wavefunctions become asymmetric at about @xmath110,resulting in a slight bow in the full curve of fig . 13 .
the fact that the asymmetry is less pronounced in this system can be understood as follows : in the three - disks system the main part of the electron is located in the middle disk .
this is in contrast to the two - disks system , where the electron wavefunction has equal parts sitting in the two disks .
even when an asymmetry is induced in the present three - disks system , the electron will still be mainly located in the middle disk , and therefore the hole will sit more tightly at the radial boundary than was the case for the two coupled disks .
when we increase the interdot distance @xmath5 up to @xmath118 , the hole will start to be confined between the disks .
15 shows that there are no angular momentum transitions for this system and it resembles very much the narrow disk - like system ( see fig . 2 ) . in this figure
the solid curve denotes the @xmath7 state , whereas the dashed curve denotes the @xmath74 curve .
we turn again to the wavefunctions to explain this behaviour .
the electron is sitting mainly in the middle dot , this time with smaller extent into the upper and lower dots ( figs .
15(a ) and ( b ) for respectively @xmath63 and @xmath64 ) .
the contourplot for the hole wavefunction ( fig .
16(a ) ) shows that already at @xmath119 the hole is completely situated between the disks .
this is the preferred place for the hole , as it tends to sit as close as possible to the electron . further increasing the magnetic field ,
squeezes the wavefunction more in the radial direction ( fig .
as there is still enough space between the disks , there is no need for the hole to jump to a higher angular momentum state , and @xmath7 remains the ground state over the whole @xmath8-region . for this case ,
no asymmetry occurs up to @xmath64 .
this means that this system is quite stable against spontaneous symmetry breaking .
this stability is due to the fact that again the electron is strongly located in the middle disk , and that the hole is now sitting between the disks instead at the radial boundary .
we now turn back to the single disk systems , both the disk - like and the pillar - like , where we did not talk about broken symmetry solutions . however also in this case they can be found .
we find that the disk - like system , i.e. @xmath58 and @xmath59 , is perfectly stable up to @xmath64 and we find no asymmetric behaviour .
this is in perfect agreement with the result for the three coupled disks with @xmath120 . indeed , for the disk - like system we have again the electron sitting in the disk with the hole above and below the disks , which is quite similar to the second three - disks system . for the pillar - like system , i.e. @xmath121 and @xmath69
, we do find an asymmetry , but only for the @xmath7 state and at very high magnetic fields , i.e. starting around @xmath122 .
the difference in energy between the states with symmetric and antisymmetric wavefunctions is only @xmath123 for @xmath73 .
the ground state however remains symmetric over the total considered @xmath8-region .
thus the wavefunctions , as plotted in fig . 4 , remain unchanged .
we studied an exciton in a type ii quantum disk of radius @xmath0 and thickness @xmath38 for the confinement potential we took a hard wall of finite height , with the hole located in the barrier , which is only confined by the interaction with the electron , confined inside the dot .
we calculated the exciton energy and wavefunction using the hartree - fock mesh method , which allows us to start without any knowledge of the single hole wavefunction .
we limit ourselves to the study of a _ model _ system in which strain effects are neglected . in the first part
, we examined the case of an exciton in a single disk .
first we investigated the exciton properties in the absence of a magnetic field .
the calculation of the probability for the hole wavefunction to sit at the radial boundary tells us that we can distinguish two regimes : disk - like systems with @xmath17 where the hole will prefer to sit above and below the quantum disk , and pillar - like systems with @xmath56 where the hole will be located at the radial boundary of the disk . applying a magnetic field along the @xmath14-direction results in a different behaviour for these two systems . for the disk - like system ,
the hole is squeezed in the radial direction by the magnetic field , and the ground state of the system is the @xmath7 state for all values of @xmath124 for the pillar - like system , the magnetic field pushes the hole closer to the disk boundary , which forms a barrier for the hole
. therefore it will be energetically more favourable for the hole to jump to a higher angular momentum @xmath6 state , as then the wavefunction is able to relax away from the radial boundary . with increasing magnetic field
, we find successive @xmath6 transitions . the investigation of a system of two vertically coupled disks was done in order to compare its magnetic field dependence with the pillar - like single - disk system . as was expected , we find again angular momentum transitions of the ground state . moreover , a new feature appeared in this study , namely a spontaneous symmetry breaking , induced by both the coulomb interaction and the magnetic field .
we found though that with increasing magnetic field , it is still preferable for the hole to jump to a higher angular momentum state . investigation of a system of two vertically coupled truncated cones learned us that also in this case angular momentum transitions occur , although the system is highly asymmetric .
a system of three vertically coupled disks was investigated in the third part of the paper .
it was shown that the system with small interdot distance @xmath5 behaves similarly to that of the single pillar - like disk .
again we found angular momentum transitions with increasing magnetic field .
for a larger interdot distance however , the hole wavefunction tends to sit between the disks . in this case
, an increasing magnetic field does not lead to angular momentum transitions anymore , and the @xmath7 state remains the ground state .
no spontaneous symmetry broken states are found as the ground state .
this is similar to the single disk system where symmetry broken states occur only as excited states .
k. l. j. is supported by the `` instituut voor de aanmoediging van innovatie door wetenschap en technologie in vlaanderen '' ( iwt - vl ) and b. p. is a post - doctoral researcher with the flemish science foundation ( fwo - vl . ) .
discussions with m. hayne , m. tdi and a. matulis are gratefully acknowledged .
part of this work was supported by the fwo - vl , the belgian interuniversity attraction poles ( iuap ) , the flemish concerted action ( goa ) programmes , the university of antwerp ( vis ) and european commission growth programme nanomat project , contract no .
gsrd - ct-2001 - 00545 .
the hartree integral expresses the interaction effect of one particle on another , and is given by @xmath125 which contains a @xmath14-dependence which was absent in our previous work ( see appendix of ref .
@xcite ) . we can remove the @xmath126-dependence , since we deal with a cylindrical symmetric system .
the integral over the angle becomes the complete elliptic integral of the first kind , which converts eq .
( a1 ) into @xmath127 the radial integral has to be solved numerically .
we use a polynomial approximation for the elliptic function @xcite , namely @xmath128-[b_{0}+b_{1}x^{\prime } + b_{2}x^{\prime 2}]\ln ( x^{\prime } ) , \eqnum{a3}\ ] ] with @xmath129 and where the coefficients @xmath130 and @xmath131 are given in ref .
the commonly used trapezoidal rule will give bad results , as this elliptic function implies the appearance of a logaritmic divergence in the integrand .
therefore we introduced the so - called ` logaritmically weighted method ' which takes care of this problem .
generally , the following integral can be considered : @xmath132 which , after transformation , becomes @xmath133 with @xmath134 the discretization step and @xmath135 the number of steps . if we replace @xmath136 by @xmath137\left ( x / h\right ) , $ ] we can write ( a5 ) as @xmath138c_{i}(r , z)\ } , \eqnum{a6}\ ] ] and the remaining problem is the calculation of the coefficients @xmath139 and @xmath140 the integrals which determine the coefficients can be solved exactly , which leads to the following results : @xmath141 with @xmath142 and @xmath143 - 2r , \eqnum{a9}\ ] ] with @xmath144 r. heitz , n.n .
ledentsov , d. bimberg , m.v .
maximov , a.yu .
egorov , v.m .
ustinov , a.e .
zhukov , zh.i .
alferov , g.e .
cirlin , i.p .
shoshnikov , n.d .
zakharov , p. werner , and u. gsele , appl .
lett . * 74 * , 1701 ( 1999 ) .
we allowed for broken symmetry solutions by manually pushing the electron single particle wavefunction 0.1% more to the positive @xmath14-direction with respect to the negative @xmath14-direction in the first iteration step . | the properties of an exciton in a type ii quantum dot are studied under the influence of a perpendicular applied magnetic field .
the dot is modelled by a quantum disk with radius @xmath0 , thickness @xmath1 and the electron is confined in the disk , whereas the hole is located in the barrier .
the exciton energy and wavefunctions are calculated using a hartree - fock mesh method .
we distinguish two different regimes , namely @xmath2 ( the hole is located at the radial boundary of the disk ) and @xmath3 ( the hole is located above and below the disk ) , for which angular momentum @xmath4 transitions are predicted with increasing magnetic field .
we also considered a system of two vertically coupled dots where now an extra parameter is introduced , namely the interdot distance @xmath5 . for each @xmath6 and for a sufficient large magnetic field ,
the ground state becomes spontaneous symmetry broken in which the electron and the hole move towards one of the dots .
this transition is induced by the coulomb interaction and leads to a magnetic field induced dipole moment .
no such symmetry broken ground states are found for a single dot ( and for three vertically coupled symmetric quantum disks ) .
for a system of two vertically coupled truncated cones , which is asymmetric from the start , we still find angular momentum transitions .
for a symmetric system of three vertically coupled quantum disks , the system resembles for small @xmath5 the pillar - like regime of a single dot , where the hole tends to stay at the radial boundary , which induces angular momentum transitions with increasing magnetic field . for larger @xmath5 the hole can sit between the disks and the @xmath7 state remains the groundstate for the whole @xmath8-region .
[ theorem]acknowledgement [ theorem]algorithm [ theorem]axiom [ theorem]claim [ theorem]conclusion [ theorem]condition [ theorem]conjecture [ theorem]corollary [ theorem]criterion [ theorem]definition [ theorem]example [ theorem]exercise [ theorem]lemma [ theorem]notation [ theorem]problem [ theorem]proposition [ theorem]remark [ theorem]solution [ theorem]summary |
for convenience in read , a list of contents is given as follows .
i. preliminaries 1 .
contents 2 .
introduction 3 . symbols \ii . matrix equivalence and lattice structure 1 .
stp of matrices 2 .
matrix equivalence of matrices 3 .
lattice structure on @xmath2 4 .
monoid and quotient monoid 5
. group structure of @xmath3 6 .
semi - tensor addition and vector space structure of @xmath4 \iii .
topology on equivalence space 1 .
topology via sub - basis 2 .
fibre bundle structure on @xmath2 3 .
coordinate frame on @xmath2 4 .
inner product on @xmath2 5 .
@xmath4 as a matric space 6 .
sub - space of @xmath4 \iv .
differential structure on equivalence space 1
. bundled manifold 2 .
@xmath5 functions on @xmath4 3 . generalized inner product 4 .
vector fields 5 .
integral curves 6 .
forms 7 .
tensor fields v. lie algebra structure structure on square equivalence space 1 . lie algebra on square equivalence space 2 . bundled lie algebra 3 . bundled lie sub - algebra 4 .
further properties of @xmath6 \vi .
lie group on nonsingular equivalence space 1 .
bundled lie group 2 . relationship with @xmath6 3 .
lie subgroup of @xmath7 4 .
symmetric group \vii .
vector equivalence 1 .
equivalence of vectors of different dimensions 2 .
vector space structure on vector equivalence space 3 .
inner product and linear mappings 4 .
type-1 invariant subspace 5 .
type-2 invariant subspace 6 .
higher order linear mapping 7 .
invariant subspace on vector equivalence space \viii .
conclusion matrix theory and calculus are two classical and fundamental mathematical tools in modern science and technology .
there are two mostly used operators on the set of matrices : conventional matrix product and matrix addition .
roughly speaking , the object of matrix theory is @xmath8 , where @xmath9 is the set of all matrices .
unfortunately , unlike the arithmetic system @xmath10 , in matrix theory both @xmath11 " and @xmath12 " are restricted by the matrix dimensions . precisely speaking :
consider two matrices @xmath13 and @xmath14 .
then the product " , @xmath15 , is well posed , if and only if , @xmath16 ; the addition " @xmath17 , is defined , if and only if , @xmath18 and @xmath19 .
though there are some other matrix products such as kronecker product , hadamard product etc . , but they are of different stories @xcite . the main purpose of this paper is to develop a new matrix theory , which intendes to overcome the dimension barrier by extending the matrix product and matrix addition to two matrices which do not meet the classical dimension requirement . as generalizations of the classical ones , they should be consistent with the classical definitions .
that is , when the dimension requirements of two argument matrices in classical matrix theory are satisfied , the newly defined operators should coincide with the original one .
because of the extension of the two fundamental operators , many related concepts can be extended .
for instance , the characteristic functions , the eigenvalues and eigenvectors of a square matrix can be extended to certain non - square matrices ; lei algebraic structure can also be extended to dimension - varying square matrices .
all these extensions should be consistent with the classical ones . in one word
, we are developing the classical matrix theory but not violating any of the original matrix theory . when we were developing the extended operators we meet a serious problem :
though the extended operators are applicable to certain sets of matrices with different dimensions , they fail to be vector space anymore .
this drawback is not acceptable , because it blocked the way to extend many nice algebraic or geometric structures in matrix theory , such as lie algebraic structure , manifold structure etc .
, to the enlarged set , which includes matrices of different dimensions . to overcome this obstacle , we introduced certain equivalence relations .
then the quotient spaces , called the equivalence spaces , become vector spaces .
two equivalence relations have been proposed .
they are _ matrix equivalence _ ( me ) and _ vector equivalence _ ( ve ) . then many nice algebraic and geometric structures have been developed on the matrix equivalence spaces .
they are briefly introduced as follows : * lattice structure : the elements in each equivalent class form a lattice .
the class of spaces with different dimensions also form a lattice .
the former and the latter are homomorphic .
the lattices obtained for me and ve are isomorphic .
* topological structure : a topological structure is proposed to the equivalence space , which is built using topological sub - base .
it is then proved that under this topology the equivalence space is hausdorff ( @xmath20 ) space and is second countable . *
inner product structure : an inner product is proposed on the equivalence space . the norm ( distance ) is also obtained . it is proved that the topology induced by this norm is the same as the topology produced by using topological sub - base .
* fiber bundle structure .
a fiber bundle structure is proposed for the set of matrices ( as total space ) and the equivalent classes ( as base space ) .
the bundle structure is named the discrete bundle , because each fiber has discrete topology . *
bundle manifold structure : a ( dimension - varying ) manifold structure is proposed for the equivalence space .
its coordinate charts are constructed via the discrete bundle .
hence it is called a bundled manifold .
* bundled lie algebraic structure : a lie algebra structure is proposed for the equivalence space .
the lie algebra is of infinite dimensional , but almost all the properties of finite dimensional lie algebras remain available . *
bundled lie group structure : for the equivalence classes of square nonsingular matrices a group structure is proposed .
it has also the dimension - varying manifold structure .
both algebraic and geometric structures are consistent and hence it becomes a lie group .
the relationship of this lie group with the bundled lie algebra is also investigated . under vector equivalence ,
all the vectors form a vector space @xmath0 , and any matrix @xmath1 can be considered as an linear operator on @xmath0 .
a very important class of @xmath1 , called the bounded operator , is investigated in details . for a bounded operator @xmath1 , which could be non - square
, its characteristic function is proposed .
its eigenvalue and the corresponding eigenvector are obtained .
@xmath1-invariant subspace has been discussed in detail .
this work is motivated by the semi - tensor product ( stp ) .
the stp of matrices was proposed firstly and formally in 2001 @xcite .
then it has been used to some newton differential dynamic systems and their control problems @xcite , @xcite , @xcite .
a basic summarization was given in @xcite .
since 2008 , stp has been used to formulate and analyze boolean networks as well as general logical dynamic systems , and to solve control problems for those systems .
it has then been developed rapidly .
this is witnessed by hundreds of research papers .
a multitude of applications of stp include ( i ) logical dynamic systems @xcite , @xcite , @xcite ; ( ii ) systems biology @xcite , @xcite ; ( iii ) graph theory and formation control @xcite , @xcite ; ( iv ) circuit design and failure detection @xcite , @xcite , @xcite ; ( v ) game theory @xcite , @xcite , @xcite ; ( vi ) finite automata and symbolic dynamics @xcite , @xcite , @xcite ; ( vii ) coding and cryptography @xcite , @xcite ; ( viii ) fuzzy control @xcite , @xcite ; ( ix ) some engineering applications @xcite , @xcite ; and many other topics @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ; just to name a few . as a generalization of conventional matrix product , stp
is applicable to two matrices of arbitrary dimensions .
in addition , this generalization keeps all fundamental properties of conventional matrix product available .
therefore , it becomes a very conventional tool for investigating many matrix computation related problems .
up to this time , the main effort has been put on its applications . now
when we start to explore the mathematical foundation of stp , we found that the most significant characteristic of stp is that it overcomes the dimension barrier .
after serious thought , it can be seen that in fact stp is defined on equivalent classes .
following this thought of train , the matrix theory on equivalence space emerges .
the outline of this new matrix theory is presented in this manuscript .
for statement ease , we first give some notations : 1 .
@xmath21 : set of natural numbers ( i.e. , @xmath22 ) ; 2 .
@xmath23 : set of integers ; 3 .
@xmath24 : set of rational numbers , ( @xmath25 : set of positive rational numbers ) ; 4 .
@xmath26 : field of real numbers ; 5 .
@xmath27 : field of complex numbers ; 6 .
@xmath28 : certain field of characteristic number @xmath29 ( particularly , we can understand @xmath30 or @xmath31 ) . 7 .
@xmath32 : set of @xmath33 dimensional matrices over field @xmath28 .
when the field @xmath28 is obvious or does not affect the discussion , the superscript @xmath28 can be omitted , and as a default : @xmath30 can be assumed .
@xmath34 ( @xmath35 ) : the set of columns ( rows ) of @xmath1 ; @xmath36 ( @xmath37 ) : the @xmath38-th column ( row ) of @xmath1
@xmath39 , @xmath40 ; 10 .
@xmath41 : the @xmath38-th column of the identity matrix @xmath42 ; 11 .
@xmath43 ; 12 .
@xmath44 is a logical matrix , if @xmath45 .
the set of @xmath46 logical matrices is denoted as @xmath47 ; 13 .
assume @xmath48 . then @xmath49 $ ]
it is briefly denoted as @xmath50.\ ] ] 14 .
let @xmath51 , and @xmath52 , @xmath53 .
then @xmath1 is called a boolean matrix .
denote the set of @xmath33 dimensional boolean matrices by @xmath54 .
set of probabilistic vectors : @xmath55 16 .
set of probabilistic matrices : @xmath56 17 .
@xmath57 : @xmath58 divides @xmath59 . 18 .
@xmath60 : the greatest common divisor of @xmath61 .
@xmath62 : the least common multiple of @xmath61 . 20 .
@xmath63 ( @xmath64 ) : the left ( right ) semi - tensor product of matrices . 21 .
@xmath65 ( @xmath66 ) : the left ( right ) vector product of matrices . 22 .
@xmath67 ( @xmath68 , @xmath69 ) : the matrix equivalence ( ( left , right ) matrix equivalence ) . 23 .
@xmath70 ( @xmath71 , @xmath72 ) : the vector equivalence ( ( left , right ) vector equivalence ) .
the set of all matrices : @xmath73 25 .
@xmath74 26 .
the set of matrices : @xmath75 27 .
lattice homomorphism : @xmath76 28 .
lattice isomorphism : @xmath77 29 .
vector order : @xmath78 30 .
vector space order : @xmath79 31 .
matrix order : @xmath80 32 .
matrix space order : @xmath81 33 . the overall matrix quotient space : @xmath82 34 .
the @xmath83-matrix quotient space : @xmath84 35 .
the @xmath85-vector quotient space : @xmath86
this section gives a brief review on stp .
we refer to @xcite , @xcite , @xcite for details .
[ d2.2.1 ] let @xmath13 , @xmath14 , and @xmath87 be the least common multiple of @xmath88 and @xmath89
. then 1 .
the left stp of @xmath1 and @xmath90 , denoted by @xmath91 , is defined as @xmath92 where the @xmath93 is the kronecker product @xcite ; 2 .
the right stp of @xmath1 and @xmath90 is defined as @xmath94 in the following we mainly discuss the left stp , and briefly call the left stp as stp .
most of the properties of left stp have their corresponding ones for right stp .
please also refer to @xcite or @xcite for their major differences .
[ r2.2.2 ] if @xmath16 , both left and right stp defined in definition [ d2.2.1 ] degenerate to the conventional matrix product .
that is , stp is a generalization of the conventional matrix product .
hence , as a default , in most cases the symbol @xmath63 can be omitted ( but not @xmath64 ) .
the following proposition shows that this generalization not only keeps the main properties of conventional matrix product available , but also adds some new properties such as certain commutativity .
associativity and distribution are two fundamental properties of conventional matrix product .
when the product is generalized to stp , these two properties remain available .
[ p2.2.3 ] 1 .
( distributive law ) @xmath95 2 .
( associative law ) @xmath96 the following proposition is inherited from the conventional matrix product .
[ p2.2.4 ] 1 .
@xmath97 2 .
assume @xmath1 and @xmath90 are invertible , then @xmath98 the following proposition shows that the stp has certain commutative property .
[ p2.2.5 ] given @xmath13 . 1 .
let @xmath99 be a column vector
. then @xmath100 2 .
let @xmath99 be a row vector .
then @xmath101 to explore further commutating properties , we introduce a swap matrix .
[ d2.2.6 ] @xcite a swap matrix @xmath102}\in { \cal m}_{mn\times mn}$ ] is defined as follows : @xmath103}&=&\d_{mn}[1,m+1,\cdots,(n-1)m+1;\\ ~&~&2,m+2,\cdots,(n-1)m+2;\\ ~&~&\cdots~;~m,2m,\cdots , nm ] .
\end{array}\end{aligned}\ ] ] the following proposition shows that the swap matrix is orthogonal .
[ p2.2.7 ] @xmath104}^t = w_{[m , n]}^{-1}=w_{[n , m]}.\end{aligned}\ ] ] its fundamental function is to swap two factors .
[ p2.2.8 ] 1 .
let @xmath105 , @xmath106 be two column vectors
. then @xmath107}\ltimes x\ltimes y = y\ltimes x.\end{aligned}\ ] ] 2 .
let @xmath105 , @xmath106 be two row vectors .
then @xmath108}=y\ltimes x.\end{aligned}\ ] ] the swap matrix can also swap two factor matrices of the kronecker product @xcite , @xcite .
[ p2.2.9 ] let @xmath13 and @xmath14 .
then @xmath109}(a\otimes b)w_{[q , n]}=b\otimes a.\end{aligned}\ ] ] [ r2.2.10 ] assume @xmath110 and @xmath111 are square matrices .
then ( [ 2.2.13 ] ) becomes @xmath112}(a\otimes b)w_{[p , n]}=b\otimes a.\end{aligned}\ ] ] as a consequence , @xmath113 and @xmath114 are similar .
the following example is an application of proposition [ p2.2.9 ] .
[ e2.2.11 ] prove @xmath115 assume @xmath110 is a square matrix and @xmath116 . note that @xmath117 where @xmath118}$ ] . then @xmath119w = e^a\otimes i_k .
\end{array}\ ] ] [ r2.2.12 ] comparing the product of numbers with the product of matrices , two major differences are ( i ) matrix production has dimension restriction while the scalar product has no restriction ; ( ii ) the matrix product is not commutative while the scalar product is .
when the conventional matrix product is extended to stp , these two weaknesses have been eliminated in certain degree .
first , the dimension restriction has been removed .
second , in addition to proposition [ p2.2.5 ] , which shows certain commutativity , the use of swap matrix also provides certain commutativity property .
all these improvements make the stp more convenient in use than the conventional matrix product .
the set of all matrices ( over certain field @xmath28 ) is denoted by @xmath9 , that is @xmath120 it is obvious that stp is an operator defined as @xmath121 . observing the definition of stp carefully , it is not difficult to find that when we use stp to multiply @xmath1 with @xmath90 , instead of @xmath1 itself , we modify @xmath1 by different sizes of identity to multiply different @xmath90 s .
in fact , stp multiplies an equivalent class of @xmath1 , precisely , @xmath122 , with an equivalent class of @xmath90 , i.e. , @xmath123 . motivated by this fact
, we first propose an equivalence over set of matrices , called the identity equivalence ( @xmath68 or @xmath69 ) .
then stp can be considered as an operator over the equivalent classes .
we give a rigorous definition for the equivalence .
[ d2.3.1 ] let @xmath124 be two matrices . 1 .
@xmath1 and @xmath90 are said to be left matrix equivalent ( lme ) , denoted by @xmath125 , if there exist two identity matrices @xmath126 , @xmath127 , @xmath128 , such that @xmath129 2 .
@xmath1 and @xmath90 are said to be right matrix equivalent ( rme ) , denoted by @xmath130 , if there exist two identity matrices @xmath126 , @xmath127 , @xmath128 , such that @xmath131 [ r2.3.2 ] it is easy to verify that the lme @xmath68 ( similarly , rme @xmath69 ) is an equivalence relation .
that is , it is ( i ) self - reflexive ( @xmath132 ) ; ( ii ) symmetric ( if @xmath125 , then @xmath133 ) ; and ( iii ) transitive ( if @xmath125 , and @xmath134 , then @xmath135 ) . [ d2.3.3 ] given @xmath136 . 1 . the left equivalent class of @xmath1 is denoted by @xmath137 2
the right equivalent class of @xmath1 is denoted by @xmath138 3 .
@xmath1 is left ( right ) reducible , if there is an @xmath126 , @xmath139 , and a matrix @xmath90 , such that @xmath140 ( correspondingly , @xmath141 ) .
otherwise , @xmath1 is left ( right ) irreducible .
[ l2.3.4 ] assume @xmath142 and @xmath143 , where @xmath144 , @xmath145 and @xmath146 are co - prime , and @xmath147 then there exists a @xmath148 such that @xmath149 _ proof_. split @xmath150 into equal size blocks as @xmath151 where @xmath152 , @xmath153 .
then we have @xmath154 note that @xmath145 and @xmath146 are co - prime . comparing the entries of both sides of ( [ 2.3.3 ] ) , it is clear that ( i ) the diagonal elements of all @xmath155 are the same ; ( ii ) all other elements are zero . hence @xmath156 .
similarly , we have @xmath157 . but
( [ 2.3.1 ] ) requires @xmath158 , which is the required @xmath159 .
the conclusion follows .
@xmath160 [ t2.3.5 ] 1 . if @xmath125 , then there exists a matrix @xmath161 such that @xmath162 2 . in each class
@xmath163 there exists a unique @xmath164 , such that @xmath165 is left irreducible . _
proof_. 1 .
assume @xmath125 , that is , there exist @xmath166 and @xmath167 such that @xmath168 without loss of generality , we assume @xmath145 and @xmath146 are co - prime .
otherwise , assume their greatest common divisor is @xmath169 , the @xmath145 and @xmath146 in ( [ 2.3.5 ] ) can be replaced by @xmath170 and @xmath171 respectively .
+ assume @xmath13 and @xmath14 .
then @xmath172 since @xmath145 and @xmath146 are co - prime , we have @xmath173 split @xmath1 and @xmath90 into block forms as @xmath174 where @xmath175 and @xmath176 , @xmath177 , @xmath178 . now
( [ 2.3.5 ] ) is equivalent to @xmath179 according to lemma [ l2.3.4 ] , we have @xmath180 and @xmath181 .
define @xmath182 equation ( [ 2.3.4 ] ) follows .
2 . for each @xmath183 we can find @xmath184 irreducible such that @xmath185 . to prove it is unique , let @xmath186 and @xmath187 is irreducible and @xmath188 .
we claim that @xmath189 .
since @xmath190 , there exists @xmath191 such that @xmath192 since both @xmath184 and @xmath187 are irreducible , we have @xmath193 , which proves the claim .
@xmath160 [ r2.3.6 ] theorem [ t2.3.5 ] is also true for @xmath69 with obvious modification .
[ r2.3.7]for statement ease , we propose the following terminologies : 1 . if @xmath140 , then @xmath90 is called a divisor of @xmath1 and @xmath1 is called a multiple of @xmath90 .
if ( [ 2.3.5 ] ) holds and @xmath194 are co - prime , then the @xmath161 satisfying ( [ 2.3.4 ] ) is called the greatest common divisor of @xmath1 and @xmath90 .
moreover , @xmath195 is unique .
if ( [ 2.3.5 ] ) holds and @xmath194 are co - prime , then @xmath196 is called the least common multiple of @xmath1 and @xmath90 .
moreover , @xmath197 is unique .
( refer to fig.1 ) 4 .
consider an equivalent class @xmath198 , denote the unique irreducible element by @xmath165 , which is called the root element .
all the elements in @xmath198 can be expressed as @xmath199 @xmath200 is called the @xmath38-th element of @xmath198 .
hence , an equivalent class @xmath198 is a well ordered sequence as : @xmath201 [ fig.1 ] and @xmath195,title="fig : " ] next , we modify some classical matrix functions to make them available for the equivalence class . [ d4.2.7 ] 1 .
let @xmath110 .
then a modified determinant is defined as @xmath202^{1/n}.\end{aligned}\ ] ] 2 .
consider an equivalence of square matrices @xmath198 , the determinant " of @xmath198 is defined as @xmath203 [ p4.2.8 ] ( [ 4.2.9 ] ) is well defined , i.e. , it is independent of the choice of the representative @xmath1 . _
proof_. to see ( [ 4.2.9 ] ) is well defined , we need to check that @xmath204 implies @xmath205 .
now assume @xmath206 , @xmath207 and @xmath208 , then @xmath209^{1/k\b}=\left[\det(\lambda)\right]^{1/k},\ ] ] @xmath210^{1/k\a}=\left[\det(\lambda)\right]^{1/k}.\ ] ] it follows that ( [ 4.2.9 ] ) is well defined .
@xmath160 [ d4.2.9 ] 1 .
let @xmath110 .
then a modified trace is defined as @xmath211 2 .
consider an equivalence of square matrices @xmath198 , the trace " of @xmath198 is defined as @xmath212 similar to definition [ d4.2.7 ] , we need and can easily prove ( [ 4.2.10 ] ) is well defined .
these two functions will be used frequently in the sequel .
next , we show that @xmath213 has a lattice structure .
[ d2.3.8 ] @xcite a poset @xmath214 is a lattice if and only if for every pair @xmath215 both @xmath216 and @xmath217 exist .
let @xmath218 .
if @xmath90 is a divisor ( multiple ) of @xmath1 , then @xmath90 is said to be proceeding ( succeeding ) @xmath1 and denoted by @xmath219 ( @xmath220 ) .
then @xmath80 is a partial order for @xmath198 .
[ t2.3.9 ] @xmath221 is a lattice . _
proof_. assume @xmath218 .
it is enough to prove that the @xmath222 defined in ( [ 2.3.4 ] ) is the @xmath223 , and the @xmath224 defined in ( [ 2.3.601 ] ) is the @xmath225 . to prove @xmath226 we assume @xmath227 and @xmath228
, then we need only to prove that @xmath229 .
since @xmath227 and @xmath228 , there exist @xmath230 and @xmath231 such that @xmath232 and @xmath233 .
now @xmath234 hence @xmath235 it follows that @xmath236 since @xmath145 and @xmath146 are co - prime , we have @xmath237 and @xmath238 , where @xmath239 .
then we have @xmath240 that is @xmath241 hence , @xmath229 . to prove @xmath242 assume @xmath243 and @xmath244
. then we can prove @xmath245 in a similar way .
@xmath160 [ d2.3.10 ] @xmath198 is said to possess a property , if every @xmath246 possesses this property .
the property is also said to be consistent with the equivalence relation . in the following
some easily verifiable consistent properties are collected .
[ p2.3.11 ] 1 .
assume @xmath136 is a square matrix .
the following properties are consistent with the matrix equivalence ( @xmath68 or @xmath69 ) : * @xmath1 is orthogonal , that is @xmath247 ; * @xmath248 ; * @xmath249 ; * @xmath1 is upper ( lower ) triangle ; * @xmath1 is strictly upper ( lower ) triangle ; * @xmath1 is symmetric ( skew - symmetric ) ; * @xmath1 is diagonal ; 2 .
assume @xmath250 , @xmath251 , and @xmath252 the following property is consistent with the matrix equivalence : @xmath253 [ r2.3.12 ] as long as a property is consistent with an equivalence , then we can say if the equivalent class has the property or not .
for instance , because of proposition [ p2.3.11 ] we can say @xmath198 is orthogonal , @xmath254 , etc . denote by @xmath255
then it is clear that we have a partition as @xmath256 where @xmath25 is the set of positive rational numbers .
[ d2.7.0 ] 1 .
let @xmath257 , @xmath89 and @xmath85 are co - prime and @xmath258
. then we denote by @xmath259 and @xmath260 as @xmath261 and @xmath262 components of @xmath83 .
2 . denote the spaces of various dimensions in @xmath2 as @xmath263 assume @xmath264 , @xmath265 , @xmath266 , and @xmath267 , then @xmath268 , where @xmath269 .
one sees easily that we can define an embedding mapping @xmath270 as @xmath271 in this way , @xmath272 can be considered as a subspace of @xmath273 . the order determined by this space - subspace relation
is denoted as @xmath274 if ( [ 2.7.2 ] ) holds , @xmath272 is called a divisor of @xmath273 , and @xmath273 is called a multiple of @xmath272 .
denote by @xmath275 and @xmath276 . using the order of ( [ 2.7.2 ] ) , @xmath2 has the following structure .
[ t2.7.1 ] 1 . given @xmath277 and @xmath278 .
the greatest common divisor is @xmath279 , and the least common multiple is @xmath280 .
( please refer to fig .
2 ) 2 . assume @xmath204 , @xmath281 and @xmath282 .
then their greatest common divisor @xmath283 , and their least common multiple @xmath284 . ] next , we define @xmath285 from above discussion , the following result is obvious : [ p2.7.2 ] consider @xmath2 .
the followings are equivalent : 1 .
@xmath272 is a subspace of @xmath273 ; 2 .
@xmath145 is a factor of @xmath146 , i.e. , @xmath267 ; 3 .
@xmath286 4 .
@xmath287 using the order @xmath81 defined by ( [ 2.7.2 ] ) , it is clear that all the fixed dimension vector spaces @xmath277 , @xmath288 form a lattice .
[ p2.7.3 ] @xmath289 is a lattice with @xmath290 the following properties are easily verifiable .
[ p2.7.3 ] consider the lattice @xmath289 . 1 .
it has a smallest ( root ) subspace @xmath291 , where @xmath292 are co - prime and @xmath258 .
that is , @xmath293 but there is no largest element .
the lattice is distributive , i.e. , @xmath294 3 . for any finite set of spaces @xmath295 , @xmath296 .
there exists a smallest supper - space @xmath297 , @xmath298 , such that * @xmath299 * if @xmath300 then @xmath301 [ d2.7.4 ] let @xmath302 and @xmath303 be two lattices . @xmath302 and
@xmath303 are said to be lattice homomorphic , denoted by @xmath304 , if there exists a mapping @xmath305 satisfying the following condition : @xmath306 @xmath307 is called the homomorphism of @xmath302 and @xmath303 .
moreover , if @xmath307 is one to one and onto , @xmath302 and @xmath303 are said to be isomorphic and @xmath307 is an isomorphism , denoted by @xmath308 .
assume @xmath281 is irreducible , define @xmath309 as @xmath310 then it is easy to verify the following result .
[ p2.7.5 ] the mapping @xmath309 defined in ( [ 2.7.6 ] ) is a lattice homomorphism from @xmath311 to @xmath312 .
next , we consider the @xmath2 for different @xmath83 s .
it is also easy to verify the following result .
[ p2.7.6 ] define a mapping @xmath313 as @xmath314 the mapping @xmath315 is a lattice isomorphism . [ d2.7.7 ] let @xmath302 be a lattice and @xmath316 .
if @xmath317 is also a lattice , it is called a sub - lattice of @xmath302 . [ r2.7.701 ]
let @xmath318 be an injective ( i.e. , one - to - one ) lattice homomorphism .
then @xmath319 is a lattice isomorphism .
hence @xmath320 is a sub - lattice of @xmath303 .
if we identify @xmath321 with @xmath320 , we can simply say that @xmath321 is a sub - lattice of @xmath322 .
[ d2.7.8 ] let @xmath302 and @xmath303 be two lattices .
the product order @xmath323 defined on the product set @xmath324 is : @xmath325 if and only if @xmath326 and @xmath327 .
[ t2.7.9 ] let @xmath302 and @xmath303 be two lattices .
then @xmath328 is also a lattice , called the product lattice of @xmath302 and @xmath303 . _
proof_. let @xmath329 and @xmath330 be two elements in @xmath331 .
denote by @xmath332 and @xmath333
. then @xmath334 , @xmath335 . to see @xmath336
let @xmath337 , @xmath335 .
then @xmath338 and @xmath339 , @xmath335 .
it follows that @xmath340 and @xmath341 .
that is , @xmath342 .
we conclude that @xmath343 similarly , we set @xmath344 and @xmath345 , then we can prove that @xmath346 @xmath160 finally , we have the following result . [ e2.7.10 ]
consider the product of two lattices @xmath347 and @xmath348 .
define a mapping @xmath349 as @xmath350 assume @xmath351 and @xmath352 , then @xmath353 and @xmath354 , and by the definition of product lattice , we have @xmath355 since @xmath356 , we have @xmath357 that is , @xmath307 is a lattice homomorphism .
it is obvious that @xmath307 is not injective .
@xmath358 can not be considered as a sub - lattice of @xmath359 .
a monoid is a semigroup with identity .
we refer readers to @xcite , @xcite , @xcite for concepts and properties .
recall that @xmath360 we have the following algebraic structure . [ p2.4.1 ] the algebraic system @xmath361 is a monoid .
_ proof_. the associativity comes from the property of @xmath63 ( refer to ( [ 2.2.4 ] ) ) .
the identity element is @xmath362 .
@xmath160 one sees easily that this monoid covers numbers , vectors , and matrices of arbitrary dimensions . in the following some of its useful sub - monoids
are presented : * @xmath363 : @xmath364 where @xmath365 and @xmath366 .
+ it is obvious that @xmath367 .
( in this section @xmath368 means @xmath1 is a submonoid of @xmath90 ) .
this sub - monoid is useful for calculating the product of tensors over @xmath369 dimensional vector space @xcite .
it is particularly useful for @xmath369-valued logical dynamic systems @xcite , @xcite .
when @xmath370 , it is used for boolean dynamic systems .
+ in this sub - monoid the stp can be defined as follows : [ d2.4.2 ] 1 .
let @xmath371 be a column vector , @xmath372 a row vector .
* assume @xmath373 ( denoted by @xmath374 ) : split @xmath375 into @xmath376 equal blocks as @xmath377^t,\ ] ] where @xmath378 , @xmath379 .
define @xmath380 * assume @xmath381 ( denoted by @xmath382 ) : split @xmath383 into @xmath88 equal blocks as @xmath384,\ ] ] where @xmath385 , @xmath379 .
define @xmath386 2 .
assume @xmath13 , @xmath14 , where @xmath387 ( denoted by @xmath388 ) , or @xmath389 ( denoted by @xmath390 ) .
then @xmath391 where @xmath392 [ r2.4.3 ] 1 .
it is easy to prove that when @xmath390 or @xmath393 for some @xmath394 , this definition of left stp coincides with definition [ d2.2.1 ] . though this definition is not general ,
it has clear physical meaning .
particularly , so far this definition covers almost all the applications .
2 . unfortunately ,
this definition is not suitable for right stp .
this is a big difference between left and right stps . *
@xmath0 : @xmath395 + it is obvious that @xmath396 . + this sub - monoid consists of column vectors . in this sub - monoid
the stp is degenerated to kronecker product .
we denote by @xmath397 the sub - monoid of row vectors .
it is also clear that @xmath398 .
* @xmath399 : @xmath400 + it is obvious that @xmath401 .
this sub - monoid consists of all logical matrices .
it is used to express the product of logical mappings .
* @xmath402 : @xmath403 + it is obvious that @xmath404 .
this monoid is useful for probabilistic logical mappings .
* @xmath405 : @xmath406 + it is obvious that @xmath407 .
we use it for @xmath369-valued logical mappings .
next , we define the set of short " matrices as @xmath408 and its subset @xmath409 then we have the following result .
[ p2.4.3 ] @xmath410 _ proof_. assume @xmath13 , @xmath14 and @xmath411 , then @xmath412 and @xmath413 .
let @xmath414 . then @xmath415 .
it is easy to see that @xmath416 , so @xmath417 .
the second part is proved . as for the first part ,
assume @xmath418 .
then @xmath419\\ ~&\geq&\operatorname{rank}\left[\left(a\otimes i_{t / n}\right)\left(b\otimes i_{t / p}\right)\right.\\ ~&~&\left.\left(b^t(bb^t)^{-1}\otimes i_{t / p}\right)\right]\\ ~&=&\operatorname{rank}\left[\left(a\otimes i_{t / n}\right)\left(i_p\otimes i_{t / p}\right)\right]\\ ~&=&\operatorname{rank}\left(a\otimes i_{t / n}\right)=mt / n . \end{array}\ ] ] hence , @xmath420 .
@xmath160 similarly , we can define the set of tall " matrices @xmath421 and the set of matrices with full column rank @xmath422 .
we can also prove that @xmath423 next , we consider the quotient space @xmath424 [ d2.4.4 ] @xcite 1 . a nonempty set @xmath425 with a binary operation @xmath426 is called an algebraic system .
assume @xmath67 is an equivalence relation on an algebraic system @xmath427 .
the equivalence relation is a congruence relation , if for any @xmath428 , @xmath429 and @xmath430 , we have @xmath431 [ p2.4.5 ] consider the algebraic system @xmath432 with the equivalence relation @xmath433 .
the equivalence relation @xmath67 is congruence .
_ let @xmath434 and @xmath435 .
according to theorem [ t2.3.5 ] , there exist @xmath436 and @xmath437 such that @xmath438 denote @xmath439 then @xmath440\otimes i_{\xi}. \end{array}\ ] ] similarly , we have @xmath441\otimes i_{\eta}.\ ] ] hence we have @xmath442 .
@xmath160 according to proposition [ p2.4.5 ] , we know that @xmath63 is well defined on the quotient space @xmath443 . moreover , the following result is obvious : [ p2.4.6 ] 1 .
@xmath444 is a monoid .
2 . let @xmath445 be a sub - monoid .
then @xmath446 is a sub - monoid of @xmath443 , that is , @xmath447 since the @xmath425 in proposition [ p2.4.6 ] could be any sub - monoid of @xmath9 .
all the aforementioned sub - monoids have their corresponding quotient sub - monoids , which are the sub - monoids of @xmath443 .
for instance , @xmath448 , @xmath449 , etc .
are the sub - monoids of @xmath443 .
[ p2.5.1 ] assume @xmath450 and @xmath451 then @xmath452 .
that is , the operation @xmath63 is a mapping @xmath453 _ proof_. assume @xmath13 and @xmath14 , where @xmath454 and @xmath455 , and @xmath414
. then @xmath456 @xmath160 .
[ d2.5.3 ] 1 .
define @xmath457 then @xmath3 is closed under operator @xmath63 .
2 . define @xmath458 then @xmath459 is also closed under operator @xmath63 .
@xmath460 is said to be power equivalent , denoted by @xmath461 , if there exists an integer @xmath462 such that both @xmath463 .
denote @xmath464 [ r2.5.4 ] it is obvious that @xmath63 is consistent with @xmath465 .
hence @xmath63 is well defined on the set of equivalent classes as @xmath466 then we have the following group structure .
[ t2.5.5 ] @xmath467 is a group , which is isomorphic to @xmath468 .
precisely , assume @xmath469 then @xmath470 is defined as @xmath471 which is a group isomorphism .
[ d2.6.1 ] let @xmath472 .
precisely , @xmath13 , @xmath14 , and @xmath473 . set @xmath474
. then 1
. the left sta of @xmath1 and @xmath90 , denote by @xmath475 , is defined as @xmath476 we also denote the left semi - tensor subtraction as @xmath477 2 .
the right sta of @xmath1 and @xmath90 , denote by @xmath478 , is defined as @xmath479 we also denote the right semi - tensor subtraction as @xmath480 [ r2.6.2 ] let @xmath481 be one of the four binary operators .
then it is easy to verify that 1 . if @xmath472 , the @xmath482 ; 2 .
if @xmath1 and @xmath90 are as in definition [ d2.6.1 ] , then @xmath483 ; 3 .
set @xmath484 , then @xmath485 and @xmath486 .
so @xmath487 can also be defined by using column numbers respectively , e.g. , @xmath488 etc .
[ t2.6.4 ] consider the algebraic system @xmath489 , where @xmath490 and @xmath433 ( or @xmath491 and @xmath492 ) .
then the equivalence relation @xmath67 is a congruence relation with respect to @xmath487 . _
proof_. we prove one case , where @xmath493 and @xmath433 .
proofs for other cases are similar .
assume @xmath494 and @xmath495 .
set @xmath496 and @xmath497 , then @xmath498 @xmath499 where @xmath500 , @xmath501 , @xmath502 are certain numbers .
now consider @xmath503 .
assume @xmath504 , @xmath505 , @xmath506 .
then we have @xmath507\otimes i_{\xi}. \end{array}\end{aligned}\ ] ] similarly , we have @xmath508\otimes i_{\zeta}.\end{aligned}\ ] ] ( [ 2.6.9 ] ) and ( [ 2.6.10 ] ) imply that @xmath509 .
@xmath160 define the left and right quotient spaces @xmath510 and @xmath511 respectively as @xmath512 @xmath513 according to theorem [ t2.6.4 ] , the operation @xmath475 ( or @xmath514 ) can be extended to @xmath510 as @xmath515 similarly , we can define @xmath478 ( or @xmath516 ) on the quotient space@xmath511 as @xmath517 the following result is important , and the verification is straightforward .
[ t2.6.5 ] using the definitions in ( [ 2.6.13 ] ) ( correspondingly , ( [ 2.6.14 ] ) ) , the quotient space @xmath518 ( correspondingly , @xmath519 ) is a vector space .
[ r2.6.6 ] as a consequence , @xmath520 ( or @xmath521 ) is an abelian group .
[ r2.3.13 ] recall example [ e2.2.11 ] , it shows that the exponential function @xmath522 is well defined on the quotient space @xmath523 . since each @xmath524 has a unique left ( or right ) identity irreducible element @xmath165 ( or @xmath525 ) such that @xmath526 ( or @xmath527 ) , in general , we can use the irreducible element , which is also called the root element of an equivalent class , as the representation of this class .
but this is not compulsory . for notational and statement ease ,
hereafter we consider @xmath510 only unless elsewhere stated . as a convention
, the omitted script ( @xmath528 " or@xmath529 ") means @xmath528 .
for instance , @xmath530 , @xmath433 , @xmath531 etc .
this subsection builds step by step a topology on quotient space @xmath4 using a sub - basis .
first , we consider the partition ( [ 2.5.2 ] ) , it is natural to assume that each @xmath2 is a clopen subset in @xmath9 , because distinct @xmath83 s correspond to distinct shapes of matrices . now inside each @xmath2 we assume @xmath532 are co - prime and @xmath533 .
then @xmath534 where @xmath535 because of the similar reason , we also assume each @xmath277 is clopen .
overall , we have a set structure on @xmath9 as @xmath536 [ d0.2.1 ] a natural topology on @xmath9 , denoted by @xmath537 , consists of 1 . a partition of countable clopen subsets @xmath277 , @xmath257 , @xmath538 ; 2 .
the conventional euclidean @xmath539 topology for @xmath277 .
next , we consider the quotient space @xmath424 it is clear that @xmath540 moreover , ( [ 0.2.2 ] ) is also a partition .
hence each @xmath4 can be considered as a clopen set in @xmath443 .
we are , therefore , interested only in constructing a topology on each @xmath4 .
[ d0.2.2 ] 1 .
consider @xmath541 as an euclidean space @xmath539 with conventional euclidean topology .
assume @xmath542 is an open set .
define a subset @xmath543 as follows : @xmath544 2 .
let @xmath545 3 .
using @xmath546 , we construct a set of subsets @xmath547 as @xmath548 taking @xmath549 as a topological sub - basis , the topology generated by @xmath425 is denoted by @xmath550 , which makes @xmath551 a topological space .
( we refer to @xcite for a topology produced from a sub - basis . ) + note that the topological basis consists of the set of finite intersections of @xmath552 .
[ fig.0.1 ] : an element in topological basis , title="fig : " ] [ r0.2.3 ] 1 .
it is clear that @xmath550 makes @xmath553 a topological space .
the topological basis is @xmath554 3 .
fig 3 depicts an element in the topological basis .
here @xmath555 , @xmath556 are two open discs with rational center and rational radius .
then @xmath557 and @xmath558 are two elements in the sub - basis , and @xmath559 is an element in the basis .
[ t0.2.4 ] the topological space @xmath553 is a second countable , hausdorff ( or @xmath20 ) space . _
proof_. to see @xmath553 is second countable , it is easy to see that @xmath546 is countable
. then @xmath560 , as countable union of countable set , is countable .
finally , @xmath561 , as the finite subset of a countable set , is countable .
next , consider @xmath562 .
let @xmath563 and @xmath564 be their irreducible elements respectively . if @xmath565 for the same @xmath38 , then we can find two open sets @xmath566 , @xmath567 , such that @xmath568 and @xmath569 . then by definition , @xmath570 and @xmath571 , @xmath572 .
finally , assume @xmath573 , @xmath574 and @xmath575 .
let @xmath576 .
then @xmath577 since @xmath578 , we can find @xmath579 , @xmath567 and @xmath580 and @xmath581 .
that is , @xmath582 and @xmath583 separate @xmath198 and @xmath584 .
@xmath160 if we consider @xmath585 as a product topological space , then @xmath550 is the quotient topology of the standard product topology on the product space @xmath9 defined by ( [ 0.2.5 ] ) .
( we refer to @xcite for product topology . )
[ d0.3.1 ] @xcite a bundle is a triple @xmath586 , where @xmath587 and @xmath90 are two topological spaces and @xmath588 is a continuous map . @xmath587 and @xmath90 are called the total space and base space respectively . for each @xmath589
, @xmath590 is called the fibre of the bundle at @xmath589 . observing the two topologies @xmath537 and @xmath550 constructed in previous subsection ,
the following result is obvious : [ p0.3.2 ] @xmath591 , is a bundle , where @xmath592 is the natural projection , i.e. , @xmath593 [ fig.0.2 ] [ r0.3.3 ] 1 . of course
, @xmath594 is also a bundle .
but it is of less interest because it is a discrete union of @xmath591 , @xmath257 .
2 . consider an equivalent class @xmath595 , where @xmath165 is irreducible .
then the fibre over @xmath198 is a discrete set : @xmath596 hence this fiber bundle is named discrete bundle .
3 . fig 4 illustrates the fiber bundle structure of @xmath591 . here
@xmath563 and @xmath564 are irreducible @xmath597 and @xmath598 .
their fibers are depicted in the fig 4 .
we can define a set of cross sections @xcite @xmath599 as : @xmath600 it is clear that @xmath601 , where @xmath602 is the identity mapping on @xmath4 .
next , we consider some truncated sub - bundles of @xmath591 .
[ d0.3.3 ] assume @xmath365 . 1
. set @xmath603}_{\mu}:=\left\{m_{m\times n}\;|\ ; m_{m\times n}\in { \cal m}_{\mu } , ~\mbox{and}~m|k\mu_y\right\}.\end{aligned}\ ] ] then @xmath604}_{\mu}$ ] is called the @xmath369-upper bounded subspace of @xmath2 .
@xmath605}_{\mu}:={\cal m}^{[\cdot , k]}_{\mu}/\sim$ ] is called the @xmath369-upper bounded subspace of @xmath4 .
the natural projection @xmath606}_{\mu}\ra \sigma^{[\cdot , k]}_{\mu}$ ] is obviously defined .
then we have the following bundle structure . [ p0.3.4 ] @xmath607},\pr,\sigma_{\mu}^{[\cdot , k]}\right)$ ]
is a sub - bundle of @xmath591 .
precisely speaking , the following graph ( [ 0.3.2 ] ) is commutative , where @xmath608 and @xmath609 are including mappings .
( [ 0.3.2 ] ) is also called the bundle morphism .
[ r0.3.5 ] 1 . in a truncated sub - algebra
there is a maximum cross section @xmath611 and a minimum cross section ( i.e. , root leaf ) @xmath612 , where @xmath292 are co - prime and @xmath258 .
2 . let @xmath613 , @xmath614 be a finite set of cross sections of @xmath591 .
set @xmath615 , then there exists a smallest truncation @xmath616}$ ] , which contains @xmath613 , @xmath614 as its cross sections .
[ d0.3.6 ] 1 .
define @xmath617}_{\mu}:=\left\{m\in { \cal m}_{\mu}^s\;\big|\ ; k|s\right\},\ ] ] which is called the @xmath369-lower bounded subspace of @xmath2 .
2 . define the quotient space @xmath618}_{\mu}:={\cal m}^{[k,\cdot]}_{\mu}/\sim,\ ] ] which is called the @xmath369-lower bounded subspace of @xmath4 .
3 . assume @xmath267 .
define @xmath619}_{\mu}:={\cal m}^{[\a,\cdot]}_{\mu}\bigcap { \cal m}_{\mu}^{[\cdot,\b]},\ ] ] which is called the @xmath620$]- bounded subspace of @xmath2 .
4 . define the quotient space @xmath621}_{\mu}:={\cal m}^{[\a,\b]}_{\mu}/\sim,\ ] ] which is called the @xmath620$]-bounded subspace of @xmath4 .
[ r0.3.7 ] proposition [ p0.3.4 ] is also true for the two other truncated forms : @xmath622},\pr,\sigma_{\mu}^{[k,\cdot]}\right)$ ] and @xmath623},\pr,\sigma_{\mu}^{[\a,\b]}\right)$ ] respectively . precisely speaking , in ( [ 0.3.2 ] )
if both @xmath604}_{\mu}$ ] and @xmath616}$ ] are replaced by @xmath624}_{\mu}$ ] and @xmath625}$ ] respectively , or by @xmath626}_{\mu}$ ] and @xmath627}$ ] respectively , ( [ 0.3.2 ] ) remains commutative .
it is obvious that @xmath628 is an infinite dimensional vector space .
since each @xmath629 has finite coordinate expression , we may try to avoid using a basis with infinite elements . to this end
, we construct a set of consistent " coordinate frames as @xmath630
. then @xmath198 can be expressed by @xmath631 , @xmath632 , and so on .
moreover , @xmath633 is a subset ( or @xmath634 is part of coordinate elements in @xmath635 ) .
then it seems that @xmath198 can always be expressed in @xmath634 no matter which representative is chosen .
the purpose of this section is to build such a set of consistent coordinate sub - frames , which form an overall coordinate frame .
assume @xmath264 , @xmath265 , @xmath266 , and @xmath267 , then @xmath268 , where @xmath269 .
the order determined by this space - subspace relation is denoted as @xmath636 one sees easily that we can define an embedding mapping @xmath270 as @xmath637 in this way , @xmath272 can be considered as a subspace of @xmath273 . our purpose is to make @xmath272 a coordinate subspace , which means it is generated by part of coordinate variables of @xmath273 . to this end
, we build a set of orthonormal basis on @xmath273 as follows : assume @xmath638 and @xmath639 . splitting @xmath640 into @xmath641 blocks , where each block is of dimension @xmath642 , yields @xmath643
then for each @xmath644 we construct a basis , which consists of three classes : * class 1 : @xmath645 where @xmath646 that is , for @xmath647-th block , at each non - diagonal position @xmath648 , set it to be @xmath362 , and all other entries to be @xmath29 . * class 2 : @xmath649 that is , at each @xmath647-th block , set @xmath650 as a basis element . * class 3 : @xmath651 that is , set @xmath652 as the rest of basis elements of the diagonal subspace of @xmath647-th block , which are orthogonal to @xmath653 .
let @xmath654 .
the frobenius inner product is defined as @xmath655 correspondingly , the frobenius norm is defined as @xmath656 using frobenius inner product , it is easy to verify the following result .
[ p0.4.1 ] 1 . set @xmath657 then @xmath658 is an orthonormal basis for @xmath647-th block ; 2 . set @xmath659
then @xmath90 is an orthonormal basis for @xmath273 .
[ e0.4.2 ] consider @xmath660 . for any @xmath661 , we split @xmath1 as @xmath662
then we build the orthonormal basis block - wise as @xmath663 the orthonormal basis as proposed above is @xmath664 assume @xmath665 and @xmath666 and @xmath667 . using ( [ 0.4.7 ] ) , matrix @xmath1 can be expressed as @xmath668 when @xmath272 is merged into @xmath273 as a subspace , matrix @xmath669 can be expressed as @xmath670 [ d0.4.3 ] assume @xmath671 and @xmath672 .
the embedding mapping @xmath673 is defined as @xmath674 to be consistent with this , we define the projection as follows : [ d0.4.4 ] assume @xmath675 , where @xmath672 and @xmath676 .
the projection @xmath677 is defined as @xmath678 where @xmath679 according to the above construction , it is easy to verify the following : [ p0.4.5 ] the composed mapping @xmath680 is an identity mapping . let @xmath681 .
it is well known that the frobenius inner product of @xmath1 and @xmath90 is @xmath682 the frobenius norm is @xmath683 the following lemma comes from a straightforward computation .
[ l0.5.1 ] let @xmath684 .
then @xmath685 [ d0.5.2 ] let @xmath472 , where @xmath686 and @xmath687 .
then the weighted inner product of @xmath688 is defined as @xmath689 where @xmath690 is the least common multiple of @xmath145 and @xmath146 .
using lemma [ l0.5.1 ] and definition [ d0.5.2 ] , we have the following property .
[ p0.5.201 ] let @xmath472 , if @xmath1 and @xmath90 are orthogonal , i.e. , @xmath691 , then @xmath692 and @xmath693 are also orthogonal .
now we are ready to define the inner product on @xmath4 .
[ d0.5.3 ] let @xmath694 .
their inner product is defined as @xmath695 the following proposition shows that ( [ 0.5.3 ] ) is well defined .
[ p0.5.4 ] definition [ d0.5.3 ] is well defined .
that is , ( [ 0.5.3 ] ) is independent of the choice of the representatives @xmath1 and @xmath90 . _
proof_. assume @xmath563 and @xmath564 are irreducible .
then it is enough to prove that @xmath696 assume @xmath597 and @xmath598 .
let @xmath697 denote by @xmath698 , @xmath699 , and @xmath700 . using ( [ 0.5.2 ] ) , we have @xmath701 @xmath160 [ d0.5.5 ] @xcite a real or complex vector space @xmath375 is an inner - product space , if there is a mapping @xmath702 , denoted by @xmath703 , satisfying 1 .
@xmath704 2 .
@xmath705 where the bar stands for complex conjugate .
3 . @xmath706 4 .
@xmath707 by definition it is easy to verify the following result . [ t0.5.6 ] the vector space @xmath708 with the inner product defined by ( [ 0.5.3 ] ) is an inner product space .
next , we define the norm of @xmath629 . @xmath709 the following is some standard results for inner product space .
[ t0.5.7 ] assume @xmath694 .
then we have the following 1 .
( schwarz inequality ) @xmath710 3 .
( parallelogram law ) @xmath711 note that the above properties show that @xmath4 is a normed space .
finally , we present the generalized pythagorean theorem : [ t0.5.8 ] let @xmath712 , @xmath713 be an orthogonal set .
then @xmath714 a natural question is that is @xmath4 a hilbert space ? "
unfortunately , this is not true .
this fact is shown in the following counter - example . [ e0.5.9 ] define a sequence of elements , denoted as @xmath715 , as follows : @xmath716 is arbitrary .
define @xmath717 inductively as @xmath718 where @xmath719 is defined as @xmath720 first , we claim that @xmath721 is a cauchy sequence .
let @xmath722 .
then @xmath723 then we prove by contradiction that it does not converge to any element .
assume it converges to @xmath724 , it is enough to consider the following three cases : * case 1 , assume @xmath725 and @xmath726 . then @xmath727 similar to ( [ 0.5.10 ] )
we can prove that @xmath728 hence the sequence can not converge to @xmath729 . * case 2 , assume @xmath725 and @xmath730 .
note that @xmath731 is orthogonal to @xmath732 , then it is clear that @xmath733 similar to ( [ 0.5.10 ] ) , we can prove that @xmath734 hence the sequence can not converge to @xmath724 . * case 3 , @xmath735 , where @xmath736 is odd . corresponding to case 1 , we assume @xmath737
. then we have @xmath738 and @xmath739 so the sequence can not converge to @xmath729 . + corresponding to case 2 , assume @xmath740 .
using proposition [ p0.5.201 ] , a similar argument shows that the sequence can not converge to @xmath729 too . using the norm defined in previous section
one sees easily that @xmath4 is a matric space : [ t0.6.1 ] @xmath4 with distance @xmath741 is a matric space . [ t0.6.2 ]
consider @xmath4 .
the topology deduced by the distance @xmath742 , denoted by @xmath743 is exactly the same as the topology @xmath550 defined in definition [ d0.2.2 ] . _
proof_. assume @xmath744 and @xmath745 .
then there exists a ball @xmath746 such that @xmath747 , where @xmath748 .
assume @xmath749 and @xmath750 .
now we can construct a ball @xmath751 , where @xmath752 .
note that @xmath753 is a sub - basis element of @xmath550 , and hence is an open set in @xmath754 . by continuity , as @xmath752
small enough , @xmath755 implies @xmath756 .
that is , @xmath757 which means @xmath758
. hence , @xmath759 .
conversely , assume @xmath760 . then there exists a basic open set @xmath761 such that @xmath762 .
express @xmath763 where @xmath764 , @xmath765 .
for each @xmath200 , we can find @xmath766 , where @xmath767 , @xmath765 .
choosing @xmath768 small enough such that @xmath769 then we have @xmath770 that is , @xmath771 .
hence , @xmath772 .
we conclude that @xmath773 .
@xmath160 [ d0.6.3 ] @xcite 1 .
a topological space is regular ( or @xmath774 ) if for each closed set @xmath375 and @xmath775 there exist open neighborhoods @xmath776 of @xmath262 and @xmath777 of @xmath375 , such that @xmath778 .
a topological space is normal ( or @xmath779 ) if for each pair of closed sets @xmath375 and @xmath383 there exist open neighborhoods @xmath777 of @xmath375 and @xmath780 of @xmath383 , such that @xmath781 .
since a matric space is regular and normal , as a corollary of theorem [ t0.6.2 ] , we have the following result .
[ c0.6.4 ] the topological space @xmath553 , defined in definition [ d0.2.2 ] , is both regular and normal .
note that @xmath782 finally , we show some properties of @xmath4 .
[ p0.6.5 ] @xmath4 is convex .
hence it is arcwise connected . _
proof_. assume @xmath694 .
then it is clear that @xmath783.\ ] ] so @xmath4 is convex .
let @xmath159 go from @xmath362 to @xmath29 , we have a path connecting @xmath198 and @xmath584 .
@xmath160 [ p0.6.6 ] @xmath4 and @xmath784 are isometric spaces . _
proof_. consider the transpose : @xmath785 then it is obvious that @xmath786 hence the transpose is an isometry .
moreover , its inverse is itself .
consider the @xmath369-upper bounded subspace @xmath616}$ ] .
we have [ p0.7.1 ] @xmath616}$ ] is a hilbert space . _
proof_. since @xmath616}$ ] is a finite dimensional vector space and any finite dimensional inner product space is hilbert @xcite . @xmath160
[ p0.7.2 ] @xcite let @xmath587 be an inner product space , @xmath787 be a hilbert subspace
. 1 . for each @xmath788 there exists a unique @xmath789 , called the projection of @xmath262 on @xmath790 , such that @xmath791 2 .
@xmath792 + is the subspace orthogonal to @xmath790 .
3 . @xmath793 + where @xmath794 stands for orthogonal sum .
using above proposition , we consider the projection : @xmath795}$ ] .
let @xmath796 .
assume @xmath797 , @xmath798 .
then the norm of @xmath799 is : @xmath800 set @xmath638 @xmath639 , and @xmath801 .
we split @xmath1 as @xmath802 where @xmath803 , @xmath804 .
set @xmath805 then it is easy to calculate that @xmath806 we conclude that [ p0.7.3 ] let @xmath629 , @xmath807}$ ] . precisely speaking , @xmath796 . using the above notations ,
the projection of @xmath198 is @xmath808 where @xmath669 is defined in ( [ 0.7.6 ] ) .
we give an example to depict this .
[ e0.7.4 ] given @xmath809 we consider its projection to @xmath810}$ ] .
denote by @xmath811 . using formulas ( [ 0.7.6])-([0.7.7 ] )
, we have @xmath812 then we have @xmath813 where @xmath814 . \end{array}\ ] ] it is easy to verify that @xmath815 and @xmath198 are mutually orthogonal .
we also have @xmath625}$ ] and @xmath627}$ ] ( where @xmath267 ) as metric subspaces of @xmath4 .
finally , we would like to point out that since @xmath4 is an infinity dimensional vector space , it is possible that @xmath4 is isometric to its proper subspace .
for instance , consider the following example .
[ e0.7.5 ] consider a mapping @xmath816}$ ] defined by @xmath817 .
it is clear that this mapping satisfies @xmath818 that is , @xmath2 can be isometrically embedded into its proper subspace .
unlike the conventional manifolds , which have fixed dimensions , this section explores a new kinds of manifolds , called the bundled manifold . intuitively speaking , it is a fiber bundle , which has fibers belonging to manifolds of different dimensions . to begin with ,
the following definition is proposed , which is a mimic to the definition of an @xmath88 dimension manifold @xcite .
[ d3.2.1 ] let @xmath819 be a topological space . 1 .
an open set @xmath820 is said to be a simple coordinate chart , if there is an open set @xmath821 and a homeomorphism @xmath822 .
the integer @xmath823 is said to be the dimension of @xmath824 .
an open set @xmath820 is said to be a bundled coordinate chart , if there exist finite simple coordinate charts @xmath825 with homeomorphisms @xmath826 , @xmath827 , @xmath828 is set @xmath824 depending , such that @xmath829 3 .
let @xmath824 , @xmath830 be two bundled coordinate charts .
@xmath831 , @xmath832 .
@xmath824 and @xmath830 are said to be @xmath5 comparable if for any @xmath825 and @xmath833 , as long as their dimensions are equal , they are @xmath5 comparable .
( where @xmath529 could be @xmath834 , that is they are @xmath835 comparable ; or @xmath836 , which means they are analytically comparable . ) [ r3.2.2 ] in definition [ d3.2.1 ] for a bundled coordinate chart @xmath837 , we can , without loss of generality , assume @xmath838 , @xmath827 are distinct . because simple coordinate charts of same dimension can be put together by set union @xmath839 .
hereafter , this is assumed .
[ fig.3.1 ] to depict the bundle coordinate chart , we refer to fig 5 , where two bundled coordinate charts @xmath824 and @xmath830 are described .
note that @xmath840 and @xmath841 , where @xmath842 are simple coordinate charts . also , @xmath824 and @xmath830 are coordinate neighborhoods of @xmath262 and @xmath261 respectively .
next , we assume that @xmath843 , @xmath844 , and @xmath845 .
@xmath824 and @xmath830 are @xmath5 comparable , if and only if the two mappings @xmath846 are @xmath5 . as a convention
, we assume @xmath847 , @xmath848 .
[ d3.2.2 ] a topological space @xmath322 is a bundled @xmath5 ( or @xmath835 , or analytic , denoted by @xmath849 ) manifold , if the following conditions are satisfied . 1 .
@xmath322 is second countable and hausdorff .
2 . there exists an open cover of @xmath322 , described as @xmath850 where each @xmath851 is a bundled coordinate chart .
moreover , any two bundled coordinate charts in @xmath852 are @xmath5 comparable .
3 . if a bundled coordinate chart @xmath853 is comparable with @xmath851 , @xmath854 , then @xmath855 .
it is obvious that the topological structure of @xmath4 with natural ( @xmath539 ) coordinates on each cross section ( or leaf ) meets the above requirements for a bundled manifold .
hence , we have the following result . [ t3.2.3 ]
@xmath4 is a bundled analytic manifold .
_ condition 1 has been proved in theorem [ t0.2.4 ] . for condition @xmath856 , set @xmath638 , @xmath639 and @xmath857 choosing any finite open subset @xmath858 , @xmath859 , @xmath860 , and constructing corresponding @xmath861 . set @xmath862 , where @xmath863 .
define @xmath864 then @xmath853 is an open cover of @xmath322 .
identity mappings from @xmath865 makes any two @xmath866 and @xmath867 being @xmath849 comparable . as for condition 3 , just add all bundled coordinate charts which are comparable with @xmath853 into @xmath853 , the condition is satisfied .
@xmath160 next , we consider the lattice - related coordinates on @xmath4 .
consider @xmath4 and assume @xmath638 and @xmath639 .
then @xmath4 has leafs @xmath868 where @xmath869 . consider an element @xmath870 ,
then there exists a unique irreducible @xmath871 such that @xmath872 .
now assume @xmath873 .
as defined above , @xmath874 is the root leaf of @xmath875 .
it is obvious that @xmath875 has different coordinate representations on different leafs .
but because of the subspace lattice structure , they must be consistent . fig .
6 shows the lattice - related subspaces .
any geometric objects defined on its root leaf must be consistent with its representations on all embedded spaces and projected spaces .
[ fig.3.2 ] as shown in fig .
6 the following subspaces are related : * class 1 ( embedded elements ) : starting from @xmath876 , we have @xmath877 * class 2 ( projected elements ) : let @xmath878 , where @xmath879
. then @xmath880 particularly , @xmath881 satisfies @xmath882 * class 3 ( embedded elements from projected elements ) : starting from any @xmath883 , @xmath879 , we have @xmath884 [ r3.2.4 ] 1 .
classes 1 - 3 are the set of coordinates , which are related to a given irreducible element @xmath885 .
2 . elements in class 1 are particularly important .
say , we may firstly define an root element on @xmath874 , such as @xmath886 .
then we use it to get an equivalent class , such as @xmath198 , and use the elements in this class to perform certain calculation , such as stp .
all the elements in the equivalent class , such as @xmath198 , are of class 1 .
the elements in subspace and their equivalent classes are less important .
sometimes we may concern only the elements of class 1 , say for stp etc .
4 . the subspace elements obtained by project mapping may not be uniformed " with the object obtained from the real subspaces of the original space .
more discussion will be seen in the sequel .
because of the above argument , sometimes we may consider only the equivalent classes which have their root elements defined on their root leaf .
therefore , the objects may only be defined on the multiple of the root leaf ( root space ) .
[ d3.3.1 ] let @xmath322 be a bundled manifold , @xmath888 is called a @xmath5 function , if for each simple coordinate chart @xmath889 @xmath890 is @xmath5 .
the set of @xmath5 functions on @xmath322 is denoted by @xmath891 .
assume @xmath892 , @xmath472 and @xmath204 .
then @xmath893 is well defined on @xmath4 means it is defined on different leafs consistently , and hence on leafs corresponding to @xmath1 and @xmath90 we have @xmath894 . to this end
, the @xmath893 can be constructed as follows : [ d3.3.2 ] assume @xmath893 is firstly defined on root leaf @xmath874 as @xmath895
. then we extend it to other leafs as : * step 1 .
let @xmath896 .
then @xmath897 where @xmath898 .
assume @xmath899 .
if @xmath900 , @xmath901 has already been defined in step 1 .
so we assume @xmath902 .
then @xmath903 + note that in step 2 , @xmath904 is allowed .
then it is easy to verify the following : [ p3.3.3 ] the function @xmath893 defined in definition [ d3.3.2 ] is consistent with the equivalence @xmath67 .
hence it is well defined on @xmath4 .
[ e3.3.4 ] consider @xmath905 , and assume @xmath893 is defined on its root leaf @xmath906 firstly as @xmath907 then we can determine the other expressions of @xmath893 as follows : * consider @xmath908 : @xmath909 * consider @xmath910 .
let @xmath911 .
then @xmath912 + similarly , for any @xmath913 we have @xmath914 * consider @xmath915 .
. then @xmath917 \end{array}\ ] ] + similarly , for @xmath918 ( @xmath919 ) we have @xmath920 .
\end{array}\ ] ] [ r3.3.5 ] for a smooth function @xmath893 defined firstly on @xmath874 , its extensions to both @xmath921}$ ] and @xmath922}$ ] are consistently defined .
hence , @xmath893 is completely well posed on @xmath4 .
we define the generalized frobenius inner product as follows .
[ d3.4.1 ] given @xmath13 and @xmath14 . 1 .
( special case ) : assume @xmath923 and @xmath924 .
split @xmath90 into equal blocks as @xmath925 where @xmath926 , @xmath765 ; @xmath927 .
then the generalized frobenius inner product of @xmath1 and @xmath90 is defined as @xmath928 note that here @xmath929 is the standard frobenius inner product defined in ( [ 0.5.01 ] ) .
( general case ) : assume @xmath13 and @xmath14 and let the great common divisor of @xmath930 be @xmath931 , and the great common divisor of @xmath932 be @xmath933 .
denote by @xmath934 and @xmath935 , @xmath936 and @xmath937 .
then we split @xmath1 into @xmath938 blocks as @xmath939 where @xmath940 , @xmath941 ; @xmath942 .
then the generalized frobenius inner product of @xmath1 and @xmath90 is defined as @xmath943 note that here @xmath944 is the ( case 1 ) generalized frobenius inner product defined in ( [ 3.4.1 ] ) .
[ e3.4.2 ] let @xmath945 and @xmath946 note that @xmath947 , @xmath948 , @xmath949 , @xmath950 , @xmath951 , @xmath952 .
then we split @xmath1 and @xmath90 as follows @xmath953 were @xmath954 , @xmath955 . finally , we have @xmath956 [ d3.4.3 ] assume @xmath686 and @xmath957 .
@xmath958 , @xmath959 , @xmath960 , @xmath961 , @xmath962 , @xmath963 . since @xmath964 are co - prime , denote @xmath965 , then @xmath966 and @xmath967 .
split @xmath1 as @xmath968 where @xmath969 ; and split @xmath90 as @xmath970 where @xmath971 .
then the generalized weighted inner product is defined as @xmath972 , \end{array}\end{aligned}\ ] ] where @xmath973 are defined in ( [ 0.5.2 ] ) .
[ d3.4.4 ] assume @xmath629 and @xmath974 .
then the generalized inner product of @xmath198 and @xmath584 , denoted by @xmath975 , is defined as @xmath976 of course , we need to prove that ( [ 3.4.4 ] ) is independent of the choice of representatives @xmath1 and @xmath90 .
this is verified by a straightforward computation .
next , we would like to define another inner product " called the @xmath977-inner product , where @xmath978 .
first we introduce a new notation : [ 3.4.401 ] let @xmath979 , @xmath83 is said to be superior to @xmath977 , denoted by @xmath980 if @xmath981 and @xmath982 .
the @xmath977 inner product is a mapping @xmath983 .
[ d3.4.5 ] assume @xmath984 , @xmath985 and @xmath986 , @xmath987 . denote @xmath988 , @xmath989 , @xmath990 , and @xmath991 , then the @xmath977-inner product of @xmath1 and @xmath90 is defined as follows : split @xmath992 where @xmath993 , and @xmath994 where @xmath995 .
then the @xmath977-inner product of @xmath1 and @xmath90 is defined as @xmath996 where @xmath997 [ d3.4.6 ] assume @xmath629 and @xmath974 , where @xmath998 and @xmath987 .
then the @xmath977-inner product of @xmath198 and @xmath584 is defined as @xmath999 [ r3.4.7 ] 1 .
it is easy to verify that ( [ 3.4.7 ] ) is independent of the choice of @xmath1 and @xmath90 .
hence the @xmath977-inner product is well defined .
2 . definition [ d3.4.3 ] ( or definition [ d3.4.4 ] ) is a special case of definition [ d3.4.5 ] ( correspondingly , definition [ d3.4.6 ] ) .
3 . definition [ d3.4.1 ] can not be extended to the equivalence space , because it depends on the choice of representatives .
4 . unlike the generalized inner product defined in definition [ d3.4.3 ] ( as well as definition [ d3.4.4 ] ) , the @xmath977-inner product is defined on a subset of @xmath9 ( or @xmath1000 ) . using @xmath977-inner product
, we have the following . [ p3.4.8 ] assume @xmath1001 is a linear mapping .
then there exists a matrix @xmath1002 , called the structure matrix of @xmath307 , such that @xmath1003 [ d3.5.1 ] let @xmath322 be a bundled manifold and @xmath1004 the tangent space of @xmath322 .
@xmath1005 is called a @xmath5 vector field , if for each simple coordinate chart @xmath824 , @xmath1006 is @xmath5 .
the set of @xmath5 vector fields on @xmath322 is denoted by @xmath1007 we express a vector field in a matrix form .
that is , let @xmath1008 .
then @xmath1009\in { \cal m}_{m\times n}.\ ] ] similar to smooth functions , the vector fields on @xmath2 can be defined as follows : [ d3.5.2 ] assume @xmath1010 is firstly defined on @xmath1011 as @xmath1012 , i.e. , @xmath874 is the root leaf of @xmath1010
. then we extend it to other leafs as : * step 1 .
let @xmath896 .
then for @xmath1013 where @xmath898 .
assume @xmath899 .
if @xmath900 , @xmath1014 has already been defined in step 1 .
so we assume @xmath902 .
then @xmath1015 next , we consider the computation of the related expressions of a vector field , which is originally defined on its root leaf . * to calculate ( [ 3.5.1 ] ) we first set @xmath1016 to get @xmath1017 .
then we split @xmath1018 into @xmath1019 blocks as @xmath1020 $ ] , where each @xmath1021 .
then @xmath1022\in t_x\left({\cal m}_{tp\times tq}\right)$ ] , and @xmath1023\right).\end{aligned}\ ] ] * to calculate ( [ 3.5.2 ] ) we split @xmath261 into @xmath1019 blocks as @xmath1024 $ ] , where each @xmath1025 .
then @xmath1026 is obtained by replacing @xmath262 by @xmath1027\in { \cal m}_{tp\times tq}$ ] as @xmath1028\right).\end{aligned}\ ] ] it follows that @xmath1029 then it is easy to verify the following : [ p3.5.3 ] the vector field @xmath375 defined in definition [ d3.5.2 ] is consistent with the equivalence @xmath1030}}$ ] .
hence the equivalent class @xmath1010 is well defined on @xmath1031}\right)$ ] .
[ r3.5.301 ] in fact , proposition [ p3.5.3 ] only claim that the representations on embedded supper spaces are consistent , which is obviously weaker than proposition [ p3.3.3 ] .
please refer to remark [ r3.2.4 ] and the latter remark [ r3.6.5.4 ] for the extension of @xmath1010 to the projected subspace .
[ e3.5.4 ] consider @xmath1032 .
assume @xmath375 is firstly defined on @xmath1033 as @xmath1034 where @xmath1035 and @xmath1036 @xmath1037 then we consider the expression of @xmath375 on the other cross sections : * consider @xmath1038 .
set @xmath1039 where @xmath1040 . according to ( [ 3.5.1 ] ) ,
@xmath1041 and @xmath1042 * consider @xmath1043 .
set @xmath1044 where @xmath1045 .
+ consider the projection @xmath1046 : @xmath1047 according to ( [ 3.5.2 ] ) , @xmath1048\otimes i_3,\ ] ] where @xmath1049 + similarly , for @xmath913 we have @xmath1050\otimes i_n\in t\left({\cal m}_{1/2}^n\right),\ ] ] where @xmath1051 * consider @xmath1052 .
set @xmath1053 where @xmath1054 .
consider the projection @xmath1055 : @xmath1056 according to ( [ 3.5.2 ] ) , @xmath1057 where @xmath1058 + similarly , for @xmath918 we have @xmath1059 as @xmath1060 where @xmath1061 [ d3.6.5.1 ] let @xmath1062 be a vector field .
then for each @xmath629 there exists a curve @xmath1063 such that @xmath1064 and @xmath1065 which is called the integral curve of @xmath1066 , starting from @xmath198 .
in fact , the integral curve of @xmath1066 is a bundled integral curve .
assume @xmath1067 , where @xmath1068 is irreducible . here
@xmath376 , @xmath88 may not be co - prime but @xmath1069 .
so we denote @xmath1070 .
then on this root leaf we denote @xmath1071 and the integral curve of @xmath1072 , starting from @xmath184 is a standard one , which is the cross section of the bundled integral curve @xmath1010 passing through @xmath184 .
that is , it is the solution of @xmath1073 we may denote the solution as @xmath1074 next , we consider the other cross sections of the integral curve , which correspond to the cross sections of @xmath1066 respectively .
recall definition [ d3.5.2 ] , we can get the cross sections of the bundled integral curve on each leafs by using the cross sections of @xmath1066 on corresponding leafs .
the following result is then obvious : [ t3.6.5.2 ] assume @xmath1075 .
the corresponding cross section of the bundled integral curve is the integral curve , denoted by @xmath1076 satisfying @xmath1077 where @xmath1078 is defined by ( [ 3.5.2 ] ) .
7 shows the integral curve and its projections on each leaf .
the following result comes from the construction directly .
[ t3.6.5.3 ] assume @xmath1079 , @xmath1080 , @xmath1081 , @xmath1082 , and @xmath1083 , @xmath919 .
@xmath1084 is defined firstly on @xmath1085 , where @xmath1086 .
then there exists a one - to - one correspondence between the two cross sections ( or corresponding integral curves ) .
precisely , @xmath1087\\ \phi^{\xi^{\b}}_t\left(a_{\b}\right)=\bd_k\left[\phi^{\xi^{\a}}_t\left(\bd_k(a_{\b})\right)\right ] .
\end{array}\end{aligned}\ ] ] particularly , assume the vector field is a linear vector field and @xmath1088 , then we have @xmath1089 moreover , on @xmath1090 , the cross section of the integral curve can be expressed by modifying ( [ 3.6.5.6 ] ) as @xmath1091 [ r3.6.5.4 ] * note that if @xmath1066 is firstly defined on @xmath1011 , @xmath1092 and @xmath1093 .
then @xmath1094 is obtained through the following two steps : ( i ) restrict @xmath1072 on @xmath1095 as @xmath1096 ( ii ) project @xmath1097 onto the tangent space of the subspace @xmath1098 . because of ( ii ) , @xmath1099 does not correspond to @xmath1100 .
hence , the relationship demonstrated in theorems [ t3.6.5.2 ] and [ t3.6.5.3 ] are not available for the integral curves of @xmath1094 and @xmath1072 respectively .
+ because of this argument , if @xmath1066 is firstly defined on @xmath1011 , where @xmath874 is the root leaf for @xmath1066 , then we are mainly concerning the integral curves of @xmath1078 on @xmath1101 , where @xmath1075 . *
the projections of @xmath1102 onto subspaces @xmath1094 are useful in some other problems .
for instance , assume @xmath1066 is a constant vector field .
using notations in theorem [ t3.6.5.3 ] , we set @xmath1103 where @xmath653 is defined in ( [ 0.4.4 ] ) .
then the projection of @xmath1104 on @xmath1105 , denoted by @xmath1106 , is well defined . moreover ,
( [ 3.6.5.5 ] ) becomes @xmath1107\\ \phi^{\xi^{\b}_{{\cal d}}}_t\left(a_{\b}\right)=\bd_k\left[\phi^{\xi^{\a}}_t\left(\bd_k(a_{\b})\right)\right ] .
\end{array}\end{aligned}\ ] ] [ e3.6.5.5 ] recall example [ e3.3.4 ] .
since it is a linear vector field , it is easy to calculate that the cross section on @xmath1108 can be expressed as in ( [ 3.6.5.6 ] ) , where @xmath1109 and @xmath1110 consider the cross section on @xmath1111 with @xmath1112 . then @xmath1113 an integral manifold of an involutive distribution on @xmath1114 can be defined and calculated in a similar way .
[ d3.6.6.1 ] let @xmath322 be a bundled manifold .
@xmath1115 is called a @xmath5 co - vector field ( or one form ) , if for each simple coordinate chart @xmath824 @xmath1116 is a @xmath5 co - vector field .
the set of @xmath5 co - vector fields on @xmath322 is denoted by @xmath1117 we express a co - vector field in matrix form .
that is , let @xmath1118 .
then @xmath1119\in { \cal m}_{m\times n}.\ ] ] similar to the construction of vector fields , the co - vector fields on @xmath4 can be established as follows : [ d3.6.6.2 ] assume @xmath1120 is firstly defined on @xmath1121 as @xmath1122 , where @xmath874 is the root leaf of @xmath1120
. then we extend it to other leafs as : * step 1 .
let @xmath896 . @xmath1123,\end{aligned}\ ] ] where @xmath898 .
assume @xmath899 .
if @xmath900 , @xmath1124 has already been defined in step 1 .
so we assume @xmath902 , @xmath919
. then @xmath1125.\end{aligned}\ ] ] similar to the calculation of vector fields , to calculate ( [ 3.6.6.1 ] ) we first set @xmath1126 to get @xmath1127
. then we split @xmath1128 into @xmath1019 blocks as @xmath1129 $ ] , where each @xmath1130 .
then @xmath1131\in t^*_x\left({\cal m}_{tp\times tq}\right)$ ] , where the entries are @xmath1132\right),\quad i=1,\cdots , tp ; j=1,\cdots , tq.\end{aligned}\ ] ] to calculate ( [ 3.6.6.2 ] ) we split @xmath261 into @xmath1019 blocks as @xmath1024 $ ] , where each @xmath1025 .
then @xmath1133 is obtained by replacing @xmath262 by @xmath1027\in { \cal m}_{tp\times tq}$ ] as @xmath1134\right).\end{aligned}\ ] ] it follows that @xmath1135 then it is easy to verify the following : [ p3.6.6.3 ] the co - vector field @xmath1120 defined in definition [ d3.6.6.2 ] is consistent with the equivalence @xmath1030}}$ ] .
hence it is well defined on @xmath1136}$ ] .
[ p3.6.6.4 ] let @xmath1137 and @xmath1138 .
then the action of @xmath1120 on @xmath1010 is defined as @xmath1139 similar to vector field case , if the co - vector field is firstly defined on @xmath874 , then we can assume @xmath1120 is only defined on @xmath1140 satisfying @xmath1141 the set of tensor fields on @xmath4 of covariant order @xmath145 and contravariant order @xmath146 is denoted by @xmath1142 . to avoid complexity ,
we consider only the covariant tensor , @xmath1143 .
[ d3.6.7.1 ] a covariant tensor field @xmath1144 is a multi - linear mapping @xmath1145 assume @xmath1146 is defined on root leaf at @xmath1147 as @xmath1148 , @xmath292 are co - prime and @xmath258 .
the calculation is performed as follows : construct the structure matrix of @xmath1149 as @xmath1150 where @xmath1151 consider @xmath1152 with its irreducible element @xmath1153 , where @xmath874 is the root leaf of @xmath1154 .
@xmath1155 is expressed in matrix form as @xmath1156\right):=v_s^k,\quad k=1,\cdots.\a . \end{array}\ ] ] then we have [ p3.6.7.2 ] @xmath1157 next , we calculate the expressions of @xmath1146 on other leafs .
the following algorithm can be verified to be consistent on @xmath1158}$ ] , where @xmath1159 is the root leaf of @xmath1146 .
[ a3.6.7.3 ] assume @xmath1146 is firstly defined on @xmath1160 as @xmath1161
. then we extend it to other leafs as : * step 1 .
denote @xmath1162 .
let @xmath1163 and @xmath1164 .
then @xmath1165 can be calculated by constructing @xmath1166 ( @xmath1167 ) as follows : first , split @xmath1168 into @xmath1169 blocks @xmath1170 where each block @xmath1171
. then we set @xmath1172\in { \cal m}_{(\tau p)^{\a}\times ( \tau q)^{\a}},\end{aligned}\ ] ] where @xmath1173 * step 2 .
assume @xmath1174 .
if @xmath1175 , @xmath1176 has already been defined in step 1 .
so we assume @xmath1177 , where @xmath919
. then @xmath1178 can be calculated by constructing @xmath1179,\quad z\in { \cal m}_{\ell p\times \ell q}.\end{aligned}\ ] ] [ e3.6.7.4 ] consider a covariant tensor field @xmath1180 , where @xmath1181 .
moreover , @xmath1146 is firstly defined at @xmath1182 with its structure matrix as @xmath1183 1 .
let @xmath1184 be defined at @xmath1185 as @xmath1186 evaluate @xmath1187 .
+ first , @xmath1188 then @xmath1189 2 . expressing @xmath1146 on leaf @xmath1190 :
note that @xmath1191 where @xmath1192\in { \cal m}_{4\times 6}$ ] .
+ using ( [ 3.6.7.4 ] ) , we have @xmath1193 where @xmath1194
consider the vector space of the equivalent classes of square matrices @xmath1195 .
since @xmath1000 is closed under the stp , more algebraic structures may be posed on it .
first , polynomials ; second , lie algebra structure .
then we extend some fundamental concepts of matrices to their equivalent classes .
[ d4.1.1 ] 1 .
@xmath198 is nonsingular ( symmetric , skew symmetric , positive / negative ( semi-)definite , upper / lower ( strictly ) triangular , diagonal , etc . )
if its irreducible element @xmath165 is ( equivalently , every @xmath200 is ) .
2 . @xmath198 and @xmath584 are similar , denoted by @xmath1196 , if there exists a nonsingular @xmath1197 such that @xmath1198 3 .
@xmath198 and @xmath584 are congruent , denoted by @xmath1199 , if there exists a nonsingular @xmath1197 such that @xmath1200 4 .
@xmath1201 is called the jordan normal form of @xmath198 , if the irreducible element @xmath1202 is the jordan normal form of @xmath165 .
[ d4.2.1 ] @xcite a set @xmath1203 with two operators @xmath1204 is a ring .
if the followings hold : 1 .
@xmath1205 is an abelian group ; 2 .
@xmath1206 is a monoid ; 3 .
( distributive rule ) @xmath1207 observing @xmath1208 , which consists of all square matrices , both @xmath475 ( including @xmath514 ) and @xmath63 are well defined .
unfortunately , @xmath1209 is not a group because there is no unit element . since both @xmath475 and @xmath63 are consistent with the equivalence @xmath67 , we consider @xmath1210 .
then it is easy to verify the following : [ p4.2.2 ] @xmath1211 is a ring . consider a polynomial on a ring @xmath1203 , @xmath1212 , as @xmath1213 it is obvious that this is well defined .
set @xmath1214 , then @xmath1215 defined in ( [ 4.2.1 ] ) is also well defined on @xmath1203 .
particularly , the coefficients @xmath1216 can be chosen as @xmath1217 for @xmath1218 , @xmath1219 .
then @xmath1215 is as a standard " polynomial .
unless elsewhere is stated , in this paper only such standard polynomials are considered . for any @xmath524
, the polynomial @xmath1220 is well defined , and it is clear that @xmath1221 using taylor expansion , we can consider general matrix functions .
for instance , we have the following result : [ t4.2.3 ] let @xmath1222 be an analytic function .
then @xmath1223 is well defined provided @xmath1224 is well defined .
moreover , @xmath1225 in fact , the above result can be extended to multi - variable case .
[ d4.2.4 ] let @xmath1226 be a @xmath369-variable analytic function .
then @xmath1227 is a well posed expression , where @xmath1228 , @xmath827 .
assume @xmath1229 , @xmath827 , then @xmath1230 is a realization of @xmath1227 . particularly , if @xmath1231 , @xmath379 , we call @xmath1230 a realization on @xmath529-th leaf
. similar to ( [ 4.2.3 ] ) , we also have @xmath1232 provided @xmath1230 is well defined . in the following we consider some fundamental matrix functions for @xmath1000 .
we refer to @xcite for the definitions and basic properties of some fundamental matrix functions . using these acknowledges ,
the following result is obvious : [ t4.2.5 ] let @xmath1233 in @xmath1000 ( i.e. , @xmath1 , @xmath90 are square matrices ) .
then the following hold : 1 .
@xmath1234 2 .
if @xmath198 is real skew symmetric , then @xmath1235 is orthogonal .
assume @xmath90 is invertible , we denote @xmath1236 . then @xmath1237 4 .
let @xmath688 be closed enough to identity so that @xmath1238 and @xmath1239 are defined .
then @xmath1240 many known results for matrix functions can be extended to @xmath1000 .
for instance , it is easy to prove the following euler fromula : [ p4.2.6 ] consider @xmath30 and let @xmath524 . then the euler formula holds .
that is , @xmath1241 recall the modification of trace and determinant in definitions [ d4.2.7 ] and [ d4.2.9 ] .
the following proposition shows that for the modifications the relationship between @xmath1242 and @xmath1243 @xcite remain available .
[ p4.2.10 ] assume @xmath30 ( or @xmath31 ) , @xmath524 , then @xmath1244 _ proof .
_ let @xmath1245 and @xmath1246 .
then @xmath1247 @xmath160 next , we consider the characteristic polynomial of an equivalent class , they comes from standard matrix theory @xcite . [ d4.2.11 ] let @xmath524 .
@xmath563 is its irreducible element .
then @xmath1248 is called the characteristic polynomial of @xmath198 .
the following result is an immediate consequence of the definition .
[ t4.2.12 ] ( cayley - hamilton ) let @xmath1249 be the characteristic polynomial of @xmath198
. then @xmath1250 [ r4.2.13 ] 1 .
if we choose @xmath1251 and calculate the characteristic polynomial of @xmath1252 , then @xmath1253 .
so @xmath1254 is equivalent to @xmath1255 .
2 . choosing any @xmath1256 , the corresponding minimal polynomials @xmath1257 are the same .
so we have unique minimal polynomial as @xmath1258 .
consider the vector space of the equivalent classes of square matrices @xmath1195 , this section gives a lie algebraic structure to it .
@xcite[d4.3.1 ] a lie algebra is a vector space @xmath1259 over some field @xmath28 with a binary operation @xmath1260:~g\times g\ra g$ ] , satisfying 1 .
( bi - linearity ) @xmath1261&=&\alpha [ a , c]+\beta [ b , c];\cr [ c,\alpha a+\beta b]&=&\alpha [ c , a]+\beta [ c , b ] , \end{array}\end{aligned}\ ] ] where @xmath1262 .
( skew - symmetry ) @xmath1263=-[b , a];\end{aligned}\ ] ] 3 .
( jacobi identity ) @xmath1264+[b,[c , a]]+[c,[a , b]]=0,\\ \forall a , b , c\in g. \end{array}\end{aligned}\ ] ] [ d4.3.2 ] let @xmath1265 be a discrete bundle . if 1 .
@xmath1266 is a lie algebra ; 2 .
@xmath1267 is a lie algebra , @xmath288 ; 3 .
the restriction @xmath1268 is a lie algebra homomorphism , @xmath288 , then @xmath1266 is called a bundled lie algebra .
( we refer to definition [ d0.3.1 ] and remark [ d0.3.3 ] for the concept of discrete bundle . ) on vector space @xmath1000 we define an operation @xmath1260:~\sigma\times \sigma\ra \sigma$ ] as @xmath1269:=\a\ltimes \b\lminus \b\ltimes \a.\end{aligned}\ ] ] then we have the following lie algebra : [ t4.3.3 ] the vector space @xmath1270 with lie bracket @xmath1260 $ ] defined in ( [ 4.3.4 ] ) , is a bundled lie algebra , denoted by @xmath6 . _
proof_. let @xmath1271 , where @xmath1272 .
then it is clear that @xmath1273 is a discrete bundle .
next , we prove @xmath1274\right)$ ] is a lie algebra
. equations ( [ 4.3.1 ] ) and ( [ 4.3.2 ] ) are obvious .
we prove ( [ 4.3.3 ] ) only .
assume @xmath563 , @xmath564 and @xmath1275 are irreducible , and @xmath1276 , @xmath1277 , and @xmath1278 .
let @xmath1279 .
then it is easy to verify that @xmath1280=\\ ~\left < \left[(a_1\otimes i_{t / m } ) , [ ( b_1\otimes i_{t / n}),(c_1\otimes i_{t / r})]\right]\right>. \end{array}\end{aligned}\ ] ] similarly , we have @xmath1281=\\ ~\left < \left[(b_1\otimes
i_{t / n } ) , [ ( c_1\otimes i_{t / r}),(a_1\otimes i_{t / m})]\right]\right>. \end{array}\end{aligned}\ ] ] @xmath1282=\\ ~\left < \left[(c_1\otimes
i_{t / r } ) , [ ( a_1\otimes i_{t / m}),(b_1\otimes i_{t / n})]\right]\right>. \end{array}\end{aligned}\ ] ] since ( [ 4.3.3 ] ) is true for any @xmath1283 , it is true for @xmath1284 , @xmath1285 , and @xmath1286 . using this fact and equations ( [ 4.3.5])-([4.3.7 ] ) , we have @xmath1287\lplus [ \b,[\left < c\right>,\a]]\lplus [ \left < c\right>,[\a,\b]]\\ = \left < \left[(a_1\otimes i_{t / m}),[(b_1\otimes i_{t / n}),(c_1\otimes i_{t / r})]\right]\right.\\ + ~~\left[(b_1\otimes
i_{t / n } ) , [ ( c_1\otimes i_{t / r}),(a_1\otimes i_{t / m})]\right]~~\\ + \left
. \left[(c_1\otimes i_{t / r } ) , [ ( a_1\otimes i_{t / m}),(b_1\otimes i_{t / n})]\right]\right>\\ = \left<0\right>=0 . \end{array}\ ] ] let the lie algebraic structure on @xmath1288 be @xmath1289 ( where @xmath30 or @xmath31 ) .
it follows from the consistence of @xmath475 and @xmath63 with the equivalence that @xmath1290 is a lie algebra homomorphism .
@xmath160 this section considers some useful lie sub - algebras of lie algebra @xmath6 .
assume @xmath1259 is a lie algebra , @xmath1291 is a vector subspace .
then @xmath1292 is called a lie sub - algebra if and only if @xmath1293\subset h$ ] .
[ d4.4.1 ] let @xmath1265 be a bundled lie algebra .
if 1 .
@xmath1294 is a lie sub - algebra of @xmath1266 ; 2 .
@xmath1295 is a lie sub - algebra of @xmath1267 , @xmath288 ; 3 .
the restriction @xmath1296 is a lie algebra homomorphism , @xmath288 , then @xmath1297 is called a bundled lie sub - algebra of @xmath1298 .
it is well known that there are some useful lie sub - algebras of lie algebra @xmath1289 . when @xmath1289 , @xmath1299 , are generalized to the bundled lie algebra @xmath6 ( over @xmath1000 ) , the corresponding lie sub - algebras are investigated one - by - one in this section .
* bundled orthogonal lie sub - algebra + [ d4.1.2 ] @xmath524 is said to be symmetric ( skew symmetric ) if @xmath1300 ( @xmath1301 ) , @xmath1302 .
+ the symmetric ( skew symmetric ) @xmath198 is well defined because if @xmath204 and @xmath1300 ( or @xmath1301 ) , then so is @xmath90 .
it is also easy to verify the following : + [ p4.4.2.2 ] assume @xmath198 and @xmath584 are skew symmetric , then so is @xmath1303 $ ] .
+ we , therefore , can define the following bundled lie sub - algebra . + [ d4.4.2.3 ] @xmath1304 is called the bundled orthogonal algebra .
* bundled special linear algebra + [ d4.4.2.4 ] @xmath1305 is called the bundled special linear algebra .
+ similar to the case of orthogonal algebra , it is easy to verify that @xmath1306 is a lie sub - algebra of @xmath6 . *
bundled upper triangular algebra + [ d4.4.2.5 ] @xmath1307 is called the bundled upper triangular algebra .
+ similarly , we can define bundled lower triangular algebras .
* bundled strictly upper triangular algebra + [ d4.4.2.6 ] @xmath1308 is called the bundled strictly upper triangular algebra . *
bundled diagonal algebra + [ d4.4.2.7 ] @xmath1309 is called the bundled diagonal algebra .
* bundled symplectic algebra + [ d4.4.2.8 ] @xmath1310 is called the bundled symplectic algebra .
+ @xmath1311 + where
@xmath1312 [ d4.4.2.9 ] a lie sub - algebra @xmath1313 is called an idea , if @xmath1314\in { \cal j}.\end{aligned}\ ] ] [ e4.4.2.10 ] @xmath1315 is an idea of @xmath1316 .
because @xmath1317=\operatorname{tr}(g\ltimes h\lminus h\ltimes g)=0,\quad \forall g\in \gl({{\mathbb f } } ) , \forall h\in \sl({{\mathbb f}}).\ ] ] hence , @xmath1318\subset
\sl({{\mathbb f}})$ ] .
many properties of the sub - algebra of @xmath1319 can be extended to the sub - algebra of @xmath6 .
the following is an example .
[ p4.4.2.11 ] @xmath1320 _ proof .
_ it is obvious that @xmath1321 since @xmath1322 , which means @xmath1323 the conclusion follows . @xmath160 [ e4.4.2.12 ] 1 .
denote by @xmath1324 it is obvious that @xmath1325 is a vector sub - space of @xmath6 . let @xmath1326 . then @xmath1327\a\lplus \a \left[\left
< x\right>,\left < y\right>\right]^t\\ = \left < x\right>\left < y\right>\a\lminus \left < y\right>\left < x\right>\a\\ ~+\a \left < y^t\right>\left < x^t\right>\lminus \left < x^t\right>\left < y^t\right>\a\\ = \left < x\right>\left < y\right>\a\lminus \left < y\right>\left < x\right>\a\\ ~+\a \left < y\right>\left < x\right>\a \lminus \left < x\right>\left < y\right>\a\\ = 0 . \end{array}\ ] ] hence , @xmath1328 is a bundled lie sub - algebra .
2 . assume @xmath198 and @xmath584 are congruent .
that is , there exists a non - singular @xmath1197 such that @xmath1329 .
then it is easy to verify that @xmath1330 is an isomorphism , where @xmath1331 [ d4.5.1]let @xmath1332 . 1 .
the @xmath89-truncated equivalent class is defined as @xmath1333}:=\left\{a_i\in \a\;\big|\;i|p\right\}.\end{aligned}\ ] ] 2 .
the @xmath89-truncated square matrices is defined as @xmath1334}:=\left\{a\in { \cal m}_{n\times n}\;\big|\ ; n|p\right\}.\end{aligned}\ ] ] 3 .
the @xmath89-truncated equivalence space is defined as @xmath1335}:= { \cal m}^{[\cdot , p]}/\sim.\end{aligned}\ ] ] [ r4.5.2 ] [ d4.5.3 ] @xcite
let @xmath1259 be a lie algebra . 1 .
denote the derived serious as @xmath1350 $ ] , and @xmath1351 then @xmath1259 is solvable , if there exists an @xmath1352 such that @xmath1353 .
2 . denote the descending central series as @xmath1354 $ ] , and @xmath1355,\quad k=1,2,\cdots.\ ] ] the @xmath1259 is nilpotent , if there exists an @xmath1352 such that @xmath1356 .
[ d4.5.4 ] let @xmath1357 be a sub - algebra of @xmath6 . 1 .
@xmath191 is solvable , if for any @xmath1332 , the truncated sub - algebra @xmath1358}$ ] is solvable .
2 . @xmath191 is nilpotent , if for any @xmath1332 , the truncated sub - algebra @xmath1359}$ ] is nilpotent .
[ d4.5.5 ] @xcite let @xmath1259 be a lie algebra . 1 .
@xmath1259 is simple if it has no non - trivial idea ( the only ideas are @xmath1360 and @xmath1259 itself ) .
@xmath1259 is semi - simple if it has no solvable idea except @xmath1360 .
though the following proposition is simple , it is also fundamental .
[ p4.5.6 ] the lie algebra @xmath1346}({{\mathbb f}})$ ] is isomorphic to the classical linear algebra @xmath1361 . _
proof_. first we construct a mapping @xmath1362}({{\mathbb f}})\ra \gl(p,{{\mathbb f}})$ ] as follows : assume @xmath1363}({{\mathbb f}})$ ] and @xmath563 is irreducible .
say , @xmath1336 , then by definition , @xmath1364 .
denote by @xmath1365 , then define @xmath1366 set @xmath1367}({{\mathbb f}})$ ] as @xmath1368}({{\mathbb f}}),\ ] ] then it is ready to verify that @xmath608 is a bijective mapping and @xmath1369 . by the definitions of @xmath475 and
@xmath1260 $ ] , it is obvious that @xmath608 is a lie algebra isomorphism .
@xmath160 the following properties are available for classical @xmath1319 @xcite , @xcite . using proposition [ p4.5.6 ] , it is easy to verify that they are also available for @xmath1316 .
[ p4.5.601 ] let @xmath1370 be a lie sub - algebra . 1 .
if @xmath1259 is nilpotent then it is solvable .
2 . if @xmath1259 is solvable ( or nilpotent ) then
so is its sub - algebra , its homomorphic image .
3 . if @xmath1291 is an idea of @xmath1259 and @xmath1292 and @xmath1371 are solvable , then @xmath1259 is also solvable .
[ d4.5.7 ] let @xmath1372 .
the adjoint representation @xmath1373 is defined as @xmath1374.\end{aligned}\ ] ] to see ( [ 4.5.7 ] ) is well defined , we have to prove that @xmath1375 it follows from the consistence of @xmath63 and @xmath475 ( @xmath514 ) with @xmath68 immediately . [ e4.5.701 ]
consider @xmath524 .
assume @xmath198 is nilpotent , that is , there is a @xmath1376 such that @xmath1377 .
then @xmath1378 is also nilpotent .
note that @xmath1379 if and only if @xmath1380 .
similarly , @xmath1381 if and only if @xmath1382 .
hence , we need only to show that @xmath1383 is nilpotent , where @xmath246 and @xmath110 .
it is easy to verify that @xmath1384 as @xmath1385 , it is clear that @xmath1386 , @xmath1387 .
it follows that @xmath1388 [ d4.5.8 ] 1 .
let @xmath1389 .
then the killing form @xmath1390 is defined as ( we refer to @xcite for original definition .
the following definition is with a mild modification . )
@xmath1391 2 .
assume @xmath1392 .
the killing form @xmath1393 is defined as @xmath1394 to see the killing form is well defined , we also need to prove @xmath1395 similar to ( [ 4.5.8 ] ) , it can be verified by a straightforward calculation .
because of the equations ( [ 4.5.8 ] ) and ( [ 4.5.10 ] ) , the following properties of finite dimensional lie algebras @xcite can easily be extended to @xmath6 : [ p4.5.9 ] consider @xmath1396 .
let @xmath198 , @xmath1397 , @xmath1398 , @xmath584 , @xmath1399 , @xmath1400
. then 1 .
@xmath1401 2 .
@xmath1402 3 .
@xmath1403 4 .
let @xmath1291 be an idea of @xmath1259 , and @xmath1404 .
then @xmath1405 the right hand side means the killing form on the ideal @xmath1292 .
a sub - algebra @xmath1406 is semi - simple , if and only if , its killing form is non - degenerated .
the engel theorem can easily be extended to @xmath1316 : [ t4.5.10 ] let @xmath1407 be a bundled lie sub - algebra .
assume each @xmath1408 is nilpotent , ( i.e. , for each @xmath1409 there exists a @xmath1376 such that @xmath1410 ) . 1 .
if @xmath1259 is finitely generated , then there exists a vector @xmath1411 ( of suitable dimension ) such that @xmath1412 2 .
@xmath1259 is nilpotent .
[ d4.5.11 ] @xcite let @xmath830 be an @xmath88 dimensional vector space .
a flag in @xmath830 is a chain of subspaces @xmath1413 with @xmath1414 .
let @xmath1415 be an endomorphism of @xmath830 .
@xmath1 is said to stabilize this flag if @xmath1416 lie theorem can be extended to @xmath6 as follows .
[ t4.5.12 ] assume @xmath1417 is a solvable lie sub - algebra .
then for any @xmath1418 there is a flag of ideals @xmath1419 , such that the truncated @xmath1420}$ ] stabilizes the flag .
[ c4.5.13 ] assume @xmath1417 is a lie sub - algebra .
@xmath1259 is solvable if and only if , @xmath1421 is nilpotent .
[ e4.5.14 ] consider the bundled lie sub - algebras @xmath1422 and @xmath1423 .
it is easy to verify the following : 1 .
@xmath1422 is solvable ; 2 .
@xmath1423 is nilpotent .
even though @xmath1316 is an infinite dimensional lie algebra , it has almost all the properties of finite dimensional lie algebras .
this claim can be verified one by one easily .
the reason is : @xmath6 is a union of finite dimensional lie algebras .
consider @xmath523 , we define a subset @xmath1424 we emphasize the fact that @xmath1425 is an open subset of @xmath1000 . for an open subset of bundled manifolds we have the following result .
[ p5.7.1.1 ] let @xmath322 be a bundled manifold , and @xmath1426 an open subset of @xmath322 .
then @xmath1426 is also a bundled manifold .
it is enough to construct an open cover of @xmath1426 .
starting from the open cover of @xmath322 , which is denoted as @xmath850 we construct @xmath1427 then we can prove that it is @xmath5 ( @xmath835 or @xmath849 ) comparable as long as @xmath852 is .
verifying other conditions is trivial .
@xmath160 [ c5.7.1.2 ] @xmath1425 is a bundled manifold
. note that , proposition [ p2.4.5 ] shows that @xmath1428 is well define .
[ d5.7.1.3 ] a topological space @xmath1429 is a bundled lie group , if 1 .
it is a bundled analytic manifold ; 2 .
it is a group ; 3 .
the product @xmath1430 and the inverse mapping @xmath1431 are analytic .
the following result is an immediate consequence of the definition .
[ t5.7.1.4 ] @xmath1425 is a bundled lie group . _
proof_. we already known then @xmath1425 is a bundled analytic manifold .
we first prove @xmath1425 is a group .
it is ready to verify that @xmath1432 is the identity .
moreover , @xmath1433 . the conclusion follows . using a simple coordinate chart , it is obvious that the inverse and product are two analytic mappings .
@xmath160 denote by @xmath1434 and @xmath1435 consider the bundle : @xmath1436 , where the map @xmath1437 is the national projection .
it has a natural sub - bundle : @xmath1438 via the following bundle morphism as @xmath1439 in fact , the projection leads to lie group homomorphism .
[ t5.7.2.0 ] 1 . with natural group and differential structures , @xmath1440 is a lie group .
2 . consider the projection @xmath592 .
restrict it to each leaf yields @xmath1441 then @xmath1442 is a lie group homomorphism .
3 . set the image set as @xmath1443 .
then @xmath1444 is a lie sub - group .
moreover , @xmath1445 is a lie group isomorphism .
[ d5.7.2.2 ] a vector field @xmath1446 , ( where for each @xmath1447 , @xmath1448 ) , is called a left - invariant vector field , if for any @xmath1449 @xmath1450 then it is easy to verify the following relationship between @xmath1425 and @xmath1316 : [ t5.7.2.3 ] the corresponding lie algebra of the bundled lie group @xmath1425 is @xmath1316 in the following natural sense : @xmath1451 that is , @xmath1316 is a lie algebra isomorphic to the lie algebra consists of the vectors on the tangent space of @xmath1425 at identity .
then these vectors generate the left - invariant vector fields which form the tangent space at any @xmath1449 .
let @xmath1452 be the natural mapping @xmath1453 .
then we have the following commutative picture : @xmath1454 where @xmath251 .
recall ( [ 2.2.15 ] ) , we know that the exponential mapping @xmath522 is well defined @xmath1455 . graph ( [ 5.7.2.3 ] ) also shows the relationship between @xmath1316 and @xmath1425 , which is a generalization of the relationship between @xmath1319 and @xmath1456 .
it has been discussed that @xmath1316 has some useful lie sub - algebras .
it is obvious that @xmath1425 has some lie sub - groups , corresponding to those sub - algebras of @xmath1316 .
they are briefly discussed as follows .
* bundled orthogonal lie sub - group + [ d5.7.3.1 ] @xmath1449 is said to be orthogonal , if @xmath1457 .
+ it is also easy to verify the following : + [ p5.7.3.2 ] assume @xmath198 and @xmath584 are orthogonal , then so is @xmath1458 .
+ we , therefore , can define the following bundled lie sub - group of @xmath1425 as follows .
+ [ d5.7.3.3 ] @xmath1459 is called the bundled orthogonal group . + it is easy to verify the following proposition : + [ p5.7.3.4 ] consider the bundled orthogonal group .
1 . @xmath1460 is a lie sub - group of @xmath1425 , i.e. , @xmath1461 .
2 . @xmath1462 then @xmath1463 .
3 . the lie algebra for both @xmath1460 and @xmath1464 is @xmath1465 .
* bundled special linear group + [ d5.7.3.5 ] @xmath1466 is called the bundled special linear group .
+ similar to the case of orthogonal algebra , it is easy to verify the following : + [ t5.7.3.6 ] consider the bundled special linear group .
1 . @xmath1467 is a lie sub - group of @xmath1425 , i.e. , @xmath1468 .
the lie algebra of @xmath1467 is @xmath1315 . *
bundled upper triangular group + [ d5.7.3.7 ] @xmath1469 is called the bundled upper triangular group . + [ t5.7.3.8 ] consider the bundled upper triangular group .
1 . @xmath1470 is a lie sub - group of @xmath1425 , i.e. , @xmath1471 .
the lie algebra of @xmath1470 is @xmath1422 .
* bundled special upper triangular group + [ d5.7.3.9 ] @xmath1472 is called the bundled special upper triangular group . + [ t5.7.3.10 ] consider the bundled special upper triangular group .
1 . @xmath1473 is a lie sub - group of @xmath1470 , i.e. , @xmath1474 .
the lie algebra of @xmath1473 is @xmath1423 . *
bundled symplectic group + [ d5.7.3.11 ] @xmath1475 is called the bundled symplectic group .
+ @xmath1476 + where @xmath1477 is defined in ( [ 2.3.8 ] ) . [ t5.7.3.12 ]
consider the bundled symplectic group . 1 .
@xmath1478 is a lie sub - group of @xmath1425 , i.e. , @xmath1479 .
the lie algebra of @xmath1478 is @xmath1480 .
let @xmath1481 be the @xmath369-th order symmetric group .
denote @xmath1482 [ d5.7.3.1 ] a matrix @xmath1483 is called a permutation matrix , if @xmath1484 and @xmath1485 .
the set of @xmath642 permutation matrices is denoted by @xmath1486 .
[ p5.7.3.2 ] consider the set of permutation matrix . 1 .
if @xmath1487 , then @xmath1488 2 . @xmath1489 let @xmath1490 .
define a permutation matrix @xmath1491 as @xmath1492 $ ] , where @xmath1493 the following proposition is easily verifiable .
[ p5.7.3.3 ] define @xmath1494 , where @xmath1495 is constructed by ( [ 5.7.3.2 ] ) . then @xmath608 is an isomorphism .
assume @xmath1496 .
then proposition [ p5.7.3.3 ] leads to @xmath1497 next , assume @xmath1498 , @xmath1499 , we try to generalize ( [ 5.7.3.3 ] ) .
[ d5.7.3.4 ] assume @xmath1498 , @xmath1499 .
the ( left ) stp of @xmath487 and @xmath159 is defined by @xmath1500 where @xmath1501 .
that is , @xmath1502 similarly , we can define the right stp of @xmath487 and @xmath159 .
now , it is clear that @xmath1503 is a sub - monoid . to get a bundled lie subgroup structure
, we consider the quotient space @xmath1504 then we have the following : [ t5.7.3.5 ] @xmath1105 is a discrete bundled sub - lie group of @xmath1425 .
@xmath1105 may be used to investigate the permutation of uncertain number of elements .
consider the set of vectors on field @xmath28 .
denote it as @xmath1505 where @xmath1506 is the @xmath38 dimensional vector subspace of @xmath0 .
our purpose is to build a vector space structure on @xmath0 . to this end
, we first propose an equivalence relation .
[ d6.2.1 ] 1 .
let @xmath1507 .
@xmath375 and @xmath383 are said to be vector equivalent , denoted by @xmath1508 , if there exist two one - vectors @xmath1509 and @xmath1510 such that @xmath1511 2 .
the equivalent class is denoted as @xmath1512 3 . in an equivalent class
@xmath1513 a partial order ( @xmath78 ) is defined as : @xmath1514 , if there exists a one - vector @xmath1509 such that @xmath1515 .
@xmath1516 is irreducible , if there are no @xmath383 and @xmath1509 , @xmath1517 , such that @xmath1518 .
[ r6.2.2 ] 1 .
the equivalence defined above can be seen as the left equivalence .
formally , we set @xmath1519 .
the right equivalence can be defined as follows : let @xmath1507 . @xmath375 and @xmath383 are said to be right equivalent , denoted by @xmath1520 , if there exist two one - vectors @xmath1509 and @xmath1510 such that @xmath1521 3 . the right equivalent class of @xmath375 is denoted as @xmath1522 .
of course , we have @xmath1523 .
hereafter , the left equivalence is considered as the defaulted one . that is , we always assume @xmath1524 . but with some obvious modifications one sees easily that the arguments / results in the sequel are also valid for right equivalence .
the following properties of the matrix equivalence are also true for vector equivalence .
the proofs are also similar to the matrix equivalence .
therefore , they are omitted .
[ t6.2.3 ] 1 . if @xmath1508 , then there exists a vector @xmath191 such that @xmath1525 2 . in each class
@xmath1526 there exists a unique @xmath1527 , such that @xmath1528 is irreducible .
[ r6.2.4 ] 1 .
if @xmath1529 , then @xmath383 is called a divisor of @xmath375 and @xmath375 is called a multiple of @xmath383 .
this relation determined the order @xmath1530 .
if ( [ 6.2.4 ] ) holds and @xmath194 are co - prime , then the @xmath191 satisfying ( [ 6.2.4 ] ) is called the greatest common divisor of @xmath375 and @xmath383 .
moreover , @xmath1531 is unique .
if ( [ 6.2.2 ] ) holds and @xmath964 are co - prime , then @xmath1532 is called the least common multiple of @xmath375 and @xmath383 .
moreover , @xmath1533 is unique . 4 .
consider an equivalent class @xmath1513 , denote the unique irreducible element by @xmath1528 , which is called the root element .
all the elements in @xmath1513 can be expressed as @xmath1534 @xmath1535 is called the @xmath38-th element of @xmath1513 .
hence , an equivalent class @xmath1513 is a well ordered sequence as : @xmath1536 we also have a lattice structure on @xmath1537 : [ p6.2.5 ] @xmath1538 is a lattice .
_ it is easy to verify that for @xmath1539 , @xmath1540 the conclusion follows .
@xmath160 [ p6.2.6 ] let @xmath1541 and @xmath1542 be both irreducible .
then the two lattices @xmath198 and @xmath1543 , generated by @xmath165 and @xmath1528 are isomorphic .
precisely , @xmath1544 the isomorphism is : @xmath1545 .
next , we consider the lattice structure of @xmath1546 which is obviously a partition . observing that if @xmath1547 , say @xmath1548 , then @xmath1549 is a subspace of @xmath1550 .
we give an order as @xmath1551 . under this partial order
we have a lattice structure on @xmath0 as follows : [ p6.2.7 ] @xmath1552 is a lattice with @xmath1553 we also have the following isomorphic relation : [ p6.2.8 ] define @xmath1554 as : @xmath1555 then @xmath307 is a lattice isomorphism . [ d6.2.9 ] 1 .
let @xmath1332 .
the @xmath89-lower truncated vector space is defined as @xmath1556}:=\bigcup_{\{s\;\big|\ ; p|s\}}{\cal v}_s.\end{aligned}\ ] ] 2 .
the quotient space of @xmath1557 is denoted as @xmath1558 3 .
the subspace @xmath1559}/\lra\\ ~&=&\left\{\lceil x\rceil \;\big|\ ; x_1\in { \cal v}_i\right\}. \end{array}\end{aligned}\ ] ] next , we define an addition between vectors of different dimensions .
[ d6.3.1 ] let @xmath1560 and @xmath1561 .
. then 1 .
the vector addition @xmath1563 of @xmath375 and @xmath383 is defined as @xmath1564 2 .
the subtraction @xmath1565 is defined as @xmath1566 [ r6.3.101 ] let @xmath1567 and @xmath1568 be tow row vectors .
then we define 1 .
the vector addition as @xmath1569 2 .
the subtraction as @xmath1570 [ p6.3.2 ] the vector addition @xmath1571 is consistent with the equivalence @xmath70 .
that is , if @xmath1572 and @xmath1573 , then @xmath1574 . _
proof_. since @xmath1572 , according to theorem [ t6.2.3 ] , there exists @xmath191 , say , @xmath1575 , such that @xmath1576 similarly , there exists @xmath1577 , say , @xmath1578 , such that @xmath1579 let @xmath1580 , @xmath1581 , and @xmath1582 .
then @xmath1583 + \left[\pi\otimes \operatorname{{\bf 1}}_{s}\otimes \operatorname{{\bf 1}}_{\eta/(s q)}\right]\\ ~=\left[\gamma\otimes \operatorname{{\bf 1}}_{\eta / p}\right]+ \left[\pi\otimes \operatorname{{\bf 1}}_{\eta / q}\right]\\ ~=\left(\left[\gamma\otimes \operatorname{{\bf 1}}_{\xi / p}\right]+ \left[\pi\otimes \operatorname{{\bf 1}}_{\xi / q}\right]\right)\otimes \operatorname{{\bf 1}}_{\ell}\\ ~=(\gamma\lvplus\pi)\otimes \operatorname{{\bf 1}}_{\ell}. \end{array}\ ] ] hence @xmath1584 .
similarly , we can show that @xmath1585 .
the conclusion follows .
@xmath160 [ c6.3.3 ] the vector addition @xmath1571 ( or subtraction @xmath1565 ) is well defined on the quotient space @xmath1586 , as well as @xmath1587 , @xmath288 .
that is , @xmath1588 let @xmath1589 ( or @xmath1590 ) .
then we define a scale product @xmath1591 using ( [ 6.3.3 ] ) and ( [ 6.3.4 ] ) , one sees easily that @xmath1586 becomes a vector space : [ t6.3.4 ] using the vector addition defined in ( [ 6.3.3 ] ) and the scale product defined in ( [ 6.3.4 ] ) , we have 1 . @xmath1586 is a vector space over @xmath28 ; 2 .
@xmath1587 , @xmath288 are the subspaces of @xmath1586 ; 3 .
if @xmath353 , then @xmath1587 is a subspace of @xmath1592 . let @xmath1593 be a set of vectors . then @xmath1594 the following proposition shows that the vector equivalence keeps the space - subspace relationship unchanged .
[ p6.3.5 ] assume @xmath1595 is a subspace of @xmath1506 .
then @xmath1596 is a subspace of @xmath1587 .
definition [ d6.3.1 ] can be translated to its corresponding right one as follows : [ d6.3.6 ] let @xmath1560 and @xmath1561 .
. then 1 .
the ( right ) vector addition @xmath1597 of @xmath375 and @xmath383 is defined as @xmath1598 2 .
the subtraction @xmath1599 is defined as @xmath1600 then all the arguments in this subsection and the following several subsections can be stated in a parallel way for right addition and corresponding linear spaces .
[ d6.0.1 ] let @xmath1601 , @xmath1602 , and @xmath1501 .
then the weighted inner product is defied as @xmath1603 where @xmath1604 is the conventional inner product .
say , if @xmath30 , @xmath1605 , and if @xmath31 , @xmath1606 .
[ d6.0.2 ] let @xmath1607 .
their inner product is defined as @xmath1608 it is easy to verify the following proposition , which assures the definition [ d6.0.2 ] is reasonable .
[ p6.0.3 ] equation ( [ 6.0.2 ] ) is well defined .
that is , it is independent of the choice of @xmath375 and @xmath383 . since @xmath1586 is a vector space , using ( [ 6.0.2 ] ) as an inner product on @xmath1586 , then we have the follows : the @xmath1586 with inner product defined by ( [ 6.0.2 ] ) is an inner product space .
it is not a hilbert space . _ proof_. the first part is obvious .
as for the second part , we construct a sequence as @xmath1609 then we can prove that @xmath1610 is a cauchy sequence and it does not converge to any @xmath1611 .
@xmath160 given a vector @xmath1611 , then it determined a linear mapping @xmath1612 via inner product as @xmath1613 similarly , @xmath1614 can also determine a linear mapping @xmath1615 as @xmath1616 unfortunately , the inverse is not true . because @xmath1000 is an infinite dimensional vector space but any vector in @xmath1000 has only finite dimensional representatives .
next , we consider the linear mappings on @xmath0 . it is well known that assume @xmath13 and @xmath1617 .
then the product @xmath1618 , defined as @xmath1619 , for given @xmath1 is a linear mapping .
we intend to generalize such a linear mapping to arbitrary matrix and arbitrary vector .
[ d6.4.1 ] let @xmath13 , @xmath1560 , and @xmath87 .
then the vector product , denoted by @xmath65 , is defined as @xmath1620 [ r6.4.101 ] 1 . the vector product defined in ( [ 6.4.1 ] ) is the left vector product . of course
, we can define right vector product , denoted by @xmath66 , defined as @xmath1621 2 .
in fact , @xmath65 is a combination of @xmath68 of matrices and @xmath71 of vectors , and @xmath66 is a combination of @xmath1622 of matrices and @xmath1623 of vectors .
of course , we may define two more vector products by combinations of @xmath1624 with @xmath1623 and @xmath1622 with @xmath1625 respectively .
note that when @xmath1626 , @xmath1627 .
that is , the linear mapping defined in ( [ 6.4.1 ] ) is a generalization of conventional linear mapping .
it is also true for other vector products .
4 . to avoid similar but tedious arguments , hereafter
the default vector product is @xmath65 .
the following proposition is easily verifiable .
[ p6.4.2 ] consider the vector product @xmath1628 1 .
it is linear with respect to the second variable , precisely , @xmath1629 2 .
assume both @xmath472 , then the vector product is also linear with respect to the first variable , precisely , @xmath1630 the following proposition shows that the vector product is consistent with both matrix equivalence and vector equivalence .
[ p6.4.3 ] assume @xmath204 and @xmath1508 .
then @xmath1631 _ proof_. assume @xmath1632 , @xmath207 ; @xmath1633 , @xmath1634 , where @xmath1635 and @xmath1636 .
denote @xmath1637 , @xmath1638 , and @xmath1639 .
then we have @xmath1640\otimes \left[i_k\operatorname{{\bf 1}}_k\right]\\ ~=\left(\lambda \lvtimes \gamma\right)\otimes \operatorname{{\bf 1}}_k .
\end{array}\ ] ] hence @xmath1641 similarly , we have @xmath1642 equation ( [ 6.4.2 ] ) follows .
@xmath160 the above propositions have an immediate consequence as follows : [ p6.4.4 ] the vector product @xmath1643 can be extended to @xmath1644 . particularly , each @xmath1645 determines a linear mapping on the vector space @xmath1586 .
next , we extend the vector equivalence to matrices
. [ d6.4.5 ] 1 .
assume @xmath1646 .
@xmath830 and @xmath853 are said to be vector equivalent , denoted by @xmath1647 , if there exist @xmath1648 and @xmath1649 such that @xmath1650 2 . the equivalence class of @xmath830 is denoted as @xmath1651 [ r6.4.6 ] 1 .
it is clear the equivalence defined by ( [ 6.4.3 ] ) is left vector equivalence @xmath71 .
the right vector equivalence @xmath72 can be defined similarly . moreover ,
the right vector equivalence class is denoted by @xmath1652 .
a matrix can be considered as a linear mapping , and correspondingly the matrix equivalence is considered .
a matrix can also be considered as a vector subspace generated by its columns . in this way
its vector equivalence , defined by ( [ 6.4.3])-([6.4.4 ] ) , makes sense .
denote the equivalence class as @xmath1653 all the results about vector space @xmath0 can be extended to @xmath1654 for @xmath1352 .
they are briefly summarized as follows : [ p6.4.7 ] 1 .
assume @xmath1655 , @xmath1656 , @xmath1647 and @xmath1657 .
then we said @xmath830 is a divisor of @xmath853 or @xmath853 is a multiple of @xmath830 .
moreover , an order is determined by this as @xmath1658 .
2 . @xmath1659 is a lattice .
3 . assume @xmath1657 , then an order is given in @xmath1660 as @xmath1661 4 .
@xmath1662 is a lattice .
assume @xmath1655 is irreducible .
then @xmath1663 defined as @xmath1664 is a lattice homomorphism . the operators @xmath1571 and @xmath65 can also be defined in a similar way as for vectors : [ d6.4.8 ] 1 .
assume @xmath1665 and @xmath1666 , @xmath1562 .
the vector addition is defined as @xmath1667 2 .
let @xmath13 , @xmath437 , and @xmath87 . then the vector product of @xmath1 and @xmath830 is defined as @xmath1668 we can denote the equivalence class of @xmath1654 as @xmath1669 for vector product and vector addition we also have the distributive law with respect to these two operators for matrix case .
[ p6.4.9 ] @xmath1670 is distributive .
precisely , let @xmath472 and @xmath1671 , @xmath1672 .
then @xmath1673 @xmath1674 _ proof_. we prove ( [ 6.7.11 ] ) .
the proof of ( [ 6.7.12 ] ) is similar .
denote @xmath1675 , @xmath1676 , @xmath1677 , and @xmath1678 .
then for ( [ 6.7.11 ] ) we have @xmath1679 \left[c\otimes { \bf 1}_{\xi / p}\right]\\ ~=a\left(a\otimes i_{\xi / m}\right)\left(c\otimes { \bf 1}_{\xi / p}\right)+b\left(b\otimes i_{\xi / n}\right)\left(c\otimes { \bf 1}_{\xi / p}\right)\\ ~=a\left[\left(a\otimes i_{s / m}\right)\left(c\otimes { \bf 1}_{s / p}\right)\right]\otimes { \bf 1}_{\xi / s}\\ ~+b\left[\left(b\otimes i_{t / n}\right)(c\otimes { \bf 1}_{t / p})\right]\otimes { \bf 1}_{\xi / t}\\ ~=a\left[\left(a\otimes
i_{s / m}\right)\left(c\otimes { \bf 1}_{s / p}\right)\right]\\~\lvplus b\left[\left(b\otimes i_{t / n}\right)(c\otimes { \bf 1}_{t / p})\right]\\ ~=rhs . \end{array}\ ] ] @xmath160 finally , we can extend the distributive law to equivalence spaces as follows .
[ c6.4.10 ] @xmath1680 is distributive .
precisely , let @xmath694 and @xmath1681 . then @xmath1682 @xmath1683 given @xmath281
, we seek a subspace @xmath1684 which is @xmath1 invariant .
[ d6.5.1 ] let @xmath1684 be a vector subspace . if @xmath1685 , @xmath425 is called an @xmath1-invariant subspace . moreover , if @xmath1686 , it is called the type 1 invariant subspace ; otherwise , it is called the type 2 invariant subspace .
this subsection considers the type 1 invariant subspace only .
[ p6.5.2 ] let @xmath1687 and @xmath1688 .
then @xmath425 is @xmath1-invariant , if and only if , * @xmath1689 * @xmath1690 , where @xmath38 satisfies @xmath1691 _ proof_. ( necessity ) assume @xmath1692 , by definition we have @xmath1693 hence we have @xmath1694 it follows from ( [ 6.5.3 ] ) that @xmath1695 since @xmath1696 and @xmath1697 are co - prime and the left hand side of ( [ 6.5.4 ] ) is an integer , we have @xmath1698 .
it follows that @xmath1699 ( sufficiency ) assume ( [ 6.5.2 ] ) holds , then ( [ 6.5.3 ] ) holds .
it follows that @xmath1700 @xmath160 assume @xmath1687 and @xmath1688 , a natural question is : can we find @xmath1 such that @xmath425 is @xmath1-invariant ? according to proposition [ p6.5.2 ] , we know that it is necessary that @xmath1698 .
let @xmath1701 be the prime divisors of @xmath1702 , then we have @xmath1703 where @xmath292 are co - prime and @xmath1704 , @xmath1705 , @xmath379 .
now it is obvious that to meet ( [ 6.5.2 ] ) it is necessary and sufficient that @xmath1706 where @xmath1707 . summarizing the above argument we have the following result .
[ p6.5.3 ] assume @xmath1687 , @xmath1688 .
@xmath425 is @xmath1-invariant , if and only if , ( i ) @xmath1698 , and ( ii ) @xmath281 , where @xmath38 satisfies ( [ 6.5.6 ] ) .
[ r6.5.4 ] assume @xmath1708 , @xmath1688 .
using proposition [ p6.5.3 ] , it is clear that @xmath425 is @xmath1-invariant , if and only if , @xmath1709 with @xmath1710 particularly , when @xmath1711 then @xmath1712 is a number .
so it is a scale product of @xmath425 , i.e , for @xmath1713 we have @xmath1714 .
when @xmath1715 then @xmath1716 .
so we have @xmath1717 .
this is the classical linear mapping on @xmath1718 .
we call these two linear mapping the standard linear mapping .
the following example shows that there are lots of non - standard linear mappings .
[ e6.5.401 ] 1 .
assume @xmath1719 , @xmath1720 .
then @xmath1698 , @xmath1721 .
@xmath1722 . using ( [ 6.5.5 ] ) , @xmath1723 and @xmath1724 . according to ( [ 6.5.6 ] )
@xmath1725 , where @xmath1726 , @xmath1727 or @xmath1728 .
according to proposition [ p6.5.3 ] , @xmath1729 is @xmath1-invariant , if and only if , @xmath1730 or @xmath1731 .
, we give a numerical example : assume @xmath1732 then @xmath1733 is @xmath1-invariant space .
for instance , let @xmath1734 then @xmath1735 motivated by ( [ 6.5.9 ] ) , we give the following definition .
[ d6.5.5 ] assume @xmath136 and @xmath1611 .
if @xmath1736 then @xmath145 is called an eigenvalue of @xmath1 , and @xmath375 is called an eigenvector of @xmath1 with respect to @xmath145 .
[ e6.5.6 ] recall example [ e6.5.401 ] .
in fact it is easy to verify that matrix @xmath1 in ( [ 6.5.7 ] ) , as a linear mapping on @xmath1733 , has @xmath1737 eigenvalues : precisely , @xmath1738 correspondingly , the first eigenvector is @xmath375 , defined in ( [ 6.5.8 ] ) .
the other @xmath1739 eigenvectors are @xmath1740 , @xmath1741 , @xmath1742 ( this is a root vector ) , @xmath1743 , and @xmath1744 , respectively .
assume @xmath1688 is @xmath1-invariant with respect to @xmath65 , that is , @xmath1745 then the restriction @xmath1746 is a linear mapping on @xmath425 .
it follows that there exists a matrix , denoted by @xmath1747 , such that @xmath1746 is equivalent to @xmath1748 .
we state it as a proposition .
[ p6.5.7 ] assume @xmath136 and @xmath1688 is @xmath1-invariant .
then there exists a unique matrix @xmath1747 , such that @xmath1746 is equivalent to @xmath1748 .
precisely , @xmath1749 @xmath1748 is called the realization of @xmath1 on @xmath1750 .
[ r6.5.8 ] 1 . to calculate @xmath1748 from @xmath1
in fact , it is clear that @xmath1751}\right)=a\lvtimes \d_t^i , \quad i=1,\cdots , t.\end{aligned}\ ] ] 2 .
consider example [ e6.5.401 ] again .
@xmath1752\left[\d_6 ^ 1\otimes \operatorname{{\bf 1}}_2\right]\\ ~=(1,1,0,0,0,0)^t . \end{array}\ ] ] similarly , we can calculate all other columns . finally , we have @xmath1753 finally , given a matrix @xmath281 , we would like to know whether it has ( at least one ) type-1 invariant subspace @xmath1688 ?
then as a corollary of proposition [ p6.5.2 ] , we can prove the following result .
[ c6.5.9]assume @xmath281 , then @xmath1 has ( at least one ) type 1 invariant subspace @xmath1688 , if and only if , @xmath1754 _ proof . _
according to proposition [ p6.5.2 ] , @xmath1698 is obvious necessary .
we prove it is also sufficient .
assume @xmath38 is factorized into its prime factors as @xmath1755 and correspondingly , @xmath1696 is factorized as @xmath1756 where @xmath89 is co - prime with @xmath38 ; @xmath1757 is factorized as @xmath1758 where @xmath85 is co - prime with @xmath38 .
using proposition [ p6.5.2 ] again , we have only to prove that there exists at least one @xmath1757 satisfying ( [ 6.5.2 ] ) .
calculate that @xmath1759 to meet ( [ 6.5.2 ] ) a necessary condition is : @xmath89 and @xmath85 are co - prime .
next , fix @xmath1760 , we consider two cases : ( i ) @xmath1761 : then on the lhs ( left hand side ) of ( [ 6.5.2 ] ) we have factor @xmath1762 and on the rhs of ( [ 6.5.2 ] ) we have factor @xmath1763 . hence , as long as @xmath1764 , we can choose @xmath1765 to meet ( [ 6.5.2 ] ) .
( ii ) @xmath1766 : then on the lhs we have factor @xmath1767 and on the rhs we have factor @xmath1763 . hence , as long as @xmath1768 , ( [ 6.5.2 ] ) is satisfied . @xmath160 using above notations , we also have the following result [ c6.5.10 ] assume @xmath1698 .
then @xmath1718 is @xmath1-invariant , if and only if , ( i ) for @xmath1764 , the corresponding @xmath1769 ; ( ii ) for @xmath1770 , the corresponding @xmath1768 .
[ e6.5.11 ] recall example [ e6.5.401 ] again . 1 .
since the matrix @xmath1 defined in ( [ 6.5.7 ] ) is in @xmath1771 .
we have @xmath1772 , @xmath1773 , @xmath1774 and hence @xmath1775 .
according to corollary [ c6.5.10 ] , @xmath1688 is @xmath1-invariant , if and only if , @xmath1776 and @xmath85 is co - prime with @xmath1772 .
hence @xmath1777 are type-1 invariant subspaces of @xmath1 .
2 . assume @xmath1778 .
then the restriction is @xmath1779}=\begin{bmatrix } 1 & 0 & -1 & 0 & 0 & 0 & 0 & 0 & 0 & 0\\ 1 & 0 & 0 & -1 & 0 & 0 & 0 & 0 & 0 & 0\\ 0 & 1 & 0 & -1 & 0 & 0 & 0 & 0 & 0 & 0\\ 0 & 1 & 0 & 0 & -1 & 0 & 0 & 0 & 0 & 0\\ 0 & 0 & 1 & 0 & -1 & 0 & 0 & 0 & 0 & 0\\ 0 & 0 & 0 & 0 & 0 & 1 & 0 & 0 & 0 & 0\\ 0 & 0 & 0 & 0 & 0 & 1 & 0 & 0 & 0 & 0\\ 0 & 0 & 0 & 0 & 0 & 0 & 1 & 0 & 0 & 0\\ 0 & 0 & 0 & 0 & 0 & 0 & 1 & 0 & 0 & 0\\ 0 & 0 & 0 & 0 & 0 & 0 & 0 & 1 & 0 & 0 \end{bmatrix}\ ] ] the eigenvalues are @xmath1780 the corresponding eigenvectors are : @xmath1781 note that @xmath1782 , @xmath1783 are two root vectors . that is , @xmath1784 [ r6.5.11 ] in fact , the set of type-1 @xmath1-invariant subspace depends only on the shape of @xmath1 .
hence we define @xmath1785 we can also briefly call @xmath1786 the set of @xmath277-invariant subspaces . before ending this subsection
we consider a general type-1 invariant subspace @xmath1686 .
the following proposition is obvious .
[ p6.5.12 ] if @xmath1686 is an @xmath1-invariant subspace , then @xmath1718 is also an @xmath1-invariant subspace .
because of proposition [ p6.5.12 ] searching @xmath425 becomes a classical problem . because we can first find a matrix @xmath1787 , which is equivalent to @xmath1788 .
then @xmath425 must be a classical invariant subspace of @xmath1789 .
denote the set of type-1 @xmath1-invariant subspaces as @xmath1790 assume @xmath281 . then @xmath1791 . to assure @xmath1792 , through this subsection we assume @xmath1698 .
we give a definition for this .
[ d6.6.0 ] 1 .
assume @xmath1687 .
@xmath1 is said to be a bounded operator ( or briefly , @xmath1 is bounded , ) if @xmath1698 .
2 . a sequence @xmath1793 ,
is called an @xmath1 generated sequence if @xmath1794 3 .
a finite sequence @xmath1795 , is called an @xmath1 generated loop if @xmath1796 .
[ l6.6.1 ] assume there is an @xmath1 generated loop @xmath1550 , @xmath1797 , @xmath1798 , @xmath1799 , @xmath1550 , as depicted in ( [ 6.6.1 ] ) .
@xmath1800 then @xmath1801 _ proof_. assume @xmath281 . according to definition [ d6.4.1 ] , we have the following dimension relationship : @xmath1802 next , we @xmath1803 we conclude that @xmath1804 equivalently , we have
@xmath1805 it follows that @xmath1806 define @xmath1807 where @xmath1808 . then from the last equation of ( [ 6.6.3000 ] ) we have @xmath1809 that is , @xmath1810 using ( [ 6.6.4 ] ) , and the expression @xmath1811 we have @xmath1812 from above it is clear that @xmath1813 next , using last two equation in ( [ 6.6.3000 ] ) , we have @xmath1814 similar to the above argument , we have @xmath1815 hence @xmath1816 to meet this requirement , it is necessary that @xmath1817 continuing this process , finally we have @xmath1818 combining ( [ 6.6.4 ] ) with ( [ 6.6.7 ] ) yields that @xmath1819 that is , @xmath1820
. @xmath160 [ t6.6.2 ] a finite dimensional subspace @xmath1684 is @xmath1-invariant , if and only if , @xmath425 has the following structure : @xmath1821 where @xmath1822 _ proof_. since @xmath425 is of finite dimension , there are only finite @xmath1823 such that @xmath1824 now for each @xmath1825 we construct @xmath1826 for certain @xmath1827 .
note that @xmath425 is @xmath1-invariant , if for all @xmath1828 , we have @xmath1829 , then this @xmath1830 is the end element in the sequence . otherwise , we can find a successor @xmath1831 .
note that since there are only finite @xmath1832 , according to lemma [ l6.6.1 ] , starting from @xmath1833 there are only finite sequence of differen @xmath1832 till it reach an @xmath1-invariant @xmath1834 ( equivalently , @xmath1-invariant @xmath1835 ) .
the claim follows .
@xmath160 [ d6.7.1 ] let @xmath281 , @xmath1718 is @xmath1-invariant subspace of @xmath0 .
that is , @xmath1836 is a linear mapping .
the higher order linear mapping of @xmath1 , is defined as @xmath1837}\lvtimes x:=a\lvtimes x,\quad x\in { \cal v}_t\\ a^{[k+1]}\lvtimes x:=a\lvtimes\left(a^{[k]}\lvtimes x\right),\quad k\geq 1 .
\end{cases}\end{aligned}\ ] ] [ d6.7.2 ] let @xmath1611 .
the @xmath1-sequence of @xmath375 is the sequence @xmath1838 , where @xmath1839 using notations ( [ 6.5.15])-([6.5.17 ] ) , we have the following result .
[ l6.7.3 ] assume @xmath1 is bounded .
then @xmath1840 , if and only if , for each @xmath1841 , one of the following is true : @xmath1842 or @xmath1843 _ proof_. @xmath1844 and @xmath1845 then @xmath1846 and @xmath1847 now @xmath1840 is assured by ( [ 6.5.2 ] ) , which leads to @xmath1848 1 . :
@xmath1849 , ( [ 6.7.4 ] ) leads to @xmath1850 .
hence , we have @xmath1851 2 . :
@xmath1852 , which leads to @xmath1853 .
@xmath1854 , which assures ( [ 6.7.4 ] ) .
combining case 2 and case 3 yields @xmath1855 note that when @xmath1850 , if @xmath1856 we have ( [ 6.7.6 ] ) .
hence @xmath1856 is not necessary .
the conclusion follows .
@xmath160 the following result is important . [ t6.7.4 ]
let @xmath281 be bounded .
then for any @xmath1611 the @xmath1-sequence of @xmath375 will enter a @xmath1857 at finite steps . _
proof_. assume @xmath1858 . using notations ( [ 6.5.15])-([6.5.17 ] ) , it is easy to calculate that after one step the @xmath1760-th index of @xmath1757 becomes @xmath1859 assume @xmath1850 , this component already meets the requirement of lemma [ l6.7.3 ] .
assume for some @xmath1760 , @xmath1860 and @xmath1861 , then we have @xmath1862 hence after finite times , say @xmath369 , the @xmath1760-th index of @xmath1863 , denoted by @xmath1864 , satisfies @xmath1865 will satisfy ( [ 6.7.3 ] ) , and as long as ( [ 6.7.3 ] ) holds , @xmath1866 @xmath1867 . hence , after finite steps either ( [ 6.7.2 ] ) or ( [ 6.7.3 ] ) ( or both ) is satisfied . then at the next step the sequence enters into @xmath1857 .
@xmath160 [ d6.7.5 ] given a polynomial @xmath1868 a matrix @xmath136 and a vector @xmath1869 . 1 .
@xmath1215 is called an @xmath1-annihilator of @xmath375 , if @xmath1870}\lvtimes x \lvplus c_{n-1 } a^{[n-1]}\lvtimes x\\ ~&~&\lvplus \cdots \lvplus c_1a\lvtimes x\lvplus c_0=0 . \end{array}\end{aligned}\ ] ] 2 .
assume @xmath1871 is the @xmath1-annihilator of @xmath375 with minimum degree , then @xmath1871 is called the minimum @xmath1-annihilator of @xmath375 .
the following result is obvious : [ p6.7.6 ] the minimum @xmath1-annihilator of @xmath375 divides any @xmath1-annihilator of @xmath375 .
the following result is an immediate consequence of theorem [ t6.7.4 ] .
[ c6.7.7 ] assume @xmath1 is bounded , then for any @xmath1611 there exists at least one @xmath1-annihilator of @xmath375 .
_ according to theorem [ t6.7.4 ] , there is a finite @xmath369 such that @xmath1872}x\in { \cal v}_{s}$ ] with @xmath1873 being @xmath1-invariant . now in @xmath1873 assume the minimum annihilator polynomial for @xmath1872}x$ ] is @xmath1871 , then @xmath1874 is an @xmath1-annihilator of @xmath375 .
@xmath160 [ e6.7.8 ] 1 .
assume @xmath1875 .
since @xmath1876 , we know any @xmath375 does not have its @xmath1-annihilator .
+ now assume @xmath1877 , where @xmath1878 , and @xmath1879 are co - prime .
then it is easy to see that the @xmath1 sequence of @xmath1880 has the dimensions , @xmath1881 , which are : @xmath1882 , @xmath1883 @xmath1798 @xmath1884 , @xmath1885 , @xmath1886 , @xmath1798 .
it can not reach a @xmath1857 .
2 . given @xmath1887
we try to find the minimum @xmath1-annihilator of @xmath375 .
set @xmath1888 .
it is easy to see that @xmath1889 hence , we can find the annihilator of @xmath375 in the space of @xmath1733 . calculating @xmath1890
it is easy to verify that @xmath1891 are linearly independent .
moreover , @xmath1892 the minimum @xmath1-annihilator of @xmath1893 follows as @xmath1894 denote @xmath1653 we extend the vector product to the equivalence spaces .
[ d6.8.1 ] let @xmath136 and @xmath1895
. then we define @xmath1896 as @xmath1897 the following proposition shows that ( [ 6.8.1 ] ) is well defined .
[ p6.8.2 ] ( [ 6.8.1 ] ) is independent of the choice of @xmath1 and @xmath90 . _
proof_. assume @xmath563 is irreducible and @xmath1898 ; @xmath564 is also irreducible and @xmath1899 . @xmath1900 and @xmath1901 .
set @xmath1902 , @xmath1903 , and @xmath1904 .
then @xmath1905 @xmath160 precisely speaking , because @xmath0 is not a vector space , invariant subspace " is not a rigorous subspace .
but @xmath1906 is a vector space .
it is easy to see that the results about @xmath0 , and @xmath1654 can be extended to @xmath1907 and @xmath1908 .
for instance , @xmath629 is a bounded operator on @xmath1907 if and only if , @xmath1698 .
matrix theory is one of the most fundamental and useful tools in modern science and technology .
but one of the major weaknesses is its dimension restriction . to overcome this barrier ,
the purpose of this paper is to set up a framework for an almost dimension - free matrix theory .
first , we review the stp ( @xmath63 ) , which extends the conventional matrix product to overall matrices @xmath9 .
the related monoid structure for @xmath1909 is obtained .
the matrix equivalence @xmath67 is proposed .
a lattice structure over each equivalence class is obtained .
the equivalence space , as the quotient space @xmath1910 is discussed .
second , the set of overall matrices is partitioned into subspaces as @xmath1911 .
the sta ( @xmath475 ) is proposed . under this addition
the quotient spaces @xmath1912 become vector spaces .
certain geometric and algebraic structures are revealed .
including topological structure , inner product structure .
differential manifold structure , etc .
particularly , when @xmath1913 , ( corresponding to square matrices ) we have extended the lie algebra and lie group theory to @xmath1208 . a fiber bundle structure , called the discrete bundle ,
is proposed for @xmath2 and the extended lie group and lie algebra .
finally , the set of overall vectors @xmath0 are considered as a universal vector space , based on the vector equivalence @xmath70 .
a matrix @xmath1 of any dimension can be considered either a linear mapping on @xmath0 , or a subspace generated by its columns .
the @xmath1-invariant subspace is discussed in details .
many key concepts such as eigenvalue / eigenvector , characteristic polynomial of a matrix have been extended from square matrices to no - square matrices .
it was said by asimov that only in mathematics is there no significant correction - only existence .
once the greeks had developed the deductive method , they were correct in what they did , correct for all the time . "
@xcite all extensions we did in this paper consist with the classical ones .
that is , when the dimension restrictions required by the classical matrix theory are satisfied the new operators proposed in this paper coincide with the classical ones .
there are many questions remain for further discussion .
for instance , is it possible to construct an equivalence over @xmath9 , which is consistent with certain matrix product , such that the quotient space becomes a vector space ? 1 .
equivalence 1 : + [ d7.1.1 ] let @xmath124 .
@xmath1 is said to be equivalent to @xmath90 , if there exist @xmath1914 , @xmath1915 , @xmath1648 , @xmath1649 such that @xmath1916 + it is easy to verify that @xmath1917 is an equivalence relation .
moreover , similar to matrix equivalence or vector equivalence , we have the following result : + [ t7.1.2 ] assume @xmath1918 , then there exists a @xmath161 , such that @xmath1919 + hence the lattice structure similar to matrix equivalence exists .
+ it may be considered as a combination of matrix and vector equivalences .
unfortunately , ( i ) it is not consistent with stp ( @xmath63 ) ; ( ii ) the quotient space is not a vector space .
2 . equivalence 2 : + [ d7.1.3 ] let @xmath124 .
@xmath1 is said to be equivalent to @xmath90 , if there exist @xmath1920 , @xmath1921 , @xmath1648 , @xmath1649 such that @xmath1922 + the lattice structure can also be determined in a similar way .
moreover , the quotient space is a vector space .
unfortunately , a proper product , which is consistent with the equivalence , is unknown . 1 .
more geometric structure on equivalence space could be interesting .
for instance , a riemannian geometric structure or a symplectic geometric structure may be posed on the equivalence space .
2 . under the stp and @xmath475 ( @xmath514 ) , for any @xmath1687 and @xmath1923 , @xmath1924=a\ltimes b\lminus b\ltimes a\end{aligned}\ ] ] is well defined .
moreover , the three requirements ( [ 4.3.1])-([4.3.3 ] ) in definition [ d4.3.1 ] can also be satisfied ( under obvious modification ) .
exploring the properties of this generalized lie algebra is challenging and interesting .
one may be more interested in its applications .
for instance , can we use the extended structure proposed in this paper to the analysis and control of certain dynamic systems ?
particularly , we may consider the following special cases : 1 .
consider a dynamic system @xmath1925 where @xmath1926 satisfies @xmath1927 where @xmath1928 is a vector field on @xmath1929 .
what can we say about this system ?
is it possible to extend this system to the equivalent space @xmath1000 ?
2 . a dimension - varying dynamic control system as @xmath1930 where @xmath1931 . + what can we say about this , say , controllability ?
observability etc . ? | motivated by semi - tensor product(stp ) of matrices , certain equivalences of matrices are proposed , which allow the matrix operators being more general , beyond the dimension restrictions .
precisely speaking , it allows the matrix product being available for two arbitrary matrices , and matrix addition being available for certain matrices of different dimensions .
in addition , the characteristic function , eigenvalue / eigenvector etc
. are also defined for a large class of non - square matrices .
all these extensions are the generalization of classical ones
. that is , when the matrices meet the dimension requirement for the classical definitions , the extended definitions coincide with the classical ones . to this end , a new matrix product , called the stp ,
is first reviewed , which extends the classical matrix product to two arbitrary matrices .
under stp the set of matrices becomes a monoid ( semi - group with identity ) .
some related structures and properties are investigated .
then the generalized matrix addition is also presented , which extends the classical matrix addition to a class of two matrices with different dimensions .
two kinds of equivalences are introduced , which are matrix equivalence and vector equivalence .
the lattice structure has been established for each equivalence . under each equivalence , the corresponding quotient space becomes a vector space . under matrix equivalence
, many algebraic , geometric , and analytic structures have been posed to the quotient space , which include ( i ) lattice structure ; ( ii ) inner product and norm ( distance ) ; ( iii ) topology ; ( iv ) a fiber bundle structure , called the discrete bundle ; ( v ) bundled differential manifold ; ( vi ) bundled lie group and lie algebra . under vector space equivalence , vectors of different dimensions form a vector space @xmath0 , and a matrix @xmath1 of arbitrary dimension is considered as an operator ( linear mapping ) on @xmath0 .
when @xmath1 is a bounded operator ( not necessarily square but includes square matrices as a special case ) , the generalized characteristic function , eigenvalue and eigenvector etc .
are defined . in one word
, this new matrix theory overcomes the dimensional barrier in certain sense .
it provides much more freedom for using matrix approach to practical problems .
semi - tensor product / addition(sta / sta ) , vector product / addition(vp / va ) , matrix / vector equivalence ( me / ve ) , lattice , topology , fiber bundle , bundled manifold / lie algebra / lie group(bm / bla / blg ) . |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.